I7 6700 vs i5 6600k: Intel Core i5-6600K vs Intel Core i7-6700: What is the difference?

Subscribe to our channel in Yandex.Zen or telegram channel @overclockers_news — these are convenient ways to follow new materials on the site. With pictures, extended descriptions and no ads.

Compare Intel Core i7 6700K vs. Intel Core i5 6600K

Comparison of Intel Core i5-7600K vs. Intel Core i7-7700K processors by all known specs in categories: General, Performance, Memory, Graphics, Graphical Interfaces, Picture Quality in Graphics, Graphics API Support, Compatibility, Peripherals, Security and Reliability, Technology, Virtualization. Analysis of processor performance by benchmarks: PassMark — Single thread mark, PassMark — CPU mark, Geekbench 4 — Single Core, Geekbench 4 — Multi-Core, 3DMark Fire Strike — Physics Score, CompuBench 1.5 Desktop — Face Detection (mPixels/s), CompuBench 1.5 Desktop — Ocean Surface Simulation (Frames/s), CompuBench 1.5 Desktop — T-Rex (Frames/s), CompuBench 1.5 Desktop — Video Composition (Frames/s), CompuBench 1. 5 Desktop — Bitcoin Mining (mHash/s), GFXBench 4.0 — Car Chase Offscreen (Frames), GFXBench 4.0 — Manhattan (Frames), GFXBench 4.0 — T-Rex (Frames), GFXBench 4.0 — Car Chase Offscreen (Fps), GFXBench 4.0 — Manhattan (Fps), GFXBench 4.0 — T- Rex (Fps).

Intel Core i5-7600k

VS

Intel Core i7-7700K

Benchmark

General result

Based on 13 tests: Intel Core i7 63 Core i5 6600k

100

Geekbench 3 (Multi-Core)

Intel Core i7 6700K faster at 16.03%

Intel Core i7 6700k

17750 9000

Intel Core i5 6600000 9000

003

Intel Core i7 6700k

33.21 Megheshi/C

Intel Core i5 6600k

32.43 Megheshi/C

COMPUBENCH 1.5 (T-RREX) 9000 I5 Core 9000 9000 9000 9000%.

1.63 FPS

Intel Core i5 6600k

1.9 FPS

Advantages

Reasons to choose the Intel Core i5-7600K

• Processor Release, The difference in release 1 YEAR (S)
• Approximately 56% increase in maximum core temperature: 100°C vs 64°C
• Approximately 2% increase in PassMark — Single thread mark performance: 2574 vs 2530
• Approximately 2% increase in Geekbench 4 — Single Core performance more: 1169 vs 1148
• About 30% better performance in GFXBench 4. 0 benchmark — Manhattan (Frames): 4291 vs 3294
• 2.3 times greater performance in GFXBench 4.0 benchmark — T-Rex (Frames): 75 PassMark — Single thread mark 2574 vs 2530 Geekbench 4 — Single Core 1169 vs 1148 GFXBench 4.0 — Manhattan (Frames) 4291 vs 3294 GFXBench 4.0 — T-Rex (Frames) 7595 vs 3360 GFXBench 4.0 — Manhattan (Fps) 4291 vs 3294 GFXBench 4.0 — T-Rex (Fps) 7595 vs 3360

  Reasons to choose Intel Core i7-6700K

  • 4 more threads: 8 vs 4
  • PassMark performance — CPU mark about 30% higher: 8956 vs 6877 Multi
  • Geek-bench4 performance About 16% more Core: 4458 vs 3827
  • About 25% more performance in 3DMark Fire Strike — Physics Score: 5050 vs 4042
  • About 46% more performance in CompuBench 1. 5 Desktop — Face Detection (mPixels/s) more: 6.217 vs 4.244
  • CompuBench 1.5 Desktop — Ocean Surface Simulation (Frames/s) approximately 17% faster: 101.85 vs 87.113
  • CompuBench 1.5 Desktop — T-Rex (Frames/s) approximately 44% faster: 0.744 vs 0.516
  • Performance in CompuBench 1.5 Desktop — Video Composition (Frames/s) about 12% more: 2.978 vs 2.661
  • Performance in CompuBench 1.5 Desktop — Bitcoin Mining (mHash/s) about 3% more: 7.352 vs 7.129
  • About 22% more performance in GFXBench 4.0 benchmark — Car Chase Offscreen (Frames): 1760 vs 1441
  • About 22% more performance in GFXBench 4.0 benchmark — Car Chase Offscreen (Fps): 1760 vs 1441
  9 PASSMARK — CPU MARK — CPU MARK0180

