Amd ryzen 5 2600x vs i7 7700: AMD Ryzen 5 2600X vs Intel Core i7-7700

Functions

The price ratio

9. 5 /10

2 Votes

reviews is not

Games

/10

2 VOTES

reviews yet not

performance

10.0 /10

2 Votes

Reviews yet not

Reliability

/10

2 Votes

reviews so far

9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000

Energy efficiency

10.0 /10

2 votes

No reviews yet

Performance

1.cpu speed

6 x 3.6GHz

4 x 3.6GHz

CPU speed indicates how many processing cycles per second a processor can perform, given all its cores (processors). It is calculated by adding the clock speeds of each core or, in the case of multi-core processors, each group of cores.

2nd processor thread

More threads result in better performance and better multi-tasking.

3.speed turbo clock

4.25GHz

4.2GHz

When the processor is running below its limits, it can jump to a higher clock speed to increase performance.

4. Unlocked

✔AMD Ryzen 5 2600X

✖Intel Core i7-7700

Some processors come with an unlocked multiplier and can be easily overclocked for better performance in games and other applications.

5.L2 Cache

More L2 scratchpad memory results in faster results in CPU and system performance tuning.

6.L3 cache

More L3 scratchpad memory results in faster results in CPU and system performance tuning.

7.L1 cache

More L1 scratchpad memory results in faster results in CPU and system performance tuning.

8.core L2

0.5MB/core

0.25MB/core

More data can be stored in the L2 scratchpad for access by each processor core.

9.core L3

2.67MB/core

2MB/core

More data can be stored in L3 scratchpad for access by each processor core.

Memory

1.RAM speed

2933MHz

2400MHz

Can support faster memory which speeds up system performance.

2.max memory bandwidth

43.71GB/s

38.4GB/s

This is the maximum rate at which data can be read from or stored in memory.

3.DDR version

DDR (Dynamic Random Access Memory, Double Data Rate) is the most common type of RAM. New versions of DDR memory support higher maximum speeds and are more energy efficient.

4 memory channels

More memory channels increase the speed of data transfer between memory and processor.

5.Maximum memory

Maximum memory (RAM).

6.bus baud rate

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

The bus is responsible for transferring data between various components of a computer or device.

7.Supports memory troubleshooting code

✖AMD Ryzen 5 2600X

✖Intel Core i7-7700

Memory error recovery code can detect and repair data corruption. It is used when necessary to avoid distortion, such as in scientific computing or when starting a server.

8.eMMC version

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

Unknown. Help us offer a price. (Intel Core i7-7700)

A newer version of eMMC — built-in flash memory card — speeds up the memory interface, has a positive effect on device performance, for example, when transferring files from a computer to internal memory via USB.

9.bus frequency

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

Unknown. Help us offer a price. (Intel Core i7-7700)

The bus is responsible for transferring data between various components of a computer or device

Geotagging

1. PassMark result

This test measures processor performance using multi-threading.

2. PassMark result (single)

This test measures processor performance using a thread of execution.

3.Geekbench 5 result (multi-core)

Unknown. Help us offer a price. (Intel Core i7-7700)

Geekbench 5 is a cross-platform benchmark that measures the performance of a multi-core processor. (Source: Primate Labs,2022)

4. Cinebench R20 result (multi-core)

Unknown. Help us offer a price. (Intel Core i7-7700)

Cinebench R20 is a benchmark that measures the performance of a multi-core processor by rendering a 3D scene.

5.Cinebench R20 result (single core)

Unknown. Help us offer a price. (Intel Core i7-7700)

Cinebench R20 is a test to evaluate the performance of a single core processor when rendering a 3D scene.

6.Geekbench 5 result (single core)

Unknown. Help us offer a price. (Intel Core i7-7700)

Geekbench 5 is a cross-platform benchmark that measures the single-core performance of a processor. (Source: Primate Labs, 2022)

7. Blender test result (bmw27)

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

Unknown. Help us offer a price. (Intel Core i7-7700)

The Blender benchmark (bmw27) measures CPU performance by rendering a 3D scene. More powerful processors can render a scene in a shorter time.

8.Blender result (classroom)

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

Unknown. Help us offer a price. (Intel Core i7-7700)

The Blender (classroom) benchmark measures CPU performance by rendering a 3D scene. More powerful processors can render a scene in a shorter time.

9.performance per watt

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

Unknown. Help us offer a price. (Intel Core i7-7700)

This means that the processor is more efficient, resulting in more performance per watt of power used.

Functions

1.uses multithreading

✔AMD Ryzen 5 2600X

✔Intel Core i7-7700

processor cores into logical cores, also known as threads. Thus, each core can run two instruction streams at the same time.

2. Has AES

✔AMD Ryzen 5 2600X

✔Intel Core i7-7700

AES is used to speed up encryption and decryption.

3. Has AVX

✔AMD Ryzen 5 2600X

✔Intel Core i7-7700

AVX is used to help speed up calculations in multimedia, scientific and financial applications, and to improve the performance of the Linux RAID program.

SSE 4th version

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

SSE is used to speed up multimedia tasks such as image editing or audio volume control. Each new version contains new instructions and improvements.

5.Has F16C

✖AMD Ryzen 5 2600X

✔Intel Core i7-7700

F16C is used to speed up tasks such as image contrast adjustment or volume control.

6 bits transmitted at the same time

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

Unknown. Help us offer a price. (Intel Core i7-7700)

NEON provides faster media processing such as MP3 listening.

7. Has MMX

✔AMD Ryzen 5 2600X

✔Intel Core i7-7700

MMX is used to speed up tasks such as adjusting image contrast or adjusting volume.

8.Has TrustZone

✖AMD Ryzen 5 2600X

✖Intel Core i7-7700

Technology is integrated into the processor to ensure device security when using features such as mobile payments and video streaming through Digital Rights Management (DRM) ).

9.interface width

Unknown. Help us offer a price. (AMD Ryzen 5 2600X)

Unknown. Help us offer a price. (Intel Core i7-7700)

The processor can decode more instructions per clock (IPC), which means that the processor performs better

Price comparison

Cancel

Which CPUs are better?

AMD Ryzen 5 2600X or Intel Core i7-7700

• Home
• >

Core i7-7700

68.4

Performance in games and similar applications, according to our tests.

The performance of 4 cores, if any, and performance per core has the greatest impact on the result, since most games do not fully use more than 4 cores.

The speed of caches and working with RAM is also important.

Office speed

Ryzen 5 2600X

70.7 (+1.7%)

Core i7-7700

69. 5

Performance in day to day activities such as browsing and office applications.

The performance of 1 core has the greatest impact on the result, since most of these applications use only one, ignoring the rest.

Similarly, many professional applications such as various CADs ignore multi-threaded performance.

Speed ​​in heavy applications

Ryzen 5 2600X

48.1 (+12.3%)

Core i7-7700

42.2

Performance in resource-intensive tasks loading a maximum of 8 cores.

The performance of all cores and their number have the greatest impact on the result, since most of these applications willingly use all the cores and increase the speed accordingly.

At the same time, certain periods of work can be demanding on the performance of one or two cores, for example, applying filters in the editor.

Data obtained from tests by users who tested their systems with and without overclocking. Thus, you see the average values ​​corresponding to the processor.

Speed ​​of numerical operations

Simple household tasks

Ryzen 5 2600X 65.9 Core i7-7700 67.2 (+1.9%)

Demanding games and tasks

Ryzen 5 2600X 50.6 (+14.6%) Core i7-7700 43.2

Extreme

Ryzen 5 2600X 17.3 (+37.6%) Core i7-7700 10.8

Different tasks require different CPU strengths. A system with few fast cores and low memory latency will be fine for the vast majority of games, but will be inferior to a system with a lot of slow cores in a rendering scenario.

We believe that a minimum of 4/4 (4 physical cores and 4 threads) processor is suitable for a budget gaming PC. At the same time, some games can load it at 100%, slow down and freeze, and performing any tasks in the background will lead to a drop in FPS.

Ideally, the budget shopper should aim for a minimum of 4/8 and 6/6. A gamer with a big budget can choose between 6/12, 8/8 and 8/16. Processors with 10 and 12 cores can perform well in games with high frequency and fast memory, but are overkill for such tasks. Also, buying for the future is a dubious undertaking, since in a few years many slow cores may not provide sufficient gaming performance.

When choosing a processor for your work, consider how many cores your programs use. For example, photo and video editors can use 1-2 cores when working with filtering, and rendering or converting in the same editors already uses all threads.

Data obtained from tests by users who tested their systems both with overclocking (maximum value in the table) and without (minimum). A typical result is shown in the middle, the more filled in the color bar, the better the average result among all tested systems.

Benchmarks

Benchmarks were run on stock hardware, that is, without overclocking and with factory settings. Therefore, on overclocked systems, the points can noticeably differ upwards. Also, small performance changes may be due to the BIOS version.

Cinebench R23 Single Core

AMD Ryzen 5 2600X

1074

Intel Core i7-7700

1114 (+3.6%)

Cinebench R23 Multi Core

AMD Ryzen 5 2600X

7514 (+27.8%)

Intel Core i7-7700

5426

Cinebench R20 Single Core

AMD Ryzen 5 2600X

421 (+1.7%)

Intel Core i7-7700

414

Cinebench R20 Multi Core

AMD Ryzen 5 2600X

3048 (+29.9%)

Intel Core i7-7700

2138

Cinebench R15 Single Core

AMD Ryzen 5 2600x

Intel Core I7-7700

9000 (+5. 5.5%) Cinebench R15 Multi Core

AMD Ryzen 5 2600X

1380 (+34.7%)

Intel Core i7-7700

901

Geekbench 5 Single Core

AMD Ryzen 5 2600X

1052

Intel Core i7-7700

1066 (+1.3%)

Geekbench 5 Multi Core

AMD Ryzen 5 2600X

5782 (+29.4%)

Intel Core i7-7700

4081

Blender 2.81 bmw27

AMD Ryzen 5 2600X

278.4

Intel Core i7- 7700

387.7 (+28.2%)

Geekbench 3 Single Core

AMD Ryzen 5 2600X

4823 (+0.9%)

Intel Core i7-7700

4779

Geekbench 3 Multi Core

AMD Ryzen 5 2600X

22745 (+25.6%)

Intel Core i7-7700

16932

Cinebench R111.5

AMD Ryzen 5 2600x

2.08

Intel Core i7-77700

281 C. R11.5

AMD Ryzen 5 2600x

14.4 (+31.2%)

Intel Core i7-7700

9,0003

Passmark

AMD Ryzen 5 2600x

14087 (+38. 8.8%) (+38.8.8%) 7700

8618

Features

The data is not yet filled in, so the tables may be missing information or missing existing functions.

Main

Manufacturer	AMD	Intel
DescriptionInformation about the processor, taken from the official website of the manufacturer.		Intel® Core™ i7-7700 Processor (8M Cache, up to 4.20 GHz)
ArchitectureCode name for the microarchitecture generation.		Kaby Lake
Process The manufacturing process, measured in nanometers. The smaller the technical process, the more perfect the technology, the lower the heat dissipation and power consumption.	No data	14
Release dateMonth and year of the processor’s availability.	09-2020	09-2019
Model Official name.		i7-7700
Cores The number of physical cores.	6	4
ThreadsNumber of threads. The number of logical processor cores that the operating system sees.	12	8
Multi-Threading Technology With Intel’s Hyper-threading and AMD’s SMT technology, one physical core is recognized as two logical cores in the operating system, thereby increasing processor performance in multi-threaded applications.	SMT (note that some games may not work well with SMT, for maximum FPS, you can disable the technology in the BIOS of the motherboard).	Hyper-threading (note that some games may not work well with Hyper-threading, you can disable the technology in the BIOS of the motherboard for maximum FPS).
Base frequencyGuaranteed frequency of all processor cores at maximum load. Performance in single-threaded and multi-threaded applications and games depends on it. It is important to remember that speed and frequency are not directly related. For example, a new processor at a lower frequency may be faster than an old one at a higher one.	3.6 GHz	3.6 GHz
Turbo frequencyThe maximum frequency of one processor core in turbo mode. Manufacturers allow modern processors to independently increase the frequency of one or more cores under heavy load, due to which performance is noticeably increased. It may depend on the nature of the load, the number of loaded cores, temperature and the specified limits. Significantly affects the speed in games and applications that are demanding on the frequency of the CPU.	4GHz	4. 2 GHz
L3 cache size The third level cache acts as a buffer between the computer’s RAM and the processor’s level 2 cache. Used by all cores, the speed of information processing depends on the volume.	No data	8
Instructions		64-bit
Extended instruction set Allows you to speed up calculations, processing and execution of certain operations. Also, some games require instruction support.		SSE4.1/4.2, AVX 2.0
Embedded Options Available Two housing versions. Standard and designed for mobile devices. In the second version, the processor can be soldered on the motherboard.	No	No
Bus frequency The speed of communication with the system.		8 GT/s DMI3
Number of QPI links		No data
TDPThermal Design Power is an indicator that determines heat dissipation in standard operation. The cooler or water cooling system must be rated for a larger value. Remember that with a factory bus or manual overclocking, TDP increases significantly.	No data	65
Cooling system specifications		PCG 2015C (65W)

Video core

Integrated graphics core Allows you to use your computer without a discrete graphics card. The monitor is connected to the video output on the motherboard. If earlier integrated graphics made it possible to simply work at a computer, today it can replace budget video accelerators and makes it possible to play most games at low settings.		Intel® HD Graphics 630
GPU base clockFrequency in 2D and idle.	No data	350
Maximum GPU frequencyMaximum 3D frequency.	No data	1150
Intel® Wireless Display (Intel® WiDi) Supports Wireless Display technology over Wi-Fi 802.11n. Thanks to it, a monitor or TV equipped with the same technology does not require a cable to connect.
Supported monitorsThe maximum number of monitors that can be connected to the integrated video core at the same time.		3

RAM

Maximum RAM Amount of RAM that can be installed on the motherboard with this processor.	No data	64
Supported type of RAM The type of RAM depends on its frequency and timings (speed), availability, price.		DDR4-2133/2400, DDR3L-1333/1600 @ 1.35V
RAM Channels The multi-channel memory architecture increases data transfer speed. On desktop platforms, two-channel, three-channel and four-channel modes are available.		2
RAM bandwidth
ECC memory Support for error-correcting memory that is used in servers. Usually more expensive than usual and requires more expensive server components. However, second-hand server processors, Chinese motherboards and ECC memory sticks, which are sold relatively cheaply in China, have become widespread.	No data	No data

PCI

PCI-E The PCI Express computer bus version. The bandwidth and power limit depend on the version. There is backward compatibility.	No data	3
PCI configuration options
Number of PCI lanes	No data

Data Security

AES-NI The AES command set extension speeds up applications that use appropriate encryption.	No data	No data
Intel® Secure Key An RDRAND instruction that allows you to create a high performance random number generator.	No data	No data

Decoration

Dimensions	No data	No data
Supported sockets	No data	No data
Maximum processors per motherboard	No data	No data

Which is better

AMD Ryzen 5 2600X

• On average, gaming performance is 1% better.
• The speed of work in office applications and browsers is increased by 1%.
• In complex multi-threaded applications, faster and outperforms by 7%.
• $30 less price
• 2 more physical cores.
• 4 more threads.