  Characteristics
  Flows	8 VS 4
  Benchmark
  Geekbench 4 — Multi-Core	4458 vs 3827
  3DMark Fire Strike — Physics Score	5050 vs 4042
  CompuBench 1. 5 Desktop — Face Detection (mPixels/s)	6.217 vs 4.244
  CompuBench 1.5 Desktop — Ocean Surface Simulation (Frames/s)	101.85 vs 87.113
  CompuBench 1.5 Desktop — T-Rex (Frames/s)	0.744 vs 0.516
  65536 MB	65536 MB

  Overview of the Intel Core i7-6700K and i5-6600K

  Perhaps, on the eve of today’s premiere, Broadwell was deprived of attention at Computex 2015. Models of the C series are not so easy to buy today. However, we are impressed with the performance of the integrated graphics system, which consists of 48 execution units (EU) and 128 MB of internal memory. In addition, the chips come with unlocked multipliers, which can attract the attention of experienced users who build highly integrated and fast HTPCs. The disadvantages include a lower thermal package and clock speed, which is why the Core i7-5775C turned out to be slower than the Core i7-479 in our tests. 0K of the previous generation.

  Today we’re revealing the features of Devil Canyon’s true successors, the Core i7-6700K and Core i5-6600K. However, our analysis will be a little incomplete, as Intel will reveal details about the Skylake architecture at a special event within the IDF, which will be held towards the end of August. Today we have several processors and motherboards based on the Z170 chipset, cool DDR4 memory kits and the desire to find out how Skylake differs from previous solutions.

  Intel Core i7-6700K and i5-6600K overview | Features

  CPU-Z with prior Skylake support identifies the Core i7-6700K as a 95W TDP chip. However, all early Intel documentation excludes the 95W category. Today, Intel claims that the Core i7-6700K and Core i5-6600K have a TDP of 91W. The chips are inserted into the LGA 1151 socket, so installing a new CPU in an old LGA 1150 board will not work. This does not upset us at all, since Skylake is architecturally different from previous Core processors and requires new system logic. In other words, the Core i7-6700K and Core i5-6600K also involve a platform upgrade. Fortunately, Intel does not overprice new chips. The Core i7-6700K should retail for around $350 and the Core i5-6600K for $243.

  Both CPUs use four physical cores, we know this for sure. As in previous lines, i7 is equipped with Hyper-Threading technology. It separates the physical cores into logical ones in order to use the available resources more efficiently. On Windows, each physical Core i7 core appears as two logical cores. Core i5 does not have this technology, and its four cores process tasks in four threads.

  The configuration of the cache system is already familiar to us. The more expensive Core i7-6700K uses 8MB of last-level cache, while the Core i5-6600K has 6MB. According to CPU-Z, the Intel Skylake architecture equips each core with 256KB of L2 cache, along with 32KB of L1 cache for data and 32KB for instructions. Neither the Core i7-6700K nor the Core i5-6600K have integrated DRAM like the Core i7-5775C and i5-5675C.

  The base clock speed of the Core i7-6700K is as much as 4 GHz. However, the Turbo Boost profile is rather modest. With one core active, the Core i7-6700K has a ceiling of 4.2 GHz. Meanwhile, the Core i5-6600K starts at 3.5GHz and goes up to 3.9GHz. We will return to overclocking later, but for now we will say that the Core i7-6700K samples available in the laboratory feel comfortable at a frequency of up to about 4.7 GHz. We also tested one processor at 4.9 and then 4.8 GHz, but it was not stable enough in a long stress test. At a frequency of five gigahertz, the system could not boot to the end.