How are they similar

• The base frequency is equal.

Testing AMD Ryzen 5 2600X and Ryzen 7 2700X processors (Pinnacle Ridge) earlier, so it’s time to update the top models of both families. Intel managed to accustom users to annual «radical changes» (at least nominally), and with a platform change at every second step — AMD at one time approached the issue a little less radically, but it seems. For example, Zambezi was released on AM3+, and a year later Vishera, on FM2 — Trinity and Richland, as part of FM2+, the company «marked» Kaveri and Godavari. What do each couple have in common? The technical process and microarchitecture — but with some modifications of the latter and the results of debugging the production process. I didn’t feel like global changes — in fact, a new stepping of crystals, taking into account evolutionary debugging during the year.

Pinnacle Ridge (Ryzen 2000 series) and Summit Ridge (CPUs of the 1000 family already familiar to us) are in a similar relationship with each other. However, in their case, the technical process has officially changed from 14 nm to 12 nm, and the microarchitecture is now called Zen + instead of just Zen. However, what it gives in practice — in practice it is necessary to check. The influence of production technology numbers is especially interesting — all semiconductor manufacturers have long had nanometers of their “own system”, and the example of TSMC is too fresh in memory, in which the density of transistors at formal 16 and 20 nm differed by less than 5% (at the same time, with on the other hand, 14 nm Samsung and 14 nm Intel differ by more than 20%). One thing is for sure: something has changed, since the “ceiling” of the clock frequency has noticeably moved away from the former 4 GHz: earlier this value had to be stormed by extreme methods, and new models sometimes go abroad and in normal mode. But what happens to power consumption in this case — you need to check: the TDP of the top model of the new family has been officially increased compared to its predecessor. In addition, the “peripheral component” of the SoC and the memory controller have been upgraded — now they correspond to “Ryzen with graphics”.

Should all of the above be taken positively, or vice versa? A priori — neutral: the situation on the market a year ago and now is very different. Then AMD should have shown a miracle — the company eliminated competition in the upper segment for several years, concentrating on Zen, and it also lingered for several months relative to the original plans, so no one would forgive her mistakes. However, there were no mistakes: in fact, new processors began to easily compete with Intel’s developments for LGA2011-3 at a price of $600 — despite the fact that even the older Ryzen 7 was priced at only $500, and even motherboards for the AM4 platform could be significantly more cheap. In the autumn, however, Intel recovered from the blow by launching the first representatives of the Coffee Lake family on the market, but here it was not without some rough edges. The company had to urgently release a new LGA1151 platform, initially almost identical in terms of consumer characteristics to the old LGA1151 — but new, incompatible with the old one. And the announcement of processors and chipsets for it lasted for several months, ending just a few days ago. Accordingly, AMD did not need to rush to respond — the company simply reduced prices for the existing Ryzen 5 and Ryzen 7 models around the New Year, since the stock for this was initially laid down. From the new AMD models, «feats» are no longer required — they just need to be a little better than the old ones for the same money . Moreover, the company already has Ryzen Threadripper on the floor above, for which Intel has not presented a clear alternative (the LGA2066 platform in the “desktop” version looks too, to put it mildly, strange, and it costs much more), and arrange intra-company competition there is no need for their AMD products.

It would also be useful to immediately mention that money is also not quite “the same”. And in both senses 🙂 The recommended prices of new processors are slightly higher than those of their predecessors, but literally by 10-20 dollars, which is logical: the processors are fully compatible, so you need to «help» partners get rid of stocks. However, now all “boxed” models also include effective coolers, and earlier this was true only for Ryzen 5 1600 and younger. It is clear that the absence of a cooler in the kit of any Ryzen 7 did not bother real enthusiasts (this category of users still prefers a separate approach, often paying more for the cooling system than middle-class processors cost), however, for the broad masses of working people its availability is more convenient. Those who don’t need a cooler can save money by choosing an OEM version (as before).

This is how it looks in the first approximation. And what exactly has changed for the buyer (and is it always in one direction) — we will now check using a couple of new Ryzen.

Test stand configuration

Processor	AMD Ryzen 5 1600X	AMD Ryzen 5 2600X	AMD Ryzen 7 1800X	AMD Ryzen 7 2700X
Core name	Summit Ridge	Pinnacle Ridge	Summit Ridge	Pinnacle Ridge
Production technology	14 nm	12 nm	14 nm	12 nm
Core frequency, GHz	3. 6/4.0	3.6/4.2	3.6/4.0	3.7/4.3
Number of cores/threads	6/12	6/12	8/16	8/16
L1 cache (total), I/D, KB	384/192	384/192	512/256	512/256
L2 cache, KB	6×512	6×512	8×512	8×512
L3 cache, MiB	16	16	16	16
RAM	2×DDR4-2666	2×DDR4-2933	2×DDR4-2666	2×DDR4-2993
TDP, W	95	95	95	105
Number of PCIe 3.0 lanes	20	20	20	20
Price	ask prices	ask prices	ask prices	ask prices

There were three Ryzen 7 models in the first generation — two remained in the second. This can be explained by the fact that, having no direct competitors among low-cost Intel processors, AMD initially set prices with a fairly wide step of $100, which resulted in three processors: «inexpensive» 1700, «fast» 1700X and «best» 1800X. Moreover, in the last pair, the performance differed by literally 5%, but the power consumption of the top-end device was already 10% lower — thanks to the use of specially selected crystals of the highest quality. Now, it makes sense to “save” those for large assemblies, such as Ryzen Threadripper or server models, since the price niche of desktop has been greatly reduced: $300-$330, and not $300-$500 like last year. But in the Ryzen 5 family there are no special changes, so it remains correct to compare models that differ only in the family number.

Processor	Intel Core i5-8600K	Intel Core i7-8700K
Core name	Coffee Lake	Coffee Lake
Production technology	14 nm	14 nm
Core frequency, GHz	3. 6/4.3	3.7/4.7
Number of cores/threads	6/6	6/12
L1 cache (total), I/D, KB	192/192	192/192
L2 cache, KB	6×256	6×256
L3 cache, MiB	9	12
RAM	2×DDR4-2666	2×DDR4-2666
TDP, W	95	95
Number of PCIe 3.0 lanes	16	16
Price	ask prices	ask prices

With whom? Of course, with Intel processors for the “updated” LGA1151, and the best in the Core i5 and Core i7 lines. Note that only they are supplied with an unlocked multiplier — unlike Ryzen, where it makes sense to choose cheaper models “without indexes” for overclocking. In normal mode, there is direct competition for top modifications. In principle, initially a little “skewed” towards AMD, since the recommended prices for 2600X/2700X are slightly lower than for 8600K/8700K, and there are no problems with the availability of inexpensive motherboards for AM4 for a long time, and “very inexpensive” support overclocking — based on the B350 chipset (but let’s not talk about sad things — the situation with the release of the second generation LGA1151 to the market, as it seems to us, has already bothered everyone), but what is it.

Processor	Intel Core i7-7800X	Intel Core i7-7820X
Core name	Skylake-X	Skylake-X
Production technology	14 nm	14 nm
Core frequency, GHz	3.5/4.0	3.6/4.3
Number of cores/threads	6/12	8/16
L1 cache (total), I/D, KB	192/192	256/256
L2 cache, KB	6×1024	8×1024
L3 cache, MiB	8.25	11
RAM	4×DDR4-2666	4×DDR4-2666
TDP, W	140	140
Number of PCIe 3.0 lanes	28	28
Price	ask prices	ask prices

And to make it not boring, we decided to add a couple of processors for the LGA2066 platform to the number of subjects. Moreover, AM4 is somewhat closer to the Intel HEDT platform than it might seem at first glance — it can also offer the buyer more than six cores and more than 16 PCIe 3.0 lanes, but much cheaper. But older models of Ryzen processors, as well as Core i7 / i9for LGA2066, integrated graphics are dispensed with — so their areas of application will largely coincide in many respects. The more interesting is the direct performance comparison.

All systems were equipped with a GeForce GTX 1070 based graphics card. The RAM in all cases was 8 GB per channel, i.e. its total amount was 16 GB for AM4 / LGA1151 and twice as much for LGA2066 (but this platform is already involved «out of competition»). Its clock speed is the maximum «official» for Intel processors and 2933 MHz for Ryzen. Accordingly, for the 2600X/2700X it is also already official — that is why we did not touch it during the first test: among other things, this makes the comparison with the processors of the previous generation more correct.

Test Method

The method is described in detail in a separate article. Here we briefly recall that it is based on the following four pillars:

• iXBT.com performance measurement methodology based on real applications of the 2017 sample

Processor Power Test Methodology

• How to Monitor Power, Temperature, and Processor Load During Testing
• 2017 Game Performance Measurement Methodology

Detailed results for all tests are available as a complete results spreadsheet (in Microsoft Excel 97-2003 format). Directly in the articles, we use already processed data. This is especially true for application tests, where everything is normalized relative to the reference system (AMD FX-8350 with 16 GB of memory, GeForce GTX 1070 graphics card and Corsair Force LE 9 SSD60 GB) and grouped according to the areas of application of the computer.

In principle, we have already prepared an updated version of the performance measurement methodology, but so far we have not begun to use it — there are more results accumulated from the «old» ones for comparison, and new processors should definitely be included in the final material. And later we will return to them again — especially since by that time there may be some improvements in terms of supporting devices with motherboard firmware (we used for testing a new model based on the X470 chipset, compatible with the 2000 series out of the box ”, but the debugging process may still not be finished yet).

We also decided to do without gaming tests today — any of the processors presented in the review is more than enough to completely «load» the used video card based on the GTX 1070 (and more powerful ones too).

iXBT Application Benchmark 2017

In principle, under loads of this kind, after the price correction, the old six-eight-core Ryzens looked good (and even at the start they had to compete with much slower Intel processors, so there were no complaints about prices too), but the new ones are a little faster. Not enough to call wow effect — but the advantage over competing designs has increased.

In this case, the Ryzen 5 1600X continued to look good in autumn and winter, but the Ryzen 7 lost some of its superiority in speed. Updating this line could return the leadership. Not unconditional like a year ago, but then Ryzen 7 was also more expensive than Core i7 — now we are talking about approximate price parity.

The degree of utilization of multithreading by these programs leaves much to be desired. This became clear after processors with more than four cores began to penetrate into the mass sector — before that tea was hot , and now programmers still need to manage the opportunities that have opened up. But this situation is only beneficial for the new family — even the old models continued to keep at the level of Skylake-X and Coffee Lake that appeared later, and the tweaked microarchitecture and improved memory controller of the new processors simply made them leaders. The Ryzen 5 2600X looks especially good, being somewhere between the Intel Core i7-5960X and i7-6900K in terms of speed: some three years ago, the former was the undisputed leader among desktop (at least conditionally desktop) processors, and both cost about $1000 . It cannot be said that these solutions are morally outdated (if only because nothing much faster has been released for a long time) — just as a result of last year’s «big leap» for such performance in video processing tasks, you will have to pay only about $ 229. Back then, just a decent board for LGA2011-3 could cost more, not to mention the starting prices of DDR4 memory.

As for working with photos, it’s impossible to talk about records here, and the performance increase in the new family is rather modest. On the other hand, it is, so the gap from the Core i7 of new generations (to put it mildly, the not so old Core i7-7700K is even slower — it roughly corresponds to the Ryzen 5 1600X in terms of points) has decreased. And the likelihood of problems such as the identified feature of one of the Photoshop filters we use (previously — we removed it from the new version of the methodology) is still lower. Just because the Ryzen 5 and Ryzen 7 processors are architecturally similar (and the APUs of the first line are similar) and do without any qualitative “cuts” — limited to quantitative ones: in the form of blocking some of the cores. Since 2011, Intel has been using interleaving: two cores, two cores plus SMT, four cores, four cores plus SMT… In laptop models, however, for a long time there were almost no processors without Hyper-Threading — only budget Celeron and Pentium. Under such conditions, the probability of missing some kind of bug when testing a new version of the software, which manifests itself only in part of the configurations (but it is stable — we have already seen that both the new Core i3 and Ryzen 3 “suffer” from it), is quite large. In the end, many families of Intel processors were affected (there were more of them in the «eighth» generation than in the «seventh»), but only budget models for AM4. Special case, of course. But the overall picture is made up of many special cases 🙂

Everything is simple here — not too complicated (for modern architectures) «integer» that scales well in terms of the number of computation threads. Accordingly, Core i5 / i7 and for the “updated” LGA1151 still could not compete even with the old Ryzen 5/7 on an equal footing, taking into account the price. And second-generation processors are even faster — albeit a little. However, it is clear that all this is true considering the price — in terms of abstract performance, Intel’s cores are still more powerful than AMD’s, so with an equal number of them, the processors of the first company will be a little faster in the end. Well, actually, and what of the buyer, who will definitely not reject the price factor?

Archiving was once considered a weak point of Ryzen. Before the advent of the LGA2066 platform, it turned out that Intel generally knows how to sometimes move in reverse . The Ryzen update is a step forward. Not too big, but, again, the Core i7-8700K managed to not only catch up, but also slightly overtake the Ryzen 7 1800X — the 2700X countered this. And the six-core Ryzen 5 looked good before — they got a little better.

An example of applications with good optimization for modern instruction sets, high requirements for the memory system, etc. etc. This manifests itself to varying degrees for everyone, but in general it is observed. As well as the “love” for a large number of physical cores, some are expressed, so here, too, the processors of the Coffee Lake family could not overtake the first generation Ryzen. The second one accelerated by more than 10%, so that the 2600X caught up with the former flagship of the AM4 platform. The i7-7820X remains the absolute leader of the processors presented in the diagram — but it is also the leader in terms of price. And at the cost of the platform too. The Core i7-7800X isn’t cheap either, by the way, but it’s starting to look like 9 every month1825 getting weirder and weirder .

The general picture is as follows. Intel managed to achieve approximate performance parity between the Core i7-8700K and the Ryzen 7 1800X, although in terms of tasks (as we have seen above) the second processor still remained faster — the release of the Ryzen 7 2700X allowed AMD to regain its leadership and «on average «. A little higher than the Core i7-7820X and other solutions for LGA2066, but this is a completely different price range, where AM4, in principle, is no longer obliged to fall — there is already Ryzen Threadripper for this. As for more budget solutions, in principle, nothing needed to be improved here — the 6/12 versus 6/6 formula allowed Ryzen 5 to look more convincing than Core i5 even after updating the latest line. Now, AMD’s leadership has only increased — in fact, the 2600X can be considered a competitor even for the Core i7-8700, although it costs less than the Core i5-8600K.

Energy consumption and energy efficiency

Unfortunately, the energy consumption is not so good — it has increased in the new family.

Moreover, this cannot be “written off” only on the board: on the “processor” line +12 V, the increase in power consumption is even greater. However, some “degree of fault” of the motherboard may also be present: if the default settings are too aggressive, then the power consumption will be higher than it could be. That is why we want to definitely return to this issue — having tested other motherboards, and, possibly, updated firmware.

However, there is a chance that nothing will change: the result can be explained by the fact that all the improvements in the manufacturing process went to increase clock frequencies and, accordingly, increase performance. In terms of energy efficiency (as we already know very well), this always hits — Core processors «suffer» from the same problems. They just have better absolute values ​​- unless, of course, we consider the LGA2066 platform 🙂 As for solutions for AM4, even if there are no improvements, there is nothing wrong with their “appetite”: in the worst case, it turns out that Ryzen 7 2700X in this parameter, it is approximately equal to the FX-8350, but much more productive than the latter — more than twice! And against the background of how much energy a “decent” modern discrete video card consumes, this is also not much, and we recall that there is still no integrated video core in older Ryzen, so APUs are intended for compact low-power systems in the company’s assortment.

Total

So, the new processors of the “2000 family” return the leadership in pure performance to the AM4 platform. More precisely, in some segments it did not lose it even after the appearance of the “updated” LGA1151, but now the parity that has been established for some time has again shifted towards Ryzen 7. As for Ryzen 5, competition was easy for six-core models of this family before — now they got even faster. In addition, do not forget about the cheaper Ryzen 5 2600 and Ryzen 7 2700, which differ from older models only in clock frequencies, but are quite favorable for overclocking (at least to the level of older ones, or even more — this is a separate issue). In any case, Intel offers only three «overclockable» processors within the LGA1151, and only one «full» Core i7: the Core i5-8600K does not have Hyper-Threading support, and the Core i3-8350K has only four processing cores, but even it not radically cheaper than the Ryzen 5 2600. And taking into account the fact that overclocking will require a motherboard based on a top-end chipset (AMD takes a more humane approach), a complete set in the case of Intel will cost more.

Nothing new in all of this, it just retains the merits of the previous Ryzen processor family. And due to the increase in performance and frequency potential, the new processors are even more attractive. A «miracle» was not required this year — just a normal evolutionary improvement of an already successful platform was more than enough. Unfortunately, the improvement was not achieved in all areas — power consumption has grown slightly along with performance. But, as mentioned above, we will definitely return to this issue — this may not be a feature of new processors, but the influence of the environment that can be corrected.

testing and detailed analysis / Sudo Null IT News

With the end of 2017, in which the Ryzen processor became one of the most successful AMD products, a natural question arose: what’s next? At the beginning of 2018, plans were announced: the second generation of Ryzen should appear in the middle of the year, after which the second generation of Threadripper will be released, on the 12nm process of GlobalFoundries. It’s not AMD’s next new microarchitecture yet, which we know will be Zen 2 on the 7nm process. This is a release of components with some improvements, plus the ability to use a manufacturing process that allows you to raise the frequency and performance limit. AMD is launching four processors today, we’ve tested them all.

Straight to the point: new processors

For readers who want to get straight to the point, AMD launches Ryzen 7 2700X, Ryzen 7 2700, Ryzen 5 2600X, and Ryzen 5 2600. at the cost of an additional 10W TDP delivers a base frequency of 3.7GHz and a turbo frequency of 4.3GHz on eight cores with simultaneous multi-threading. This is an additional +100 MHz and +300 MHz, respectively, which is above the average limits of the overclocked 1800X.

Big news: With the 2700X, AMD reduced the maximum price of the top-end AM4 Ryzen processor: the 1800X cost $499 at launch and shipped without a cooler. Yes, the 1800X has recently dropped in price to $349 to compete with powerful Intel processors. The 2700X out-classed both champions by hitting the market with a suggested retail price of $329 and bundled with the best business-class cooler: the AMD Wraith Prism RGB. AMD is pushing hard on all fronts: aggressive pricing, top performance, and the best package, all in one go, in one product.

The Ryzen 5 2600X is a six-core variant of the processor, also with an aggressive frequency strategy: 3.6 GHz and 4.2 GHz turbo. At 95W TDP and an MSRP of $229, it comes bundled with an AMD Wraith Spire cooler, which, again, is quite an impressive stock product.

The Ryzen 7 2700 and Ryzen 5 2600 are 65-watt versions of their X counterparts, offering nearly the same frequencies for $30 less. All processors will support dual-channel DDR4-29 memory33, which is higher than the maximum DDR4-2666 memory support frequency for 2017 Ryzen processors. One of the major changes is that every processor now comes with a cooler, ranging from the Silent 65W Stealth models to the large Prism RGB, each of which is sufficient to keep the processor stable in turbo mode.

AMD’s planned AM4 Ryzen line will now look like this: AMD has overhauled its product line and replaced three past generation products with two new Ryzens, perhaps based on sales figures. As will be seen from our review, the 2700X pushes the current silicon process to the limit.

The full list is ultimately a combination of Ryzen 2000 series processors (new), Ryzen 2000 series APUs, plus a couple of Ryzen 1000 series products. first generation models. So, four new 2000-series processors are now at the top of the list, but AMD often focuses on new products, so over time (quite likely) there will be a few more new 2000-series processors.0003

Other information from today’s review

Of course, our review is just beginning, because we haven’t talked about the options yet. AMD is using GlobalFoundries’ 12nm manufacturing process, and the benefits are clear. There are many firmware improvements, updated features and roles for AMD Precision Boost and XFR technologies that can have a direct impact on performance. There’s a new chipset (along with 30+ motherboards) ready to work with the new line of processors, as well as new or rebranded features like StoreMI. We want to explore how these new products fit into AMD’s long-term plans and whether they generally fit in.

Let’s consider these questions in the article. You can also find test results here.

AMD Ryzen 2000 competitor: Intel’s Coffee Lake

As part of the new product launch, AMD provided extensive information on testing its new chips. From the data presented, it was quite clear that the new processors aimed to compete with the most modern Intel processors: Coffee Lake. This contrasts with last year’s Ryzen 1000 series pitting the octa-core Ryzen 7 1800X against Intel’s 8-core Broadwell-E, around which time Intel upgraded its main line of processors to six high-frequency cores.

As a result, AMD now offers a comparison of the Ryzen 7 2700X with the Core i7-8700K and the Ryzen 5 2600X with the Core i5-8500K. This is an important point — now both major players in the x86 processor market have pitted their latest products against each other head-on. This hasn’t happened for several generations. However, some numbers have remained the same since last year’s release:

• Intel is expected to have an advantage in frequency and IPC
• AMD will not be far behind in frequency and will offer more cores at the same price

AMD’s multi-threaded crown shines particularly in internal testing, but single-thread performance still lags behind the competition. A number of new features for the Ryzen 2000 series should make things right: slightly higher IPC, higher clocks, higher TDP, and an improved dynamic boost model. We’ll take a look at them over the next few pages.

Frequency and number of cores are just part of the equation. The fact that AMD and Intel have different cache models will play a significant role. One of the things we’ll see in this analysis is cache comparisons, and the tweak AMD has made to close the gaps. In terms of price, the AMD Ryzen 7 2700X is cheaper than the i7-8700K, + adds a Wraith Prism RGB cooler that easily replaces a $30-$40 cooler, saving the consumer money.

The Ryzen 5 2600X and Core i5-8600K are more similar than their big brothers. These processors do not differ in the number of cores, although Ryzen 5 has twice as many threads. For any multithreaded workload that is able to use concurrent multithreading, this is probably an important metric. The Core i5-8600K has a slightly higher core clock and, as expected, an IPC advantage. Again, AMD offers a processor equipped with a good cooler, while Intel’s offer is a bare processor.

Overall, AMD says its new high-end processors will perform within 1-2% of the competition at 1440p gaming, but deliver a 20% boost in «creative performance.» We have several ways to test this.

Articles in this review:

1. AMD launches Ryzen 7 2700X, Ryzen 7 2700, Ryzen 5 2600X and Ryzen 5 2600
2. Speaking of 12nm and Zen+
3. Cache hierarchy improvements
4. Translated into IPC (instructions per cycle): all this for the sake of 3%?
5. Precision Boost 2 and XFR2: Need more hertz
6. New X470 chipsets and motherboards
7. StoreMI way to faster JBOD
8. Test parameters
9. CPU system tests
10. CPU Rendering Tests
11. CPU Web Tests
12. CPU Encoding Tests
13. CPU Office Tests
14. CPU Legacy Tests
15. Gaming Performance: Civilization 6
16. Gaming Performance: Shadow of Mordor
17. Gaming Performance: Rise of the Tomb Raider
18. Gaming Performance: Rocket League
19. Gaming Performance: Grand Theft Auto
20. Findings: the burden of competition

Speaking of 12nm and Zen +

One of the highlights of the launch of the Ryzen 2000 series is that these processors use the GlobalLoundries 12LP manufacturing process, following the 14LPP process of the first generation of Ryzen processors. Both AMD and GlobalFoundries discussed process differences, but it’s worth understanding that the companies’ goals are different: AMD should only promote what helps its products, while GlobalFoundries is a large semiconductor manufacturer with a large «customer base» and can provide numbers and «ideal scenario» data. This year we were invited to GlobalFoundries Fab 8 (Upstate New York) where we were able to interview Dr. Gary Patton, CTO.

Several interesting points were covered in this interview. First, the CTO doesn’t necessarily have to care what certain processes are called: their customers know the performance of a given process regardless of the advertised «nm» figure based on the development tools provided to them. Second: 12LP is just a slightly improved 14LPP process — minor changes to improve performance. The upgrade is the result of a partial optical reduction and a slight change in the rules of production in the back-end and middle part of the production process. In the past, these changes may not have been such a big deal, but GF customers want to take advantage of the improved process.

Overall, GlobalFoundries stated that its 12LP process provides a 10% improvement in performance and a 15% improvement in cell density compared to 14LPP.

This has been interpreted in various ways, such as an additional 10% frequency at the same power, or lower power for the same frequency, or as an opportunity to create smaller chips.

As part of today’s launch, AMD clarified what the transition to the 12LP process meant for the Ryzen 2000 series:

• Increase in maximum clock frequency by ~ 250 MHz (~ 6%)
• Operation of the cores in turbo mode at a frequency of 4.2 GHz
• Core voltage reduction by ~ 50 mV

AMD patiently explains that at the same frequency, the new Ryzen 2000 series processors consume 11% less power than the Ryzen 1000 series, which means +16% performance at the same power. Still, the statements are a bit ambiguous, as AMD has other new technologies in the 2000 series that will impact performance.

One interesting point is that while GF claims a 15% density improvement, AMD claims these processors have the same die size and transistor count as the previous generation. Ultimately, this seems to be counterintuitive — wouldn’t AMD want to use smaller dies to fit more chips on a wafer?

Ultimately, the new processors are almost exact copies of the old ones, both in terms of design and microarchitecture. AMD calls the design of the cores Zen+ to distinguish them from the design of the previous generation Zen, mainly due to the way the microarchitecture features are laid out on silicon. Many of the key features haven’t changed — they just take up less space, leaving empty silicon between the elements.

Here is a very rough representation of the functions bound to the data path. On the left is the 14LPP design, and each of the six functions is sized and connected to the bus. Between the cells is «dark silicon» — unused silicon that is either seen as useless or can be used as a thermal buffer between high-energy cells. On the right is a representation of the 12LP design: each of the functions has been reduced in size, simply leaving «dark silicon» between the elements (white squares show the original size of the function). In this context, the number of transistors has not changed, as has the die size. But if somewhere in the design there were thermal restrictions due to the proximity of the “hot” elements, now there are more distances between them so that the elements do not interfere with each other.

For reference, AMD lists the dimensions of these new processors as 213 mm2 containing 4.8 billion transistors, identical to the first generation silicon design. AMD has confirmed it is using 9T transistor libraries as it did with the previous generation, although GlobalFoundries also offers a 7.5T design.

So, Zen+: a new microarchitecture or a change in the technology node?

Ultimately, nothing about much of Zen+’s physical design is new. Aside from the change to the manufacturing process node and likely minor adjustments, the main improvements are in firmware and support:

• Cache latency adjustment resulting in +3% IPC
• DRAM overclocking support for DDR4-2933
• Improved voltage/frequency curves resulting in +10% performance overall
• Improved performance with Precision Boost 2
• Better thermal response with XFR2

Cache Hierarchy Improvement

The biggest internal change for the Ryzen 2000 series processors is the reduction in cache latency. AMD claims that they were able to remove one cycle from the L1 and L2 caches, several cycles from the L3 caches and improve DRAM performance. Since net base IPC is closely intertwined with caches (size, latency, bandwidth), these numbers lead AMD to claim that the new processors can deliver +3% IPC gains over the previous generation.

Numbers provided by AMD:

• 13% improvement in L1 latency (1. 10ns vs 0.95ns)
• 34% improvement in L2 latency (4.6ns vs. 3.0ns)
• 16% improvement in L3 latency (11.0ns vs. 9.2ns)
• 11% improvement in memory latency (74ns vs. 66ns on DDR4-3200)
• Support for increased DRAM frequency (DDR4-2666 vs. DDR4-2933)

Interestingly, in the official presentation, AMD mentions latency as measured in time, although in private conversations at our briefing, this was discussed in terms of clock cycles. Ultimately, latency as measured by time can take advantage of other internal improvements; however, a real engineer prefers to discuss clock cycles.