  Haswell-E-based Core i7 processors support quad-channel DDR4 memory subsystem. However, Skylake promises to make the technology more mainstream with its own dual-channel controller. The architecture officially supports DDR4 with transfer rates up to 2133 MT/s and DDR3L labeled up to 1600 MT/s. All motherboards in our lab only support DDR4. So when switching to new processors for enthusiasts, most likely, you will need not only a new motherboard, but also a set of RAM.

  K-series Skylake processors are armed with a graphics core called GT2, although it is not very interesting to enthusiasts. The company seems to have saved all the graphics details for IDF, which is understandable, so there’s not much point in discussing it today. However, the impressive benefits we saw in the GT3 on Broadwell are gone. In the intermediate solution for Skylake called HD Graphics 530, the number of EU blocks was reduced from 48 to 24, and the upper frequency bar is 1150 MHz.

  Notably, Skylake’s Gen9 core supports DirectX 12, OpenCL 2.0, OpenGL 4.4, and OpenGL ES, compared to DirectX 11.1 and OpenCL 1.2 in Gen8. When the engineers broach the subject of graphics at IDF, there will most likely be a lot of talk about power-oriented architectural optimizations. Of course, performance has not been left out either, primarily by improving the use of available bandwidth (hierarchical Z-buffer and in-stream memory color compression). Gen9 graphics coreimplemented in the Y, U, H and S lines and should cover all platforms from 4W to 91W. In terms of display interfaces, there is also much to discuss (GT4e should surprise us), but we will save the information we know for the next article.

  Both Skylake processors have 16 PCIe 3.0 lanes for connecting expansion cards and work in 1×16, 2×8 or 1×8/2×4 configurations. However, they differ from their predecessors in four lines that Intel leaves to connect the Platform Controller Hub. In past generations, the company called this interface DMI 2.0. It provided up to 2 GB/s of bidirectional bandwidth, which was sufficient for most drives, peripherals, and network interfaces. But the functionality of the 100-series chipsets has expanded, requiring more bandwidth on a given channel. Therefore, Intel has equipped the Core i7-6700K and Core i5-6600K processors with a Direct Media Interface 3.0 interface with a theoretical bandwidth of almost 4 GB / s.

  Intel Core i7-6700K and i5-6600K overview | Chipset Intel Z170

  For what specific needs is the extended channel implemented? The new Intel 100-series chipsets are perhaps even more interesting than the Core processors themselves. The company has not yet officially announced all the models, and Intel is only talking about the flagship Z170 logic. This is normal, because, from the point of view of enthusiasts, all other models differ only in the Z170 chipset feature set.

  Naturally, some features and technologies have been borrowed from the 9th series of chipsets. There is an integrated Gigabit Ethernet MAC with a single-lane connection for PHY. As before, you get six SATA 6Gb/s ports and HD Audio support.

  But there are many changes. Now the board can implement up to ten USB 3.0 ports and 14 USB 2.0 ports. The Z97, for example, offers just six USB 3.0 ports and fourteen USB 2.0 ports. Equally nice to get up to 20 PCI Express 3.0 lanes in the PCH. Thanks to software optimizations in the Intel Rapid Storage Technology driver, we ended up with the first desktop platform designed with PCIe storage in mind.

  However, you won’t get 20 PCIe lanes, six SATA ports, M.2 connectors, a full complement of USB and GbE at the same time. The Z170 is highly configurable, so motherboard manufacturers must choose how to use the available resources. Six of the 26 lanes are used for USB 3.0 (which is why the spec says «up to 20 PCIe lanes»). The manufacturer can use the remaining lines for four more USB 3.0 ports, SATA and PCIe storage. For example, using all six SATA 6 Gb / s ports «eats» six lanes. A PCIe drive occupies four lanes. On the Z170A Gaming M7 board we use for testing, MSI implemented four PCIe x1 slots and one PCIe x4 slot, an ASMedia ASM1142 USB 3.1 controller that takes two lanes, and two M.2 slots that share two more lanes. MSI provides several options for system configuration through the line switches on the PCIe slots.

  According to MSI’s instructions, the Intel RST driver still supports RAID 0, 1, 5, and 10 on SATA devices, but adds RAID 0 and 1 modes for PCIe-based SSDs in M.2 slots. You can use one of the slots in the PCH for a video card, and the increased bandwidth on DMI 3.0 will certainly improve the performance of this bundle compared to previous generation platforms. MSI claims that in this configuration, its board supports the connection of three video cards in CrossFire. But Nvidia continues to limit processors with 16 Gen 3 PCIe lanes to two cards in SLI.

  The above features are well shown in the block diagram. But apart from that, Z170 is the only chipset with official overclocking support (full support on K-series processors and partial overclocking of certain non-K-series models). With Skylake processors, Intel seems to have taken overclocking more seriously. Many limitations of previous generation architectures have been removed, so you can freely explore the potential of your processor.

  Intel Core i7-6700K and i5-6600K overview | Overclocking

  Overclockers don’t always welcome the decisions Intel makes in the pursuit of greater integration, greater efficiency, or optimized cost. We’ve already seen poor quality thermal interface material, clipping factors, and built-in voltage regulation all affect overclocking in their own way. But in general, the company pays attention to the wishes and criticism from enthusiasts.

  Sometimes changes are superficial in nature. Devil’s Canyon has implemented adjustments to Haswell’s power management and thermal efficiency. But we still ran into BCLK coefficients, because of which the base frequency could only be changed by a few megahertz. The new Skylake architecture allows Intel to rethink the operation of more significant nodes and change their behavior. So in many ways the Core i7-6700K and Core i5-6600K are a significant step forward, even though some details are not yet available. For example, it is not yet known what material Intel is using between the die and the heat spreader cover this time around.

  In the basic version, the reference frequency of the Platform Controller Hub to the PCIe bus and to the I/O interfaces is fixed at 100 MHz. A separate BCLK signal from the PCH makes it easy to tune the processor core, cache, graphics subsystem, and memory controller in 1 MHz steps up to 200 MHz.

  This signal is multiplied by several different factors to provide optimized clock speeds. For example, cores support ratios up to 83x — higher than Haswell’s 80x ceiling. However, in both cases they go far beyond the practical limits of processors. Like Haswell, Skylake’s ring bus ratios are not tied to cores. Multiplier adjustment up to 83x is also available for it, although you can lower the bus frequency if you suspect that it will have to deal with more aggressive overclocking of other components. The graphics core offers ratios up to 60x, similar to Haswell and Ivy Bridge. And if you have a Skylake mobile processor with built-in DRAM, adjustable coefficients will also provide an opportunity to improve performance through overclocking.

  Early Intel documentation suggested that memory overclocking would be available in multipliers of up to 24x (at 133 MHz) and 31x (at 100 MHz) with a top speed of around 3200 MT/s. However, in materials that appeared closer to the announcement, data transfer rates of up to 4133 MT/s were mentioned. We have memory kits capable of running at 3600 MT/s, so motherboard designers will have a lot to work on. Whereas previous architectures offered multipliers that changed the memory data rate in 200/266 GT/s increments, Skylake provides finer tuning in 100/133 MHz increments. The addition of XMP 2.0 is due to the new DDR4 specification. However, the XMP certification process itself has remained unchanged.

  The chip removed the built-in automatic voltage regulator, which many enthusiasts criticized, claiming that because of it their Haswell processors ran at elevated temperatures and required more expensive cooling systems. It remains to be seen what effect this factor has on the actual results, especially considering that we are dealing with a completely different architecture.

  Our field experience The

  Skylake has just arrived and we are currently testing prototypes. With a base frequency of 4 GHz and a peak of 4.2 GHz in Turbo Boost mode on the Core i7-6700K processor, our expectations about the scalability of the new architecture were not particularly optimistic. However, our experience suggests that 4.7 GHz may be a perfectly achievable target with minimal voltage increase.