Naturally, we’ve looked at two aspects of this equation: are cache numbers really lower, and will we get an IPC boost?

So what about the cache?

For testing, we use a memory latency checker at each step of the single core cache hierarchy. For this test we used the following:

• Ryzen 7 2700X (Zen+)
• Ryzen 5 2400G (Zen APU)
• Ryzen 7 1800X (Zen)
• Intel Core i7-8700K (Coffee Lake)
• Intel Core i7-7700K (Kaby Lake)

The most obvious comparison is between AMD processors. Here we have the first-series Ryzen 7 1800X, the Ryzen 5 2400G APU that fuses Zen cores with Vega graphics, and the new Ryzen 7 2700X processor.

This graph is logarithmic on both axes

This graph shows that the latest Ryzen 7 2700X requires fewer core cycles in each cache design phase. The biggest difference is in the latency of the L2 cache, but L3 also has a significant gain. The reason why the L2 gain is so big, especially between 1800X and 2700X, is quite curious.

When AMD first launched the Ryzen 7 1800X, L2 latency was tested and found to be 17 cycles. This was quite a lot — it turned out that the engineers initially assumed that the L2 latency would be 12 cycles, but the lack of time to configure the firmware and linking before sending the project to production forced 17 cycles to be left as the best compromise, so that the design would be workable and didn’t cause problems. With Threadripper and Ryzen APUs, AMD tweaked the design enough to achieve L2 latency of 12 cycles, however, this fact was not publicized at the time, despite the benefits it provides. Now with the Ryzen 2000 series, AMD has reduced the latency to 11 cycles. We were told that this was due to both the new manufacturing process and additional tweaks to ensure signal consistency. In our testing, we actually saw an average L2 latency of 10.4 cycles, up from 16.9cycles on Ryzen 7 1800X.

The difference in L3 latency is a little unexpected: AMD claimed a 16% reduction in latency from 11.0 ns to 9.2 ns. We saw a change from 10.7 ns to 8.1 ns, which means a decrease from 39 to 30 cycles.

Of course, we couldn’t do without comparing AMD with Intel. And the comparison turned out to be very interesting. Now the cache configurations between the Ryzen 7 2700X and the Core i7-8700K are different:

AMD has a larger L2 cache, however the AMD L3 cache is not an inclusive victim cache, which means it cannot use prefetching, unlike from the Intel L3 cache.

The result was unexpected, as it became clear that AMD has a latency advantage in L2 and L3 caches. There is a significant difference in DRAM, however, the main performance indicators here are in the lower caches.

We can expand the test to include three AMD chips, as well as Intel Lake Lake and Kaby Lake cores.

This is a graph using cycles, not time delay. Intel has a slight advantage in L1, but the larger L2 caches in AMD’s Zen designs mean that Intel will reach higher L3 latency sooner. Yet Intel gets the job done quickly with DRAM’s low latency.

Translated into IPC (instructions per cycle): all this for 3%?

Contrary to popular belief, increasing IPC is a very difficult task. Trying to ensure that every port is involved in every cycle requires wide decoders, large instruction queues, fast caches, and proper execution port configuration. This may seem easy to compile, but both physics and economics say no: the chip still needs to be thermally efficient, and it needs to make money for the company. Each processor design update will focus on what is called «low-hanging fruit»: small changes that bring the most gain for the least amount of effort. Usually, reducing cache latency is not the easiest task, and for non-semiconductor engineers (myself included), it sounds like a lot of work for little gain.

We use the following rules to test IPC. Each processor allocates four cores with no additional threads, and the power modes are disabled so that the cores only run at a certain frequency. DRAM is configured as officially supported by the processor, so in new processors it is DDR4-2933, and for the previous generation DDR4-2666. There has been a recent debate about whether this is fair or not, and here is the opinion: this is an IPC test, not a system efficiency test. Official DRAM support is part of the hardware specification, as is the size of the caches or the number of execution ports. Running two processors at the same DRAM frequency gives one of them an unfair advantage: it’s either a big overclock/underclock or a deviation from the intended design.

So, for the test, I took the new Ryzen 7 2700X, the first generation Ryzen 7 1800X and pre-Zen Bristol Ridge based on the A12-9800, based on the AM4 platform and using DDR4. We run each processor on four cores, no multithreading, at 3.0 GHz. Let’s get to the tests.

In this graph, we used the first generation Ryzen 7 1800X as a 100 percent marker and the blue bars are as Ryzen 7 2700X. The problem with trying to determine a 3% increase in IPC is that 3% can easily get lost in the noise of the benchmark run: if the cache isn’t fully set before startup, we can experience different performance. As shown above, more tests fall within the ±2% range.

However, when calculating heavy tasks, the advantage was 3-4%: there are Corona, LuxMark, CineBench and GeekBench. We didn’t include GeekBench subtest results in the chart above, but most of them show 2-5% gains.

If we take the Cinebench R15 nT result and the Geekbench memory benchmarks, the average gain across all benchmarks is +3.1% for the new Ryzen 2700X. Sounds like coins to AMD.

Going back to the Cinebench R15 nT result, which saw a 22 percent gain, we also had a few other IPC tests done at 3. 0 GHz but with 8C/16T (which we couldn’t compare to Bristol Ridge), and some other tests also showed a 20%+ gain. This is likely a sign that AMD has also adjusted the management of concurrent multithreading. This issue requires further testing.

10% overall improvement

Given the benefits of the new 12LP manufacturing process, we wondered why AMD didn’t redesign some elements of the microarchitecture to achieve an even better result. Ultimately, it turns out that the «free» frequency increase is fairly easy to transfer to the same design (as mentioned earlier, the 12LP design is based on the 14LPP with performance improvements). In the past, such a solution might not have been listed as a separate product line. So pushing a product on the same design is an easy win, allowing teams to focus on the next major core redesign.

To sum it up, AMD has previously stated its intentions for Zen+ Core — at CES earlier in the year, AMD said it wants Zen+ and future products to go beyond the «industry standard» of 7-8% performance each year.

Clearly 3% IPC is not enough, so AMD combines the performance gain with +250MHz frequency increase, which is about another 6% peak frequency gain, with better turbo performance with Precision Boost 2 / XFR 2 This is about a 10% increase, but at least on paper. Let’s see what the tests say.

Precision Boost 2 and XFR2: Need more hertz

One of the biggest changes in the new Ryzen-2000 series is the implementation of the CPU turbo mode. Up to this point (with the exception of the recent launch of the APU), processors have relied on step-by-step implementation of the function: the system determines how many threads are loaded, tries to implement a certain frequency on those cores if possible, and then consults a look-up table of the ratio of the number of threads to frequency. The goal of AMD Precision Boost 2 is to make this process more dynamic.

AMD’s slide shows this feature: the system will determine how much performance headroom is still available and turbo the processor as much as possible until it hits one of the limiting factors. These factors can be any of the following (although not limited to):

• Total peak power of the

chip

• Individual voltage / frequency response
• Thermal interactions between neighboring nuclei
• Power limits for individual cores / groups of cores
• General thermal data

The new AMD Ryzen Master 1.3 software used on the Ryzen 2000 series processor has several indicators to determine marginal factors. For the most part, the way the processor turbos and responds to the environment will be transparent to the user.

The best way to test this in action, from my point of view, is to look at the power consumption of the first and second generation Ryzen processors. We can look at the internal calculated power consumption of each core individually, since, fortunately, AMD left these registers open, and we got the following data:

This is only the power consumption of the cores, not the entire processor, which will include the DRAM controller, Infinity Fabric, and processor IO. This means we’re getting numbers that are different from the nominal TDP, but the danger here is that the Ryzen 7 2700X has a 10W TDP higher than the Ryzen 7 1800X where the 2700X draws more power and it might seem like the answer to TDP .

Plotting energy consumption gives the following picture:

Even so, it’s clear that the Ryzen 7 2700X draws more power, up to 20W, with varying thread counts. Let’s change the graph as a function of peak power:

The results are not so clear anymore: the 1800X seems to consume more as a percentage of maximum power at low threads, but the 2700X consumes more at medium threads.

It’s worth noting that the end result of Precision Boost 2 has two sides: higher performance, but also higher power consumption. Users who want to fit a low power processor into a small form factor system may want to disable this mode and revert to the standard step function for thermal control.

Note — Although the marketing name is Precision Boost 2, the internal function name in the BIOS is «Core Performance Boost». It is similar to Multi-Core Enhancement, which is a feature of some Intel motherboards designed to go beyond the limits of the processor’s turbo mode. However, this is just AMD’s standard PB2: when disabled, «Core Performance Boost» will disable PB2. We initially disabled it thinking it was a motherboard manufacturer’s tool to do some clean tests. Seems like a weird disagreement between AMD engineers and marketing.

Extended Frequency Range 2 (XFR2)

For the Ryzen 2000 series, AMD has changed how XFR works. In the previous generation, it was used on some processors, allowing them to exceed the maximum turbo frequency when the thermal situation favors higher frequencies and higher voltages in low thread states. For the new generation, XFR is still related to thermal, but now applies to any core load: if the processor temperature is up to 60ºC, the frequency can increase regardless of the maximum frequency of Precision Boost 2 (so why not get more from using PB2?). However, the core still needs to be in a suitable voltage/frequency range to maintain stability.

Some motherboards, such as the ASUS Crosshair VII Hero, have additional options to support XFR2 outside of the AMD implementation. ASUS doesn’t go into specific details, however I suspect it’s implementing a more aggressive version, perhaps by widening the voltage/frequency curve, raising the power limits, and/or adjusting the thermal limit.

New X470 chipsets and motherboards

Focus on power

For our AMD product reviews, we received two motherboards: ASUS ROG Crosshair VII Hero (Wi-Fi) and MSI X470 Gaming M7 AC. These are two high-end motherboards based on the new X470 chipset.

Katamari loves motherboards. Or just likes to sit on review samples

The new X470 chipset should take its place above the X370 chipset, although looking at the specs users might not even notice the difference. Technically, the X470 has the same PCIe and SATA support as the older X370 chipset, and AMD will be rolling out both chipsets at the same time to major motherboard manufacturers for a while. Both motherboards will use socket AM4, which AMD has supported for generations.

The main changes in the chipset relate to power consumption. Currently, the X370 chipset, built on a 55nm manufacturing process using ASMedia IP, runs at 6.8W TDP (at full load). For the X470, we’re told it’s the same process and IP, but the chip will now consume 4.8W peak and 1.9W idle. This is due to the improved power infrastructure within the chip, and AMD also claims that the overall throughput is improved. The chipset firmware is also tuned to provide better overclocked memory support and stability.

The next important component is StoreMI, which we will devote the next article to. This new option does not technically require chipset support, however the installer checks for the X470 chipset before granting a free license, otherwise the software will cost $20 and will not be AMD branded.

All X470 boards and X370 boards with the latest BIOS updates will support the new Ryzen 2nd Gen processors. Newer X370 motherboards that already have the BIOS updated will have the «Ryzen 2000 Desktop Ready» logo on the box, but the X470 motherboards support the new processors anyway.

Boot Kit AMD

For buyers of X370/B350/A320 motherboards with older firmware, AMD offers to resolve the issue through the support page. Consumers should first attempt to replace the board with a new BIOS-updated board from a retail vendor, but if unsuccessful, users with registered purchases may receive an «AMD Boot Kit» — an A-series processor on short-term rental that can update the BIOS for their new processor.

AMD will provide the kit for free if the user:

• Shows a photo of the new 300 series motherboard,
• Show photos of the new Ryzen 2000 series processor,
• With model numbers / unique serial numbers in frame and,
• A copy of the purchase invoice.

This means that users who want to take advantage of the Boot Kit will have to buy both retail and used components. The kit includes an A-series processor (Bristol Ridge) and a cooler, as well as a pre-paid return label. Such a generous motherboard BIOS update kit is an offer unheard of before. Prior to this, users had to resolve the issue through the seller and pay for the RMA. However, it can be assumed that AMD has enough A-series processors that this is not a problem, and the positive response from such a service outweighs the shipping and return costs.

Users who need a Boot Kit can follow this link to find the official data.

Motherboards X470

Each manufacturer has already announced several new motherboards for the chipset, although it is clear that this is far from a complete stack.

Most sellers will offer the X470 and X370 at the same time, and the X470 will fill the premium product niche.

ASUS ROG Crosshair VII Hero

The first motherboard we opened was ASUS ROG Crosshair VII Hero (Wi-Fi). The packaging turned out to be pretty shabby — either during delivery or during storage.

At first glance, we see some great options: a combined 12-phase power supply (probably 10 + 2) on the CPU, with two M.2 slots and enhanced PCIe to work with SLI x8 / x8. The rear I/O panel is pre-attached to the system with a shroud, and there is a small cable connecting the onboard LEDs on the motherboard.

Socket unchanged: AM4 with 1331 CPU holes. The latch mechanism is the same as the dimensions of the cooler.

ASUS has added a number of RGB connectors to the board, as well as what looks like handy voltage reading points (or points that allow you to use any 5V system, like a cold cathode backlight?).

Of the two M.2 sockets, one is PCIe 3.0 x4 from the CPU and the other is PCIe 2. 0 x4 from the chipset. There are also six SATA ports for connecting storage devices.

The board has various USB 3.1, USB 3.0 and USB 2.0 ports, although I found it rather funny that ASUS chose to label the ports «Native USB» to emphasize that this refers to the chipset and not the controller. This makes sense for VR, which requires its own ports, allowing the user to install front panel connectors into native USB connectors.

To make the premium board shine even brighter, ASUS has placed its SupremeFX sound card. It is based on a custom Realtek ALC1220A codec, Nichicon audio capacitors, EMI shields, PCB decoupling, and a software package.

On the back we have: ASUS BIOS Flashback button so that users can update BIOS without installing CPU/GPU/DRAM; Clear CMOS button; 802.11ac WiFi; two USB 2.0 ports; PS/2 combo, eight USB 3.0 ports, two USB 3.1 ports (one Type-C), Gigabit Ethernet port, and audio jacks.

MSI X470 Gaming M7 AC

Unlike the previous one, the packaging of the MSI X470 Gaming M7 AC was not damaged, and showed the image of the motherboard directly on the front side. Usually we are used to seeing the names «MSI» and «ACK» on Gaming M7 Wi-Fi motherboards, indicating the use of a Killer network controller, but not here.

At first glance, the motherboard looks less focused on «style» than ASUS, although MSI also catches the eye. The obvious features are DRAM support, plenty of power chokes, and a U-shaped heatsink that hides two M.2 slots.

A close look at the DRAM sockets shows that this is how MSI is developing its «reinforced memory slot» concept. We can argue whether the above concept makes sense (it certainly helps PCIe), but there is clearly room for aesthetics here.

I counted 14 chokes on this motherboard and this is possibly the largest power delivery option on any AM4 motherboard. The power heatsinks are not connected together, which may indicate the cost, or MSI’s confidence in power delivery efficiency. It’s worth noting that MSI uses an 8-pin power connection to the CPU here, compared to the 8+4-pin arrangement on the ASUS ROG.

One of the more esoteric features of MSI’s latest motherboards is this large knob, with numbers up to 11. This is MSI’s Game Boost overclocking feature, designed so that each turn gives a greater level of overclocking to the processor. Previously, such a function would be too heavy for most processors, in an attempt to increase the frequency as much as possible, previously we could never go beyond «2» with air cooling. Luckily, there are power off/reset buttons next to the handle.

MSI Audio Boost 6 branded audio is not much different from the competition: Realtek ALC1220 codec with dedicated audio capacitors, EMI shield and PCB connector. MSI adds software licensed from Nahimic that offers various EQ settings and additional benefits for gamers.

Like ASUS, MSI offers BIOS update function without CPU/GPU/DRAM installed. On the rear panel we see two USB 2.0 ports: a PS / 2 port, four USB 3.0 ports, a Wi-Fi 802.11ac module, two USB 3.1 ports, a Gigabit Ethernet port and audio jacks.

A few interesting notes about the back of the board — near the chipset heatsink area, MSI posted a warning that the standoff screws should not roll around the motherboard. Since most cases are designed to accommodate any motherboard form factor, users who use older cases and do not remove unnecessary support screws can cause a short circuit and possibly destruction of the hardware. Although, unless the user removes the old supports, I doubt they will bother to read the text on the back of the motherboard.

Here’s a nice addition to the motherboard: the user is told the number of PCB layers. In this case, six. To be clear, the cheapest motherboards usually have three or four layers, most mainstream motherboards will have six or sometimes eight, and high-end desktop motherboards usually have eight or ten. Motherboards for which price is not an argument, such as server systems, may have twelve. Additional options such as double copper or high humidity protection can add +50% to the cost of a bare PCB.

StoreMI way to faster JBOD

When AMD released the Ryzen APUs earlier this year, one of the minor announcements was the promotion of a software called FuzeDrive by Enmotus. The software was offered to consumers for an additional $20. For the new Ryzen-2000 series desktop processors and the X470 platform, this software has become part of the AMD software stack and can be downloaded free of charge from the AMD website under the StoreMI brand.

Cache benefits

The way StoreMI works is that a user can take a mid-range system running on a slow drive and add a small but fast drive to speed up their most frequently used files. The software creates a «storage system» from a fast drive and a slow drive, giving the user one drive with a total capacity of two drives, while the software implements pattern recognition to understand which files to place on the fast drive for maximum system speed.

Tiered storage is not new — it is used in many enterprise storage systems supported by complex software. Fast storage systems are somewhat expensive, while the volume of data is usually large: content delivery networks (CDNs) such as Netflix or Steam use tiered storage and caching. Thus, frequently used movies or games are available from storage that is near the user and at the same time is the fastest available.

In today’s computer, the fastest storage is built-in memory/DRAM. Files and programs are loaded here when the processor needs to access data or keep data «close» for constant access. This has been the case throughout the history of software. In recent years, some enthusiasts have used RAMDisks, creating disk space using RAM to get a fast repository. The only downside to this method is that data is lost on reboot because the data stored in DRAM is volatile (or not persistent).

The modern enthusiast’s machine is more likely to include some sort of solid state drive (SSD) based on NAND flash, either a drive connected via PCIe lanes as super-fast storage, or connected via a SATA port. These same users often have a rotating platter-based mechanical hard drive as back-up storage for large volumes, commonly referred to as a hard disk drive (HDD), and the speed is limited by how the system reads data from a disk that spins 7200 or 5400 times. per second. Many machines still ship with HDDs as primary and mass storage, which is a big concern for users who are attracted to the benefits of using an SSD.

With StoreMI, the user can use any configuration of PCIe SSD, SATA SSD or HDD and implement it as a tiered drive. The combined single drive will have the capacity of all the drives in the stack, and the software will manage what data should be moved to the fast drive. This process is gradual, it will take time for the software to learn which files are the most important — which means that the effect will not be noticeable immediately, but after the third or fourth download of the software or game, the system should show good results.

The StoreMI tool also allows the user to add up to 2 GB of RAM to the storage stack. This process does not add any extra capacity to the «single disk», but DRAM acts as the fastest cache and will only store copies of data stored on other disks to avoid data loss. As mentioned above, since the built-in memory is not stable, the in-memory cache for speedup will be lost on restart. In a conversation with AMD, we were told that 2 GB is a good amount of memory for a DRAM cache: due to the nature of the software, a larger cache will not show a measurable increase, according to the campaign. The software allows even systems with as little as 8 GB of total system memory to use the software.

Configurations

Unlike RST caching technology (which recently supported caching on non-boot drives), AMD’s StoreMI can be used at any time in the system’s lifecycle. For any user who wants to delay purchasing an NVMe or SATA SSD, or delay purchasing a larger drive, there is an option to take the time to add the new drive to tiered storage at a later date.

StoreMI can support almost any configuration, including boot drives and storage drives. When working with a boot drive, AMD recommends that you first install the operating system on a slower, larger drive and then add a blank SSD as a quick access tier, although the opposite is also possible for users who want to add a larger drive later (there may be additional steps to end of the process). The only difference is that the software will most likely move a large amount of data at the start of the job.

Configurations offered for StoreMI:

• HDD + DRAM
• HDD + SATA SSD
• HDD + SATA SSD + DRAM
• HDD + NVMe SSD
• HDD + NVMe SSD + DRAM
• SATA SSD + DRAM
• SATA SSD + NVMe SSD + DRAM
• NVMe SSD + DRAM

The biggest noticeable improvement should come in configuration 4 when the NVMe SSD is paired with a mechanical hard drive.

AMD states that if the storage stack extends from the native chipset to the controller-based SATA ports, the software will likely move sleep-related files to the drive on the native SATA ports for stability; this process can take up to 30 minutes.

Users can also remove drives from tiered storage if there is enough space to fit all the data on the drive that remains in the tier. The deleted disk will remain with zero data and can be removed from the system or used for other purposes.

Severe limitation: 256 GB on Fast Tier

An important point, not mentioned in our initial briefings when FuzeDrive launched with the APU, but made clear in the StoreMI user guide, is that AMD and Enmotus’ joint license stipulates that a faster drive in a tier can be no larger than 256 GB.

When you add a drive larger than 256 GB as a shortcut tier, the system automatically partitions the empty drive, offering additional capacity as a separate drive letter.

Adding a large HDD as a slow tier to an SSD boot disk will only cause problems if the SSD is larger than 256GB. Users in this case will need to migrate the operating system (using other software) from the SSD to the HDD, then boot the system from the hard drive, and add the (now empty) SSD as a fast access layer.

Some users may think that this is too much effort — adding a 3GB hard drive to a 512GB SSD shouldn’t be that hard. The solution is to leave the drives partitioned and manually move things like the Steam folder to different drives, leaving your favorite games on the SSD. However, StoreMI is more focused on systems that have a large hard drive out of the box — to add a small, fast SSD, say 64GB-128GB. Such a case is considered as the most probable.

Failure rate: JBOD disadvantages

Most options for combining disks in an array modify the element of speed (reading data on many disks at the same time), and protection against failures (data is replicated or a parity bit is introduced). Typically, storage array options combine both features, while focusing on one feature at the expense of the other. For users who need to store files, one option for bundling disks is known as JBOD or «just a bunch of disks». JBOD offers neither extra speed nor crash protection.

The JBOD array does one thing: it combines the drives into an array to show them as one shared file space, and treats the array as such. Enabling eight 10TB drives in a JBOD will be seen by the system as one 80TB drive. However, it does not read and write to disks at the same time — JBOD will simply write data sequentially, and read data from the disk on which it is located. As a result, it still works with only one drive, but if one of the drives in the JBOD array fails, the array loses integrity and is destroyed. Without specialized tools, all data from the entire array, as well as the data on that particular drive, will almost certainly be lost. And if one drive has an average failure rate, then the expected failure rate when running eight drives at the same time will be eight times higher than for each drive separately.

In this context, it is clear that StoreMI suffers from the same lack of protection against disk failures. There is no mechanism by which data is protected, and if one disk in the stack fails, data on all disks is lost. If the boot drive is a cheap low quality NAND SSD, or the hard drive is outdated, then data loss on both drives is a real possibility.

Initially, I thought this was a serious problem. For example, if a user wants to stack 10 disks, it could spell disaster. However, the severity of the problem was greatly reduced when I learned about the maximum size of the quick level of 256 GB. This means that most users will probably only be able to connect two drives in one level. AMD’s response to the threat of failure has been to show that users should use backup anyway, and no specific comments have been made on the subject of the dual drive stack’s increased failure rate compared to a single drive. Because StoreMI moves data between an SSD and a hard drive, there may be an additional issue. In this case, there are more data writes to the SSD over time than a normal user’s individual boot drive, causing the NAND to wear out faster. Neither AMD nor AnandTech take this shortcoming seriously, given that modern MLC and TLC SSDs are very good at managing bad data blocks and have built-in redundancy.

And the tests!?

We haven’t been able to directly test StoreMI at this time, focusing on other projects and upcoming events. If we have time, we will provide a full article about StoreMI.

Test parameters

Testing with Specter and Meltdown patches installed

For our testing of the new AMD Ryzen 2000 processors and Intel processors, we used the latest version of Microsoft Windows with the latest updates, as well as BIOS microcode updates, to make sure that the Specter and Meltdown vulnerabilities were patched, as they could be. This means that some of the data used in this review is not comparable to previous reviews, however over time we plan to update our database with the latest fixes.

Test bench

In accordance with our processor testing policy, we take a premium category motherboard suitable for the socket and equip the system with the appropriate amount of memory running at the manufacturer’s maximum supported frequency.

It is noted that some users dispute this approach, mentioning that sometimes the maximum supported frequency is quite low, or faster memory is available at a similar price, or that using the supported frequencies may degrade performance. While these comments make sense, very few consumers end up using memory profiles (XMP or otherwise) as they require interaction with the BIOS, and most users opt out of supported JEDEC speeds — this includes both home users and vendors who may want to reduce the margin by a couple of cents or stay within the limits set by the manufacturer. Where possible, we will expand testing to include faster memory modules, either in this review or at a later date.

Power analysis

One of the key debates around nutrition comes down to how TDP is interpreted, how it is measured, and what exactly it should mean. TDP, or Thermal Design Power, is usually used as a value for the required heat dissipation capacity of the cooler being used, rather than the power consumption. There are a few subtle physical differences between the two, but for simplicity, most users refer to TDP as the nominal power consumption of the processor.

What TDP actually means is hard to determine. For any Intel processor, the nominal TDP is the actual thermal dissipation requirements (or power consumption) when the processor is running at its base frequency. So for a chip like the Core i5-8400 which is rated at 65W, that means the 65W rating is only applicable at 2.8GHz. It’s quite surprising that the official Core i7-8700 turbo TDP is calculated at 3.8 GHz on all cores, which is much higher than the specified base frequency. In truth, if the processor is limited in firmware to 65W, we will see a maximum of 3.2GHz if all cores are loaded. This is an important point for thermally limited scenarios, but it also means that without this limit in the firmware, power consumption is not tied to TDP: Intel does not provide a TDP value above the base frequency, despite the fact that in reality the frequency (in turbo mode) is much above.

AMD calculates TDP a little differently. Previously, this was defined as the peak CPU power consumption, including turbo mode, when all cores are loaded (this is possible with a virus in the system). Now TDP is more of a measure of cooling. AMD defines TDP as the difference between CPU cover temperature and fan intake temperature divided by the minimum required cooler performance. Or to put it another way, the minimum chiller capacity is defined as temperature difference divided by TDP. As a result, we get a sliding scale: if AMD wants to define a cooler with higher thermal performance, it will lower the TDP.

For Ryzen, AMD dictates that this temperature difference is 19.8ºC (61.8ºC on the CPU when the fan intake is 42ºC), which means that for 105W TDP, the thermal performance of the cooler should be able to handle 0.189ºC per watt. For colder thermals, 0.4 ºC/W TDP would be calculated as 50W, or a value of 0.1 would give 198W.

This ultimately makes AMD’s TDP more of a measure of cooling performance than power consumption.

When testing, we are also subject to the whims of the motherboard manufacturer. Ultimately, for some processors, turbo modes are determined by a lookup table. If the system uses X cores, then the processor must run at Y frequency. Not only can motherboard manufacturers change this table with each firmware version, but Intel has also stopped making this data official. So we can’t tell if the motherboard manufacturer meets the Intel specifications or not. In some reviews, we had three different motherboard manufacturers with different lookup tables, but all three claimed to follow Intel’s specifications. Ok, it’s good when everything is simple and clear.

If this is not enough, it is worth noting that we also depend on the whim of chance. Even when two processors are made the same way, the response of the processors to voltage and frequency can actually be very different. The box stamp is just a guaranteed minimum, and actual CPU performance or thermals can vary from that very minimum to something really, really good. Both AMD and Intel go through a process called «binning» whereby each processor from the production line is tested to certain standards — if it exceeds the best standards, it is labeled as the best processor. If it does not meet these standards, it may be labeled as something else. It’s also a known fact that if a manufacturer needs more mid-range processors, they can reduce the percentage of components that meet the high standard, and the same high-quality processors will be labeled as if they were average. So the processor is a lottery.

In our testing, we read power values ​​from internal processor registers designed to estimate power consumption and apply turbo and cooler parameters. This method, strictly speaking, is not the most accurate — and therefore we will use our own multimeters. And yet the first method gives us more information than multimeters. Modern multi-core processors use different voltage plans for different parts of the processor, or even for each core, so software readings give us a good understanding of the power split for different parts of the processor. It is very convenient if the processor makes such information available, but this is not always the case. In most situations, we can get only two main important parameters: the estimated power consumption of the entire chip and the estimated power consumption of all cores (without a memory controller or interconnect).