  One of our samples even reached 4.9 GHz using a voltage setting of 1.41 V, but it was unstable under load and we consider this voltage level not comfortable for long-term operation.

  We easily found the point of failure of our Core i7-6700K using one megahertz BCLK adjustments. But to show that Intel’s more flexible BCLK settings are doing the job, we set the reference frequency to 115 MHz and took the following screenshot:

  We also spent more time testing DDR4 memory scaling. Corsair and G.Skill sent us 3200 MT/s bar sets. Corsair later offered us 3600 MT/s bars. We tested the memory on the MSI Z170A Gaming M7 board in 2133, 2400, 2666, 2933 and 3200 MHz modes. Both kits performed well with the default MSI settings of 3200 MT/s and their XMP profiles. But we weren’t able to get the claimed 3600 MHz as Windows kept crashing during the boot phase.

  A few hours before publication, MSI sent a firmware update that ensures stable operation in DRR4-3466 mode, albeit with less physical bandwidth than in 3200 mode. Sooner or later, even more aggressive memory settings will become a reality. Intel’s early overclocking guide suggested ratios for 3200 MT/s, now the company is talking about numbers like 4133 MT/s. Motherboard manufacturers are rushing to optimize their products for enthusiast RAM, which should appear shortly after the official release of Skylake, so we will return to this issue more than once.

  Bandwidth still scales with data rate. At 2133 MHz, we got >23 GB / s — not a bad result for a dual-channel controller. At the level of 3200 MHz, there were over 32 GB / s, that is, the increase was almost 40%.

  Of course, in reality, performance has not improved that much. Intel-based desktop configurations generally don’t suffer from data starvation with the tasks we use for benchmarks. So in the best case WinRAR speed increases by 6-7%. And this is in a test that is highly sensitive to the performance of the memory subsystem. Almost all other programs we use are less sensitive.

  Interestingly, the best performance is not achieved at the maximum data rate. Beyond DDR4-2933, throughput continues to go up, but in very small steps.

  Intel Core i7-6700K and i5-6600K overview | How We Tested

  Moving to LGA 1151 means we need to find a reference platform again. We already use the full range of MSI boards and have asked the company to show what they have available for the Z170 chipset. To be honest, we were impressed. The Z170A Gaming M7 looks impressive, both in terms of features and appearance.

  You can rest assured that MSI takes full advantage of the platform it works with. The CPU’s sixteen PCIe lanes can run in either x16 or x8/x8 configurations. The flexible PCN in this case supports as many as two M.2 slots, two SATA Express ports (or four SATA 6 Gb / s ports), two additional 6 Gb / s connectors, six USB 3. 0 ports, two more USB 3.1 10 Gb ports connected to two PCIe lanes. In addition to everything, the board has expansion slots with one and four lines.

  As already mentioned, Corsair and G.Skill sent in their memory kits, designed to make the most of Skylake’s memory controller. It is noteworthy that both solutions without problems reached the level of 3200 MT / s (this is already 1066 higher than the official specification of 2133 MT / s) even before the official release of the new Intel architecture. In addition, we overclocked the Corsair DDR4-3600 kit to 3466 MT/s before it crashed. We expect companies like MSI to optimize compatibility as quickly as possible.

  Most of the tests in today’s review were performed in the German laboratory of Tom’s Hardware. It uses test benches that created the 2015 processor ratings. In addition to analyzing processor performance, we’ll take a closer look at power consumption, thermal performance, and efficiency in workstation-centric applications.

  Intel Core i7-6700K and i5-6600K overview | Publishing & Multimedia

  Adobe CC

  We use Photoshop, InDesign and Illustrator from the Adobe CC software suite, and PCMark 8 Professional for workload management. Thus, we cover a relatively large range. The details of each test are presented in the tables below.

  Test results are affected by storage and background processes, as tests involve opening and closing each application and loading and saving files. For this reason, we use PCMark 8 to calculate the geometric mean of three bench tests (GEOMEAN).

  In contrast to other tests, in Adobe CC we noticed significant differences in performance under Windows 8.1 and 10. Therefore, where the differences are well outside the measurement error, we show both results. Naturally, we use the latest drivers available for each platform.

  In some applications, Intel Skylake processors showed much better performance under Windows 10 than under Windows 8.1. In the future, of course, we will return to these results, since in some cases the increase in speed was very significant. It’s also a mystery to us why AMD’s A10-7850K APU performed so well in Illustrator’s test. We repeated the test, but got the same result.

  The two Skylake processors lead the tests, as we expected. Please note that the Core i5-6600K is almost as good as, and sometimes even slightly superior to, the Core i7-6700K, despite the lower frequency. It looks like SMT slows down the overall performance in Microsoft Office a bit.

  review of Intel Core i5-6600K and Intel Core i7-6700K processors Part 1: Architecture and Platform Overview

  Let me remind you that the first part covered all the known information about the Intel Skylake desktop processors, as well as the LGA1151 platform in general. The company itself positions its new products as ideal solutions for overclockers and gamers. Plus, Intel promises a 10 percent performance increase over the previous generation. This is what I will check today.

  Specifications

  Let’s take another look at the technical specifications of the Intel Skylake CPUs. Frequencies are pleasing. Therefore, the performance should also please. However, which is not at all surprising, due to the lack of serious competition, the Core i5-6600K and Core i7-6700K have … processors from Intel. Thus, the Core i7-4790K continues to be the chip with the highest operating speed — 4.4 GHz in Turbo Boost mode.

  9L
  • Motherboard: ASUS Z170-Deluxe
  • Video Card: GIGABYTE GeForce GTX 980 Ti 6 GB GDDR5
  • RAM: DDR4-2133 2x 8 GB
  • Storage: OCZ Vertex 3 306 byte PSU1 LEPA G1600, 1600 W
  • Peripherals: Samsung U28D590D, ROCCAT ARVO, ROCCAT SAVU
  • Operating system: Windows 8.1 x64

  Calculations

  At such low frequencies for this standard, it has no advantages over DDR3. In addition, DDR4 modules tend to operate at very high latency. I repeat, the situation will change dramatically when affordable solutions with an effective frequency of 4000+ MHz appear on the market. Plus the timings will go down. This is a standard situation: both the release of DDR2 and the announcement of DDR3 were accompanied by similar problems. The graph below clearly shows that DDR4-2133 and DDR3-2133 show almost identical results. It was with this memory that Haswell and Broadwell processors were tested.

  And now let’s move on to x86 calculations. We found that the Skylake architecture is on average 5-10% faster than Haswell and Broadwell. But those studies were carried out at the same speed for all chips. Gate switching frequency affects the performance of processors is not the least. In the CINEBENCH R15 benchmark, the Core i7-6700K is 4.8% faster than the Core i7-4790K, and the Core i5-6600K is 1.3% faster than the Core i5-4690K. As you can see, in this application, the difference between these processors is negligible.

  A similar situation is observed in the Fryrender benchmark: the Core i7-6700K is ahead of the Core i7-4790K by 8.5%, and the Core i5-6600K is faster than the Core i5-4690K by 5%. True, among the Core i5 models, the “stone” of the Broadwell generation is generally the leader.

  The next benchmark is LuxMark. If we consider only the speed of the CPU component, then the Core i7-6700K and Core i7-4790K have parity. However, when using heterogeneous computing (that is, using the GPU), the Skylake processor is noticeably faster. Core i7-6700K outperforms Core i7-479 in this test pattern0K by 7%. The leader is the hybrid processor Core i5-5675C. Thanks to the L4 cache and very powerful Iris Pro 6200 graphics.

  The importance of a high clock speed in Photoshop draws attention. Therefore, both the Core i5-6600K and the Core i5-4690K are seriously ahead of the Core i5-5675C: by 17.5% and 11%, respectively. There is equality between the Core i7-6700K and the Core i7-4790K.

  Archivers respond very well to the presence of the fourth level cache. That is why the Core i5-5675C is very nimble both in the benchmark and in real data compression. In the WinRAR test, the Core i7-6700K turned out to be the leader among quad-core models. However, in real conditions, he lost to the Core i7-4790K.