There is a very noticeable difference between Intel and AMD chips, the difference between the power of the cores and the power of the whole chip. AMD’s interconnect, the Infinity Fabric, combined with other non-core chip components, consumes much more power than Intel chips. And that probably leaves more headroom for Intel to raise frequencies. As mentioned, AMD ties power consumption to its TDP value: our Ryzen 7 2700 showed very high efficiency, although we seem to be seeing average results on the Ryzen 5 2600. In contrast, the Intel Core i7-8700K breaks beyond its TDP very easily, while older Kaby Lake processors are more in line with native TDP values.

Our thanks

Thanks to Sapphire for providing us with several AMD graphics cards. We caught up with Sapphire at Computex 2016 and discussed a platform for future testing with AMD GPUs on their hardware in several upcoming projects. Sapphire donated a pair of RX 460s that we use as CPU test cards. The amount of GPU power consumed can directly impact CPU performance, especially if the CPU has to spend all of its time on the GPU. The RX 460 is a very good card for our tests as it is powerful, has low power consumption and does not require any additional power connectors. The Sapphire Nitro RX 460 2GB continues to follow the Nitro philosophy, and in this case is a powerful card at a low price. His 89The 6 SPs are clocked at 1090/1216MHz and it is paired with 2GB GDDR5 with an effective 7000MHz.

We also have to thank MSI for providing us with the GTX 1080 Gaming X 8GB GPU. Despite the size of AnandTech, providing high performance graphics cards for CPU game testing is quite a challenge. MSI has been very gracious in providing us with a couple of their high-end graphics cards. The MSI GTX 1080 Gaming X 8GB graphics card is an excellent air-cooled product, ranking below the water-cooled Seahawk but above the Aero and Armor versions. The card is big enough with two Torx fans, custom PCB design, Zero-Frozr technology, improved PWM, and a large backplate for easier cooling. The card uses the GP104-400 silicon die on a 16nm TSMC process, contains 2560 CUDA cores, and can be clocked up to 1847MHz in OC mode (or 1607-1733MHz in Silent mode). The memory interface is 8 GB GDDR5X, clocked at 10010 MHz. For a very long time, the GTX 1080 was the #1 card.

Thank you Crucial for providing us with the MX200 SSD. A critical component for our task, as the list of tests grows with new benchmarks and games, and the 1TB MX200 is a strong help. Based on Marvell’s 88S9189 controller and using a Micron chip with 16nm 128Gb MLC, these are 7mm, 2.5″ drives rated for 100K random IOP reads and with 555/500MB/s sequential read and write speeds. . The 1TB models we use here support TCG Opal 2. 0 and IEEE-1667 (eDrive) encryption and have a nominal endurance of 320TB with a three-year warranty.

Thanks to Corsair for providing the AX1200i power supplies. The AX1200i was the first power supply to offer digital monitoring and control via the Corsair Link system, but under the hood it puts out a 1200W rating at 50°C with Platinum certification. This allows for a minimum of 89-92% efficiency at 115V and 90-94% at 230V. The AX1200i is fully modular, with a larger 200mm design, dual ball bearing 140mm fan for high performance use. The AX1200i is designed as a workhorse, with 8 PCIe slots to handle 4 x GPU systems. The AX1200i also features a Zero RPM fan mode that allows you to turn off the fan if the power supply is running under 30% load.

Thanks to G.Skill for the memory. For many years, G.Skill has supported AnandTech in testing processors and motherboards, even if the review is not about memory. We reported on their high performance and high frequency RAM kits, and how every year Computex G. Skill hosts a world liquid nitrogen overclocking tournament right on the show floor.

CPU system tests

Our first set of tests are general system tests. This test suite is designed to emulate most of what people usually do with a computer, such as opening large files or processing small stacks of data. This is somewhat different from our office testing, which uses industry standard testing. Also, some of the tests here are relatively new and unusual.

FCAT processing

One of the more interesting workloads that has been used at our booth in recent quarters is FCAT, a tool we use to measure game latency due to dropped or dropped frames. The FCAT process requires including a color overlay in the game, recording the gameplay and then analyzing the video file with the appropriate software. Such software is usually single-threaded, since the video is in a primitive RAW format, which implies a large file size, and requires the movement of a large amount of data. For our test we take 9A 0 second Rise of the Tomb Raider benchmark running on a GTX 980 Ti at 1440p, which is about 21GB in size, and measuring the time it takes to process with the visual analysis tool.

FCAT is a single-threaded task only and shows the benefits of Intel’s high frequency components and high IPC. From AMD’s side, Ryzen 5 performs better than Ryzen 7, but the results are within the margin of error.

Dolphin Benchmark

Many emulators are bound by single CPU performance, and general reports tend to suggest that Haswell has greatly improved emulator performance. This benchmark runs a Wii program in which a beam traces a complex 3D scene inside a Dolphin Wii emulator. The results of this test are a fairly reliable indicator of the speed of Dolphin CPU emulation, which is a single-core intensive task that uses most aspects of the processor. Results are in minutes, with the Wii itself clocking in at 17.53 minutes (1052 seconds).

Dolphin is also a single-threaded test and has historically favored Intel processors. The new Ryzen-2000 series, with increased IPC and frequency, is moving forward, overtaking Intel’s Skylake.

3D Movement Algorithm Test v2.1

This is the latest version of the 3DPM benchmark we wrote. The goal of 3DPM is to simulate partially optimized scientific algorithms taken directly from my PhD thesis. Version 2.1 differs from 2.0 in that it passes the underlying particle structures by reference rather than by value, and reduces the number of double->float->double conversions performed by the compiler. This gives a 25% speedup compared to version 2.0, which means new data.

In this multi-threaded test, the new 8-core Ryzen 7 2700X goes head and shoulders above Intel’s 8-core Skylake-X versus the 1800X. However, the six-core Coffee Lake i7-8700K is sandwiched between the Ryzen 5 2600X and Ryzen 5 2600.

Agisoft Photoscan 1.

3:

Photoscan remains in our test suite from the previous version of the benchmarks, but we’re now running Windows 10, so features like Speed ​​Shift on the latest processors come into play. The concept of Photoscan is to convert many 2D images into a 3D model — so the more detailed the images, and the more of them, the better the model. The algorithm consists of four stages: several single-threaded and several multi-threaded, and also has some dependence on cache and memory. For some of the more diverse multi-threaded workloads, options such as Speed ​​Shift and XFR can take advantage of CPU waits or idle times, giving significant performance gains on newer microarchitectures.

Photoscan is a variable multi-threaded test, evolving from 1800X to 2700X shows that the extra TDP and Precision Boost 2 can literally shave minutes off the test. Intel’s slower mesh architecture in Skylake-X on the 8-core 7820X compared to the ring architecture of the Coffee Lake 8700K means the two smaller cores on the 8700K allow it to climb higher, and yet it still loses about four minutes to the Ryzen 7 2700X. Intel needs a big 18-core processor, the i9, to win here-7980XE.

Civilization6 AI Test

Our Civilization AI test uses the Steam version of Civilization 6 and runs an in-game AI test to process 25 rounds of late game saves. We run a benchmark on our GTX 1080 at 1080p to ensure rendering isn’t the limiting factor, and the results are given as a geometric mean over 25 rounds to get the average processing time per AI round.

Although the AI ​​test still uses multiple threads, Intel’s high single-core performance makes these processors win.

CPU Rendering Tests

Rendering benchmarks have long been a favorite of reviews and benchmarks, as the code used by rendering packs is usually optimized to squeeze every bit of performance. Sometimes rendering programs is also heavily dependent on memory — when you have many threads carrying tons of data, low latency memory can be the key to everything. Here we take a few of the usual Windows 10 rendering packs, as well as some interesting new benchmarks.

Corona 1.3

Corona is a standalone package designed to support software like 3ds Max and Maya with photorealism via ray tracing. It’s simple — you direct the rays, you get pixels. Okay, a little more complicated, but this benchmark renders a fixed scene six times, and gives results in terms of time and number of rays per second. The official benchmark tables show user results in terms of time, but I find «rays per second» to be the best metric (and in general, results where «more is better» are easier to explain). Corona likes to pile up threads, so the results are highly dependent on the number of threads.

Blender 2.78

The old man of the rendering test world, Blender is still a very popular tool. We were able to run the standard workload on the February 5 build of Blender and measure the time it takes to render the first frame of the scene. Blender is one of the biggest open source tools out there, which means both AMD and Intel are actively working to help improve the codebase, which can either benefit or hurt their own microarchitecture.

This is the multi-threaded test where Intel’s 8-core Skylake processor beats the new AMD Ryzen 7 2700X; Blender’s variable multithreading means that the grid architecture and memory bandwidth perform well here. Although in terms of price-performance ratio, the Ryzen 7 2700X easily outperforms the best performers. The Ryzen 5 2600 easily beats the Core i7-6700K.

LuxMark v3.1

As a synthetic one, LuxMark can seem somewhat unreliable as a renderer given that it’s primarily used for GPU benchmarking. However, it offers both OpenCL and standard C++ mode. In this case, besides the comparison in each core coding and IPC variant, we also see that C++ and OpenCL code show different performance on the same processors.

POV Ray 3.

7.1b4

Another regular benchmark in most kits is POV-Ray. Another ray tracer with a long history. As is often the case, during AMD’s preparations for the launch of Ryzen, the code base has been actively updated as developers make changes to the code and release new updates. Our test version was taken just before the start of such events, but over time we see that the POV-Ray code is adjusted in accordance with the new requirements.

Cinebench R15

The latest version of CineBench has also become one of those programs that have been used everywhere, in particular as a single thread performance indicator. High IPC and high frequency gives performance in ST, while having good scaling and many cores is the result of the MT test.

Intel is still the single-threaded champion in benchmarks like CineBench, but it looks like the Ryzen 7 2700X has taken the lead in the multi-threaded test.

CPU Web Tests

One problem with running web tests is the habit of modern browsers to automatically install updates. This means that in any long period of benchmarking, the «update this outside of compare state» rule will always be violated, especially when browsers start updating, if you give them even half a second to think about it. Regardless, we were able to find a number of commands to create a non-upgradable version of Chrome 56 for our 2017 test suite. And while this means we can’t keep up with the latest version of the browser, it does make the processor comparison scores comparable.

SunSpider 1.0.2:

link

The oldest web test in this part of our review is SunSpider. It’s a very simple JavaScript-based tool, and is ultimately more of a measure of IPC and memory performance than anything else, with most high-end processors showing roughly equal results. The base test runs 10 cycles and returns an average. We perform this basic test 4 times.

Mozilla Kraken 1.1:link

Kraken is another Javascript-based test using the same test suite as SunSpider but focused on more rigorous real-world use cases and libraries such as audio processing and image filters. Again, the main test body goes through 10 cycles and we run this base test four times.

Google Octane 2.0:link

Major browser vendors such as Google and Mozilla know that peak JS performance is often a critical point when compared to other OS vendors. Just like SunSpider is a very early test of JS and Kraken is a bit newer, Octane aims to be more relevant to real workloads, especially on devices with limited capabilities like smartphones and tablets.

WebXPRT 2015:link

While the previous three tests perform calculations in the background and then show the scores scored, WebXPRT is designed to better interpret the visual workloads experienced by professional users. For example, browser-based applications, graphics, image editing, data sorting and analysis, scientific analysis, and financial tools.

CPU Encoding Tests

One of the interesting elements of modern processors is encoding performance. This includes encryption/decryption as well as video transcoding from one video format to another. In the case of encryption/decryption, performance is still relevant for on-the-fly encryption of sensitive data, a process by which modern devices typically secure software. The use of video transcoding to adjust the quality, file size and resolution of a video file has been booming lately. This is necessary to fit the video to the consumer’s device, or for game streamers who wish to upload the transcoded video stream from the camera in real time. As we move into live 3D video, this task will become even more of a priority, and it turns out that the performance of certain algorithms is determined by the content input / output function.

7-Zip 9.2:link

One of the freeware archivers that allows you to evaluate the performance of processors is 7-Zip. It operates under an open source license and is a fast and easy to use tool for advanced users. We run the test mode via the command line, run four loops and get the result.

WinRAR 5.40: link

For the 2017 test package, we will use the latest version of WinRAR in our test of archivers. WinRAR is more user-friendly than 7-Zip in some aspects, so we decided to include it in the package. Instead of using benchmark mode as in the case of 7-Zip, here we take a set of files representing a total stack (33 video files in 1.37 GB, 2834 smaller files in 370 150 MB folders) of compressible and incompressible formats. The results shown are the time taken to encode the file. Due to DRAM caching, we run the test 10 times and get an average of the last five runs when the benchmark is in a steady state.

WinRAR requires a good memory base, so we see quad-channel processors top the list. The high IPC Core i7-8700K also works great.

AES Encoding

Algorithms using AES encryption have become widespread around the world as a ubiquitous tool for encryption. However, this is another test-for-CPU, and modern CPUs have the AES instruction set extension to increase encryption performance. We often look at both frequency and core scaling with this benchmark. We are using the latest version of TrueCrypt and running it in benchmark mode with 1 GB of data in DRAM. The results shown are the average GB/s for both encryption and decryption.

HandBrake v1.0.2 h364 and HEVC: link

As mentioned above, video transcoding (both encoding and decoding) is a hot topic in terms of performance as the pace of new content creation picks up. The first assumption is a change in the video formatting standard, which can occur both with loss and without loss of quality. It is also possible to reduce the quality of the video for the sake of the file size. Along with Google’s favorite codec, VP9, there are two other commonly used ones: h364, an old codec found almost everywhere and optimized for 1080p video, and HEVC (or h365), which aims to provide the same quality as h364 but at a lower file size (or more high quality for the same size). HEVC is important because it is capable of streaming video in 4k quality, which means fewer bits for the same quality content.

Handbrake is the transcoding tool of choice, so our test mode covers three areas.

Low quality / h364 resolution: here we transcode a 2 hour 640×266 h364 video and change the encoding from Main profile to High profile using the very-fast preset.

High Quality / h364 Resolution: Similar test, but this time we take a ten minute 4K double file (3840×4320) rendered at 60Hz and re-encoded from Main to High using the very-fast preset.

HEVC test: using the same video in HQ, we change the resolution and codec of the original video from 4K60 in h364 to 4K60 HEVC.