  Other applications show a similar picture. So, the Core i7-6700K is faster than the Core i7-4790K, and the Core i5-6600K is ahead of the Core i5-4690K. But in all cases, the difference between these processors is insignificant: an average of 5-10%. Sometimes even less. There is nothing surprising in this, these are the realities that have been going on for several years. And both at Intel, and at AMD.

  Integrated graphics

  As we have already found out, HD Graphics 530 has 24 unified devices in its arsenal. Four elements more than the HD Graphics 4600 used in Haswell processors. Therefore, one should not expect any tangible results from the integrated Skylake graphics. This is not the Iris Pro 6200, which can compete even with a discrete graphics card like the GeForce GTX 750. In my opinion, it makes no sense to focus on the performance of the integrated graphics of processors designed for enthusiasts and gamers. AT 9In 9% of cases, a more efficient discrete 3D card will be installed in the system unit.

  The difference in FPS between the Core i5-6600K and Core i7-6700K is due to the fact that the first HD Graphics 530 chip runs at 1100 MHz, while the second runs at 1150 MHz.

  CPU dependency

  It is much more important to know how Skylake is doing with CPU dependency. Will these chips be able to “pump” even the fastest video cards, because the product will be promoted mainly among gamers? To do this, I took the graphics adapter GeForce GTX 980 Ti and tested the test bench in three modern games. The settings are selected in such a way that 3D accelerators are not loaded to 100%. So, anti-aliasing was not used, as well as NVIDIA’s proprietary technologies.

  In Full HD, the effect of processor dependence was the least noticeable. The Core i7-4790K turned out to be the most efficient. That is, a chip based on the Haswell architecture. However, the Core i7-6700K literally breathes down his neck. The slowest was the Core i5-6600K. This stone is noticeably inferior to the Core i5-4690K, and Core i5-5675C. As for Broadwell, I can’t help but notice how responsive GTA V turned out to be to the presence of the fourth level cache. In the most resource-intensive game from the list — «The Witcher» — in general, the discrepancy in the results is due solely to the measurement error. In GTA V, the Core i5-5675C again turned out to be the leader. The Core i7-4790K is again ahead of the Core i7-6700K. But between Core i5-6600K and Core i5-4690K there is relative equality.

  Hyper-Threading is useless in modern games.

  In fact, the problem of processor dependence is relevant for less efficient chips. Especially if they don’t have an unlocked multiplier and any sort of overclocking capability. In the case of the Core i5-6600K and Core i7-6700K, everything is solved very simply. We set a higher frequency, and now the GeForce GTX 980 Ti is completely “pumped”. Skylake overclocking will be discussed further.

  Overclocking, Power Consumption, and Cooling Requirements

  Haswell processors got very hot at the time. Partially this problem was solved simultaneously with the release of cooler models of the Devil’s Canyon family. Despite the fact that the declared TDP level, for example, for the Core i7-4790K is not too high (only 88 W), its successful overclocking is hindered by high heating. There are several reasons: too elongated shape of the crystal, three-dimensional transistors, which themselves heat up unevenly, the use of a mediocre thermal interface between the chip and the heat-distributing cover. As a result, even the most monstrous coolers have difficulty cooling the Core i7-4790K with a standard overclock of 4.5 GHz. Skylake processors have an even higher TDP level — 91 W …

  However, both the Core i5-6600K and Core i7-6700K heat up noticeably less than the «devil» processors! For this test, we used the Noctua NH-D9L cooler with the fan speed locked at 1500 rpm. If the Core i7-6700K heated up to only 69 degrees with a stable frequency of 4000 MHz, then the Core i7-4790K «fired up» to 98 degrees Celsius, but at a higher stable speed of 4200 MHz. Similar results were demonstrated by a couple in the face of the Core i5-6600K and Core i5-4690K. Let me remind you that the maximum allowable temperature for Skylake, at which throttling is not activated, is 100 degrees Celsius.