For large-sized video transcoding, the previous generation HandBrake has a problem with the new Ryzen-2000 series processors, but even with this, the Core i7-8700K is ahead by a decent margin. The Core i5-8400 also performs well, but is inferior to the Ryzen processors.

CPU Office Tests

The office programs we use for benchmarking are not specific programs, but industry standard tests that carry weight in a professional environment. The purpose of these tests is to use a set of programs and methods that a typical office user might encounter, such as video conferencing, document editing, architectural modeling, and so on and so forth.

Chromium Compile (v56)

Our new compilation test uses Windows 10 Pro, VS Community 2015.3 with Win10 SDK to compile a nightly build of Chromium. We prepared a build test at the end of March 2017 and in our test we are running a new full compilation. Compilation is a typical example of a workload with variable multithreading — some compilation and linking processes are linear, while other parts are multithreaded.

Compilation testing shows that the new Ryzen-2000-series components are a big step up from the first generation, probably due to the reduced cache latency and the new precision boost. The performance per dollar of the 8700K and 2700X seems to be about the same.

PCMark8: link

Even though PCMark was first introduced in 2008/2009, Futuremark supports PCMark8 and it remains relevant in 2017. At the scale of complex tasks, PCMark focuses more on the low-percentage range of professional workloads, making it a good indicator of what people consider «office work». We run the benchmark from the command line in «traditional» mode, i.e. C++ without OpenCL, to remove the graphics card from the equation and focus purely on the processor. PCMark8 offers Home, Work, and Creative workloads, with some software tests being generic and others unique to each test suite.

PCMark 10

GeekBench5

When it comes to multi-threaded benchmarks, the best parts from Intel and AMD are head-to-head, but you can see that the previous generation quad-core processors are lagging behind.

CPU Legacy Tests

Our legacy tests are benchmarks that were once ahead of their time. Some of them are synthetic industry standards and we have data coming in for 10 years now. All of the data here has been re-run in Windows 10 and we plan to go back over multiple generations of components to see how performance has evolved.

3D Particle Movement v1

3DPM is a test we wrote that applies the basic 3D motion algorithms used in Brownian motion simulations and tests them for speed. High performance floating point, MHz and IPC will show itself in the single threaded version, while the multithreaded version works with threads and likes many cores. This is the original version, written in the style of a typical «non-computer» student coding an algorithm for a theoretical problem. It comes without any non-obvious optimizations that are not yet performed by the compiler, such as false sharing.

CineBench 11.5 and 10

Cinebench is a well-known performance measurement tool related to the MAXON Cinema 4D animation software. Cinebench has been optimized for a decade and only focuses on CPU power. This means that if there is a mismatch in bandwidth performance, Cinebench will most likely show that mismatch. It’s possible that other software doesn’t use all available CPU-intensive tools, so the test’s relevance to the real world may be purely academic, but given our large database for Cinebench, it’s hard to ignore a small five-minute test. In this test, we are running the modern version 15 as well as the old 11.5 and 10 because of our saved data.

x264 HD 3.

0

Similarly, the x264 HD 3.0 package we are using here is also stored for historical regression data. The latest version is 5.0.1 and it encodes 1080p video to a high quality x264 file. Version 3.0 performs the same test in a 720p file, and in most cases software performance hits the limit for high-end processors, but still performs well for mid-range and low end. Also, this version only runs for a few minutes, while the latest version can take over 90 minutes.

Gaming Performance: Civilization 6

So, the first game in our CPU game tests is Civilization 6. Initially launched by Sid Meier and his team, the Civ turn-based strategy series has become a cult classic. Many apologies were made for the sleepless nights of the players who couldn’t get Gandhi to go to war due to integer overflow. To tell the truth, I never played the first version, but I did play every part from the second to the sixth, including the fourth, voiced by the late Leonard Nimoy. This is a game that is easy to play but difficult to play well.

Civilization’s benchmarking has always been something of an oxymoron — for a turn-based strategy game, framerate isn’t particularly important, and in the right mood, just 5fps is enough for a good game. However, with Civilization 6, Firaxis has gone hardcore and raised the visual bar in an attempt to get you into the game. As a result, Civilization may require a new graphics card and processor from the player, especially when playing in high detail under DirectX 12.

Perhaps a more desirable result will be seen during the late game, when, in older versions, Civilization could take 20 minutes to make the AI ​​players’ turn and give control to a human. The new version of Civilization has an integrated «AI Benchmark», although it is currently not in our test portfolio yet for technical reasons we are trying to resolve. Instead, we run a graphics test that provides an example of the game’s average settings in the options.

For 1920×1080 and 4K resolutions we run the same settings. Civilization 6 has sliders for MSAA, Performance Usage and Memory Usage. The last two refer to detail and texture size, respectively, and are rated from 0 (lowest) to 5 (extreme). We run our Civ6 benchmark at position four for performance (ultra) and 0 in memory and MSAA at 2x.

For reviews where we use 8K and 16K benchmarks (Civ6 allows us to compare extreme resolutions on any monitor) on our GTX 1080, we run 8K benchmarks the same as 4K, but 16K benchmarks are set to the lowest performance setting.

MSI GTX 1080 Gaming 8G Performance

1080p

4K

8K

16K

Shadow of Mordor

The next name in our performance battle is open-world action RPG Middle Earth: Shadow of Mordor (SoM for short). The game was created by Monolith on the LithTech Jupiter EX engine with many additional add-ons. SoM goes into detail and complexity. The game’s main story was written by the same writer who wrote Red Dead Redemption, and SoM won Zero Punctuation’s Game of the Year in 2014.

The games of 2014 are quite old for modern testing, but SoM has stable code and many fans, and can still stress the player’s computer. At the time, SoM was unique in offering dynamic screen resolution, allowing users to use high resolution settings, which are then scaled down to the monitor’s capabilities. This form of natural resampling was designed to give the user a better idea of ​​what the developers wanted if you have powerful enough graphics hardware but no -4K monitor.

The game has a built-in benchmark, and we run it using a script that configures the graphics, starts the benchmark, and parses the results that the test dumps to disk. Graphics settings include standard options such as Graphical Quality, Lighting, Mesh, Motion Blur, Shadow Quality, Textures, Vegetation Range, Depth of Field, Transparency and Tessellation. There are also standard presets.

We’re running the benchmark at 1080p and native 4K using our 4K monitors at Ultra settings. The results are averaged over four runs and we output the average FPS, 99 percentile and time under analysis.

MSI GTX 1080 GAMING 8G Performance

1080p

4K

9000

Rise of the TOMB RAIDER 9002
One of the newest games in our gaming benchmark suite is Rise of the Tomb Raider (RoTR), developed by Crystal Dynamics, the sequel to the popular Tomb Raider, which was loved for its built-in automatic benchmark mode. But don’t be fooled: the benchmark mode in RoTR is very different from the past.

Visually, the previous Tomb Raider raised the bar for realism almost to the level of TressFX, and the new RoTR goes up another notch in terms of graphic fidelity. This leads to an interesting set of hardware requirements: some sections of the game tend to be GPU-limited, while others with lots of physics over long distances can be CPU-limited, depending on how the driver allocates the DirectX 12 workload.

Where the old game had one reference scene, the new game has three different scenes with different requirements: geothermal valley (1-Valley), Prophet’s Tomb (2-Prophet), and mountain spine (3rd mountain) — and we test all three. These are the three scenes that should be taken from the game, but it has been noted that the scenes like 2-Prophet shown in the benchmark may be the most CPU-limited elements of this entire level, and the scene shown represents only a small portion of that level. Because of this, we report the results for each scene on each video card separately.

The graphics settings for RoTR are similar to those of other games of this type, offering some presets or allowing the user to adjust texture quality, anisotropic filter levels, shadow quality, soft shadows, occlusion, depth of field, tessellation, reflections, foliage, bloom and features like PureHair, which the TressFX library allows you to use.

We are still testing the game at 1920×1080 and 4K using our own 4K displays. At 1080p we run the High preset, and at 4K we use the Medium preset, which still packs a significant framerate hit.

It’s worth noting that the RoTR benchmark is slightly different from our other benchmarks in that the game saves its graphics settings in the registry rather than a standard INI file, and unlike the previous game, the built-in TR benchmark cannot be invoked from the command line. Nevertheless, despite the difficulties, we prepared a script to automatically run the benchmark four times and parse the results. From the received data, we derive the average FPS, 99th percentile and time under analysis.

MSI GTX 1080 GAMING 8G Performance

1080p

4K

9000 9000

ROCKET LEAGUE

Fun simple play-and-collect games are great fun. For this reason, I’m a big fan of the Katamari franchise — you just press «start» on your controller and you roll forward, picking up items to grow. Extremely simple. Until we get a PC version of Katamari that I can test, we’ll be focusing on Rocket League.

Rocket League uses pick-up-and-play elements, allowing users to play with other people (or bots) to play football without rules on cars. The game was created on the Unreal Engine 3 engine, which, although outdated at the moment, allows users to both run the game on low-performance systems and squeeze all resources out of more powerful systems. Since its release in 2015, the game has sold over 5 million copies and seems to be a star on local networks and game shows. Users who practice playing become very serious, fighting in teams and leagues with very little customization and everyone is on the same level. Rocket League is confidently becoming one of the big names in the world of esports, and what is especially nice is when you can watch the competition directly from the game interface.

Based on these factors, and because this game is a pleasure to launch and play, we decided to find the best way to test it. Unfortunately, most of the automatic tests for games will not work here. In addition, thanks to the Unreal 3 engine, Rocket League does not have a benchmark mode. In this case, we have to develop a tight run and record the frame rate.

As already mentioned, Rocket League doesn’t have a benchmark mode, so we have to perform a series of automatic actions similar to a racing game with a fixed number of laps. We take the following approach: using Fraps to record the time it took to render each frame (and the total frame rate), we will use an automation tool to run a game with 4v4 bots, while the system must perform a series of actions during the match, for example, switching angles cameras and movement.

It just so happens that the described method is quite accurate in representing data from a real match with bots, including movement, collisions, power-ups, or even getting unexpected help, as strange as it sounds for an automated set of commands. To maintain consistency, the commands we use are not random, but also fixed in time. We also run tests on the same card (Aquadome, which is known to be a GPU heavy card due to water/transparency) and with constant vehicle settings. We start recording right after the start of the match and record for 4 minutes of game time (5 laps DIRT: Rally benchmark I think), and determine the average frame rate, 99th percentile and time under.

MSI GTX 1080 GAMING 8G Performance

1080p

4K

9000
The long-awaited iteration of the Grand Theft Auto franchise hit the shelves on April 14, 2015, with both AMD and NVIDIA making efforts to optimize the game. GTA has no graphic presets, yet the game opens up new possibilities for users and pushes the limits of modern graphics, pushing even the most powerful computers to the limit with the help of Rockstar’s Advanced Game Engine under DirectX 11. Regardless of whether the user is flying high in the mountains, where you need to draw the world from long distances, or deal with the sorted garbage in the city, when it is bent to the maximum, the game creates amazing visuals, plus hard work for both the processor and the graphics card.

For testing, we wrote several scripts for the benchmark built into the game. The internal benchmark will include five scenarios: four short panoramic scenes with variable lighting and weather effects, plus a fifth — a sequence of actions lasting about 90 seconds. We decided to use only the last scene, which includes flying in a jet, then driving a car through the city through several intersections, and at the end, a collision with a fuel truck, which explodes, as do the cars around it. It’s a great mix of long range rendering followed by action with close range rendering. And fortunately, the game produces all the necessary test results.

GTA has no graphics presets, but the user can manually adjust settings such as population density and draw distance using sliders. Some parameters like texture/shadow/shaders/water quality are switched from low quality to very high quality. Other customization options include MSAA, soft shadows, post-effects, shadow resolution, and advanced remote rendering settings. There’s a handy option at the top of the screen that shows how much VRAM the game will consume at these settings, with obvious consequences if the user requests more VRAM than what’s on the card (although there’s no obvious prompt if you have a weak GPU with lots of VRAM, e.g. R7 240 4 GB).

We end up running tests at 1920×1080 using Very High in the settings, and also at 4K using High in most of them. The result will be the averages of the four runs, average frame rate, 99th percentile, and time-under analysis.

MSI GTX 1080 GAMING 8G Performance

1080P

4K

9000

Bena

Conclusions

Throughout the adventures of AMD and Zen, the pursuit of x86 HPC has two goals: to be competitive, and to be the best. Undoubtedly, the first generation of Ryzen succeeded in being competitive, with analysts and retail outlets showcasing Ryzen processors in the top selling lists and providing analytical benchmarks where AMD’s best chips compete against Intel’s high-end desktop components. The goal for the next few years is to capture these readily available microarchitecture advantages and later capture and exploit the benefits of the new process technology. The first stage is the second generation of Ryzen, known as the Ryzen-2000 series.

In this review, we showed that AMD achieved its goal of an additional 3% increase in net performance, with our numbers showing +3.1%. Combining this with the frequency boost gained from the GlobalFoundries 12nm manufacturing process and turbo features like Precision Turbo Boost that makes most temperature limits intuitive, AMD has achieved a 10 percent performance gain across generations overall. This doesn’t sound so fair, as at first glance these are just small improvements in the production process, as well as some reasonable reduction in cache latency. If it were Intel, we would be screaming for a big and dramatic breakthrough, but AMD is still sitting on Zen design and the focus is on the next full microarchitectural update in Zen 2. Therefore, most users and journalists are giving a thumbs up in a sign of confidence in AMD now , hoping to see a bigger jump next time.

4K gaming analysis

However, everyone wants to know about the results of the Ryzen 2000 series benchmarks. We start with games, and first of all with our performance/price charts from high-end 4K gaming. Our results take the R7 1800X as our baseline «100%» and we show the geometric average across all of our 4K gaming tests.

As you can see from the results, the new Ryzen 2000-series processors are up to 1-3% better than the Ryzen 7 1800X, and even the Ryzen 5 2600. Intel processors were 0-4% above the 1800X, with both Coffee Lake processors at that +4% peak. In GPU-limited benchmarks, there is some benefit to single-core performance, but all modern processors at a reasonable frequency perform quite well.

For the 99th percentile charts, AMD’s new processors are either leveling up or have already outperformed the Ryzen 1000-series. For our Intel benchmarks, the previous generation 6700K/7700K are 3% behind the 1800X and the rest are 4% above the 1800X. Intel’s favorite, Coffee Lake, takes first place, but all processors (with the exception of the Bristol Ridge, A12-9800) have improved their results.