  As a result, the Core i5-6600K and Core i7-6700K consume less power under load than the Core i5-4690K and Core i7-4790K.

  The unlocked multiplier is an important and useful attribute of the CPU. I dare to assume that the Core i5-6600K and Core i7-6700K models will be the only chips in the entire Skylake lineup equipped with this overclocking element. The built-in power converter has now migrated back to the motherboard, so the new crystals work with a high voltage of 1.2 V. At home, when the system is running 24/7, I recommend not to increase the potential difference above 1. 4 V. In the test bench I used an ASUS motherboard. To prevent the processor frequency from being reset during overclocking, I had to activate the eighth calibration level for the CPU Load-Line Calibration option.

  As a result, the Core i5-6600K was overclocked to an absolutely stable 4600 MHz, and the Core i7-6700K to 4700 MHz. In the second case, I was able to boot the operating system even at 4900 MHz. So there is reason to believe that there are more successful models in the world. For overclocking, a maintenance-free water cooling system ENERMAX LIQTECH 240 was used. With a Core i7-6700K running at 4700 MHz, she coped with difficulty.

  Skylake lid has a polymer thermal interface.

  The excellent overclocking potential of Skylake processors is confirmed by the world records that have fallen like from a cornucopia. So, overclocker John Lam, using liquid nitrogen, overclocked the Core i7-6700K sample to 6998.88 MHz. Just a little bit was not enough to conquer the psychological mark of 7 GHz! He also managed to overclock the DDR4 memory module to an effective 4901. 6 MHz.

  But that’s not all. Skylake processors can finally be overclocked on the bus! From now on, the BCLK frequency is not locked and is not linked to other buses via the CPU Strap function in 100/125/166 MHz steps. It can be freely increased with an accuracy of 1 MHz. All this allows us to hope for two things. Firstly, from now on, overclocking of processors without an unlocked multiplier will return again. According to my observations, not only the devices in question, but also the younger models should be overclockable. Secondly, it will be easier to overclock memory and use kits with pre-installed XMP profiles. Unlocking the bus is related to two things: decoupling the PCI Express 3.0 and DMI 3.0 buses from the BCLK and getting rid of the built-in power converter. When overclocking over BCLK, you only need to increase the memory divider so as not to hit the DDR4 frequency ceiling, as well as reduce the L3 cache frequency multiplier. They retained their connection with the clock generator.

  IMPORTANT! As it turned out, overclocking of processors on the bus is relevant only for models with the letter K in the name. Unfortunately, the rest of the Skylake chips, lacking an unlocked multiplier, do not run on the bus. This is due to the fact that their DMI and BCLK bus frequencies are rigidly tied to BCLK. Therefore, a slight deviation from the reference 100 MHz will simply lead to unstable system operation. It is unfortunate that Intel did this with a line of non-overclockable processors.

  In one of the editorial columns, I had already buried overclocking. However, in this regard, Skylake processors are pleasing. The models have high clock speeds, the unlocked multiplier has not disappeared, plus Intel has returned to the origins of overclocking. It’s nice to be wrong sometimes!

  In conclusion

  Which year, one can even say by tradition, at the end of the next review of the new processor, sacred words are written in the style: the next generation of chips turned out to be faster than the previous one by just a little bit . Skylake is no exception. If you look at the results, the solutions studied in the test lab are ahead of their predecessors by an average of 5-10%. It is logical that users who already use the services of Haswell or Ivy Bridge do not make sense to immediately run to the store for a new “stone”. In addition, the entire platform will have to be changed. But for those who are going to assemble the system unit from scratch, there is a reason to take a closer look at Skylake and LGA1151.

  Core i5-6600K and Core i7-6700K integrated graphics are 20-40% faster on average than Core i5-4690K and Core i7-4790K. In this area, Intel annually demonstrates significant growth. However, the considered processors, even by the manufacturer itself, are positioned as solutions for enthusiasts. Therefore, in the vast majority of cases, a discrete video card will be installed in the motherboard along with the Skylake chip.

  The LGA1151 platform can safely be considered the most progressive among all currently relevant platforms.