Gaming analysis 1080p

A key aspect for many users is playing at a lower resolution: 1920×1080 still dominates gaming settings, no matter how much more pixels we as enthusiasts want to see. Some readers have reached out to us claiming that they are still buying the best graphics card possible, but running at 200+ FPS at low resolutions, just for speed. Low resolution benchmarks are a poor indication of future processor performance, but what we see today is what matters to us.

At this resolution, the Ryzen 7 2700X is up about +7% from the previous generation 1800X, with the new 65W components easily matching the 1800X. Users looking at the budget Ryzen 5 1600, AMD’s bestseller last year, can now look at the 2600 (+7%) or 2600X for (+10% vs. 1600)

Yet Intel wins here too. With higher IPC speeds and higher clock speeds, Coffee Lake processors are 8-10% faster than Ryzen 7 2700X, ranging from +3% to +25% depending on the model. That being said, our results show that the Ryzen 7 2700X has risen above Intel’s previous generation processors, allowing the 2700X to win by a narrow margin over Kaby Lake and a 5%+ margin over Skylake.

The 99th percentile graph at 1080p looks like a stretched version of the average frame rate graph, and it generally is: processors that perform best at 1080p perform even better at percentiles. Of all the metrics where AMD needs to be competitive, this is where the new performance boost works best, but it looks like there’s a lot of work to be done. To be honest, the results will look better if and when AMD can match Intel in terms of frequency. AMD also has a core memory DRAM response speed deficit, which we cited as a factor in previous testing at 99 percent frame rate. It will be interesting if AMD can provide a higher supported memory frequency faster than Intel, because it really matters here.

Workstation analysis

In the mainstream competition, AMD delivered eight cores and sixteen threads with the Ryzen-1000 against Intel’s Kaby Lake with four cores and eight threads. This time around, Intel is offering six cores in Coffee Lake, and now Intel’s 12 threads are up against AMD’s 16. AMD has also raised the bar for turbo frequency, but Intel is pushing the power budget far beyond the limits specified for the processor. So, in the middle range, where there were four streams against twelve, now six against twelve, however, Intel again produces higher frequencies.

In our single thread tests, the new Ryzen-2000 series now match the performance of Intel Skylake processors. In the last round of comparisons, Intel still had some advantage, but now they are equal. To AMD’s chagrin, Intel is still two generations ahead, thanks to Kaby Lake and Coffee Lake, which deliver excellent single-core performance at the expense of additional clocks. This means that the Core i5-8400 really outperforms the best AMDs in single-threading, and the Core i7-8770K is king of the hill.

A number of additional variables come into play in multi-threaded workloads. This is the nature of the thread in each core, which parts of the core are dynamic or statically separated, memory and cache management. Here AMD has been aggressive on latency and low-level cache size, but AMD’s non-inclusive L3 cache competes here with Intel’s L3 write-back cache, which is more useful, as well as lower latency main memory.

What we’re seeing here for AMD is only the Ryzen 7 2700X outshines the old Ryzen 7 1800X, with the 2700 lagging behind. In this price range, the AMD chip’s higher thread count gives it a healthy lead over the Coffee Lake i7-8700K, and it outshines the octa-core Skylake-X Core i7-7820X in most benchmarks. It is important to note that the previous generations of the Intel i7-6700K and i7-7700K processors are very far behind the competition, and even the latest 2nd generation AMD Ryzen processor, the Ryzen 5 2600.

If someone had said a few years ago that AMD would develop a second-generation Ryzen processor in 2018 that would outshine Intel’s Skylake and Kaby Lake processors, I would have laughed. But here we are talking about the success of AMD. If a user wants a chip for a multi-threaded workload, the $199 Ryzen 5 2600 is the best budget processor on the market today.
In this mid-range price competition, the Ryzen 5 2600 easily outperforms the Core i5-8400 as well. This is not a fight, but a beating.

General thoughts

The manufacturer loses somewhere, wins somewhere, but customers always win

When there is competition, each product must make its best leap forward. Carrying this burden is not easy, with existing relationships and high budgets. But every breakthrough forces the competitor to react, and that’s a win for customers. In recent years, Intel has mostly been criticized for being stagnant—minor updates to the current process while waiting for a new one; with such a long stagnation, everything that is competitive will look attractive in the market. Those inside the industry might say that while Intel releases iterations of a similar 14nm design every year, anticipating a 10nm release, AMD has taken its best step forward with Zen and Ryzen at 14nm, now following with Ryzen 2 at GF 12 nm.

This time rounding down to simple recommendations will actually be very simple.

Any user who needs high single-threaded performance or high performance 1080p gaming with a mid-range GPU can choose Intel’s Core i5-8400 as a great solution.

For hardcore enthusiasts running high-end graphics at 4K or just like their general computing, the Ryzen 2000 series looks like the best bet. At every price point, AMD can meet the needs of a 4K gamer and win in pure bandwidth.
AMD also ships a decent cooler with each processor, something Intel has neglected in recent years, which makes the product even more attractive.

Thank you for staying with us. Do you like our articles? Want to see more interesting content? Support us by placing an order or recommending to your friends, 30% discount for Habr users on a unique analog of entry-level servers that we created for you: The whole truth about VPS (KVM) E5-2650 v4 (6 Cores) 10GB DDR4 240GB SSD 1Gbps from $20 or how to share a server? (available with RAID1 and RAID10, up to 24 cores and up to 40GB DDR4).

Dell R730xd 2 times cheaper? Only here 2 x Intel Dodeca-Core Xeon E5-2650v4 128GB DDR4 6x480GB SSD 1Gbps 100 TV from $249 in the Netherlands and USA! Read about How to Build Infrastructure Corp. class with the use of Dell R730xd E5-2650 v4 servers worth 9,000 euros for a penny?

AMD Ryzen 7 1800X vs. Intel Core i7-7700K

Works With Us 3rd-Gen Ryzen will be on sale very soon, we are gearing up for this launch by revisiting a few hot CPU battles from the past couple of years which may bring additional context to current owners these processors.

Today’s shootout between Ryzen 7 1800X and Core i7-7700K. But before we get into that, some caveats. We’re testing the 1800X rather than the cheaper 1700 in part because we wanted to make it a one-on-one test and time was of the essence so we could only pick one Ryzen 7 processor. we could only choose one.

Upon release, the 1800X had an MSRP of $500, making it significantly more expensive than the $340 Core i7-7700K. This made the Ryzen 7 1700 a more direct competitor as it was priced at just $330. However, AMD soon made some price cuts. , and the 1800X dropped to roughly match a Core i7 processor.

To be precise, the 1800X dropped to $320 the same year it was released, and has since dropped to $220. We also overclock Core i7 and Ryzen 7 processors as part of our test. The R7 1700 is typically clocked at 4GHz and , therefore, should deliver the same overclocked performance you’ll see today at 1800X.

In terms of memory, we internally argued about using expensive Samsung B-Dies with low latency and ended up using G.Skill’s. FlareX DDR4-3200 CL14 memory for several reasons: these are not budget processors, so additional memory costs are out of the question, and secondly, we test the performance of gaming processors with RTX 2080 Ti to remove the processor bottleneck, so why us to do that, and then limit performance by using slower system memory.

This was our reasoning and shouldn’t bother you anyway if both processors were tested with the same memory. We tested 9 games in two resolutions: stock and overclocked, as well as several application tests. Let’s dive into this.

Application Benchmarks

Starting from this session we have Cinebench R20 which has always been good at showing two tales of Ryzen. Here we have multi-threaded performance, and despite the rather large clock speed deficit, we see Ryzen stepping over the Core i7 processor without any problems. Initially, a processor with core 8 is 55% faster, although of course it has the advantage of doubling the number of cores. However, when fully loaded, it becomes clear that the 7700K is no match for the 1800X.

Here’s the other side of the story: single-core performance. The Core i7-7700K is 22% faster out of the box and when overclocked. This gives the Intel processor a significant advantage in lightly threaded workloads.

WinRAR is more about bandwidth and memory latency than cores, and as a result the 7700K is ~8% faster in this particular workload. Not a big difference, but after watching the Cinebench R20 multi-core benchmark, it’s not the result you might expect.

Continuing, we have a benchmark that is especially important for creators and video editors. I use Premiere Pro almost every day, so I’m directly involved. Note that the lower the better as we are looking at the time it takes to encode a single Hardware Unboxed video in 4K H.264. The R7 1800X is 43% faster in this test, which takes 508 seconds out of the box. So for content creators, the CPU 8 is the obvious choice.

Similarly, the 1800X was welcomed with open arms by 3D modeling professionals. Here we can see that the 1800X is 57% faster than the 7700K out of the box when using the latest version of V-Ray, and 50% faster when both processors are overclocked.

We see a similar story with Corona: the 1800X was 57% faster out of the box, completing the task in just 131 seconds.

The last application test we’re looking at is a mixer and here the 1800X was 62% faster and 60% faster when overclocked — a huge benefit for Ryzen.

While running our Blender workload, we also measured overall system consumption, and here the Ryzen 7 1800X was pushing power consumption 26% higher, a positive result of about a 60% increase in performance. Overclocking really takes Ryzen’s efficiency out of the window, and to be honest, the 32% increase in system power isn’t worth the marginal increase in performance.

Game tests

Time for some games. First we have Assassin’s Creed: Odyssey where the 7700K was 5% faster on average and 8% faster for a low score of 1%. These differences remained virtually unchanged after overclocking. As you might expect, going to 1440p reduces the headroom as we became a bit more GPU limited, and once overclocked both CPUs were able to max out the RTX 2080 Ti. It’s worth noting that we’re not using the highest quality settings in Assassin’s Creed: Odyssey, so it’s possible to make the game a lot more GPU bound without increasing the resolution.

The Battlefield V results are interesting, here the 7700K is maxed out, and while it’s much better than the 7600K, we can see the low performance is 1% lower than the 1800X, which has significantly more breather. So in this case, the average frame rate can be misleading as 1800X delivers a smoother experience.

Even at 1440p, the 7700K still gets the most out of it and therefore delivers a significantly worse experience than the 1800X. The game is still very playable on the 7700K, but given the choice of these two CPUs for BFV, we’re sure most gamers will pick the 1800X.

Here we see that the 7700K is well versed in Shadow of the Tomb Raider, and this is entirely dependent on Hyper-Threading support. Previously when testing the 7600K, it really struggled in this title and was noticeably slower than the R5 1600. In this case, both the 7700K and 1800X delivered smooth, playable performance, but overall the Core i7 processor was faster and at a reasonable margin percentage of 11.

Transition 1440p cuts that margin, but the 7700K is still faster by a 6% margin when comparing average frame rates.

2 is another section where the 7600K struggled, but with the help of Hyper-Threading the 7700K parses great and is much faster than the 1800X at a low 1% performance comparison. The

1800X returns with a more limited 1440p resolution for GPUs, and yet the 7700K posted a comfortable 11% advantage when comparing low 1% results.

Once again we see that Far Cry is the title that Ryzen processors aim to fight. In this case, the 7700K was 100% faster than 24 out of the box and 31% faster after overclocking. The Ryzen 7 was able to keep frame rates above 60, but the 7700K was much better overall.

Even at 1440p, we can see that for high refresh rate games in Far Cry, the New Dawn 7700K is simply a much better processor for the job. Overclocking was 35% faster this time around, which is a huge difference for a processor to do it at 1440p, even with an RTX 2080 Ti.

The Ryzen 7 processor also loses in World War Z, but this time the losses are less significant as both processors allow the RTX 2080 Ti to render at 130 frames per second.

Naturally the margins are reduced at 1440p and here the 7700K was up to 8% faster. Needless to say, both processors provided excellent gaming experiences.

Rage 2 does not require a processor header, and in this header both processors provide the same average frame rate. The higher clocked 7700K did score a lower 1% result, offering around 12% more performance.

At 1440p, that headroom is kept to a minimum, and it’s fair to say that both processors provide the same gaming experience.

1st gen Ryzen is not up to the task in Hitman 2 and you will see a significant performance improvement, especially for low 1s, when moving to a part of 2nd gen Ryzen like the 2700X. The 1800X does deliver smooth playback, but the 7700K manages the same with around 18 to 20% more frames.

Even at 1440p, we’re still CPU-bound, and the 7700K delivers a slightly better gaming experience as a result.

The last game on our list is Total War: Three Kingdoms, where both processors achieve the same average frame rate, but a higher core count at 1800X delivers a noticeably better 1% performance.

This is also seen at 1440p when comparing performance out of the box, but we can see that the 4. 8 GHz multi-core overclock allows the 7700K to run faster.

What did we miss?

For modern gaming, the Core i7-7700K and Ryzen 7 1800X are generally equal. There are lightweight games where the 7700K has a significant frame rate advantage, but in all of those games, the 1800X still delivers smooth, fluid performance. Meanwhile, in more demanding games, the 7700K is starting to find its limits. When the CPU was completely disabled, the low 1% performance was particularly affected.

Frankly speaking, this situation is not so different from what we found in our first day survey two years ago, it is now a little more pronounced. Here’s a direct quote from our now biennial Ryzen 7:9 review0003

One thing we noticed is that all the games we looked at were smooth on Ryzen processors. GTA 5, for example, plays very well on the Core-7 i7700, but a slight stutter can be noticed from time to time, while the 1800X runs smoothly as silk, with no stutter from what we observed.

We found a similar situation when testing Battlefield 1. The performance of the Ryzen processors was flat while the quad-core 7700K experienced a slight hiccup from time to time. This was rare, but we didn’t notice it when using the 1800X and 1700X. But, oddly enough, that doesn’t change the fact that gamers using a high refresh rate monitor may be better served with the higher refresh rate Core-7 i6700 or 7700K.

While the gaming results may not be as strong as we hoped, they are quite competitive and this should be especially true for the Ryzen 5 and 3 series. It’s also worth noting that we’re testing extreme gaming performance here with the Titan XP at 1080p. Ryzen looks more competitive at 1440p, and certainly so when paired with a GTX 1070 or Fury X.

For the most part, we’d say this conclusion remains accurate today, though we wouldn’t recommend the 7700K over any Ryzen part. 7 in 2019even if they were sold at the same price.

Also, the 7700K left us with a bad taste in our mouths after launching early $340 2017 for an LGA1151 socket, it was actually dead that same year.