Ryzen 5 3600x vs ryzen 7 5800x: AMD Ryzen 5 3600X vs AMD Ryzen 7 5800X: What is the difference?

AMD Ryzen 5 3600X vs AMD Ryzen 7 5800X: What is the difference?

62points

AMD Ryzen 5 3600X

59points

AMD Ryzen 7 5800X

3600XT

64 facts in comparison

AMD Ryzen 5 3600X

AMD Ryzen 7 5800X

Why is AMD Ryzen 5 3600X better than AMD Ryzen 7 5800X?

5°C higher maximum operating temperature?
95°Cvs90°C
10W lower TDP?
95Wvs105W
1.33MB/core more L3 cache per core?
5.33MB/corevs4MB/core
Supports ECC memory?
6.96% cheaper?
282.77 €vs303.94 €

Why is AMD Ryzen 7 5800X better than AMD Ryzen 5 3600X?

33.33% faster CPU speed?
8 x 3.8GHzvs6 x 3.8GHz
4 more CPU threads?
16vs12
1MB bigger L2 cache?
4MBvs3MB
1.57x higher PassMark result?
28752vs18325
0. 3GHz higher turbo clock speed?
4.7GHzvs4.4GHz
128KB bigger L1 cache?
512KBvs384KB
31.18% higher PassMark result (single)?
3513vs2678
49.54% higher multi-core Geekbench 5 result?
10330vs6908

Which are the most popular comparisons?

AMD Ryzen 5 3600X

AMD Ryzen 5 5600X

AMD Ryzen 7 5800X

Intel Core i7-12700K

AMD Ryzen 5 3600X

AMD Ryzen 5 5500

AMD Ryzen 7 5800X

AMD Ryzen 9 5900X

AMD Ryzen 5 3600X

AMD Ryzen 5 3600

AMD Ryzen 7 5800X

AMD Ryzen 7 5700X

AMD Ryzen 5 3600X

AMD Ryzen 5 5600G

AMD Ryzen 7 5800X

AMD Ryzen 7 5700G

AMD Ryzen 5 3600X

AMD Ryzen 5 5500U

AMD Ryzen 7 5800X

AMD Ryzen 5 5600X

AMD Ryzen 5 3600X

AMD Ryzen 7 2700X

AMD Ryzen 7 5800X

Intel Core i5-12600K

AMD Ryzen 5 3600X

AMD Ryzen 7 3700X

AMD Ryzen 7 5800X

AMD Ryzen 7 3700X

AMD Ryzen 5 3600X

AMD Ryzen 5 4500U

AMD Ryzen 7 5800X

Intel Core i7-11700K

AMD Ryzen 5 3600X

Intel Core i5-10400

AMD Ryzen 7 5800X

Intel Core i7-10700K

AMD Ryzen 7 5800X

AMD Ryzen 7 3800X

Price comparison

Cheap alternatives

User reviews

Overall Rating

AMD Ryzen 5 3600X

2 User reviews

AMD Ryzen 5 3600X

9. 0/10

2 User reviews

AMD Ryzen 7 5800X

3 User reviews

AMD Ryzen 7 5800X

9.7/10

3 User reviews

Features

Value for money

10.0/10

2 votes

9.7/10

3 votes

Gaming

9.5/10

2 votes

9.7/10

3 votes

Performance

9.0/10

2 votes

9.7/10

3 votes

Reliability

9.5/10

2 votes

9.7/10

3 votes

Energy efficiency

9.5/10

2 votes

9.7/10

3 votes

Performance

1.CPU speed

6 x 3.8GHz

8 x 3.8GHz

The CPU speed indicates how many processing cycles per second can be executed by a CPU, considering all of its cores (processing units). It is calculated by adding the clock rates of each core or, in the case of multi-core processors employing different microarchitectures, of each group of cores.

2. CPU threads

More threads result in faster performance and better multitasking.

3.turbo clock speed

4.4GHz

4.7GHz

When the CPU is running below its limitations, it can boost to a higher clock speed in order to give increased performance.

4.Has an unlocked multiplier

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

Some processors come with an unlocked multiplier which makes them easy to overclock, allowing you to gain increased performance in games and other apps.

5.L2 cache

A larger L2 cache results in faster CPU and system-wide performance.

6.L3 cache

A larger L3 cache results in faster CPU and system-wide performance.

7.L1 cache

A larger L1 cache results in faster CPU and system-wide performance.

8.L2 core

0.5MB/core

0. 5MB/core

More data can be stored in the L2 cache for access by each core of the CPU.

9.L3 core

5.33MB/core

4MB/core

More data can be stored in the L3 cache for access by each core of the CPU.

Memory

1.RAM speed

3200MHz

It can support faster memory, which will give quicker system performance.

2.maximum memory bandwidth

47.68GB/s

This is the maximum rate that data can be read from or stored into memory.

3.DDR memory version

DDR (Double Data Rate) memory is the most common type of RAM. Newer versions of DDR memory support higher maximum speeds and are more energy-efficient.

4.memory channels

More memory channels increases the speed of data transfer between the memory and the CPU.

5. maximum memory amount

The maximum amount of memory (RAM) supported.

6.bus transfer rate

Unknown. Help us by suggesting a value. (AMD Ryzen 5 3600X)

Unknown. Help us by suggesting a value. (AMD Ryzen 7 5800X)

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

7.Supports ECC memory

✔AMD Ryzen 5 3600X

✖AMD Ryzen 7 5800X

Error-correcting code memory can detect and correct data corruption. It is used when is it essential to avoid corruption, such as scientific computing or when running a server.

8.eMMC version

Unknown. Help us by suggesting a value. (AMD Ryzen 5 3600X)

Unknown. Help us by suggesting a value. (AMD Ryzen 7 5800X)

A higher version of eMMC allows faster memory interfaces, having a positive effect on the performance of a device. For example, when transferring files from your computer to the internal storage over USB.

9.bus speed

Unknown. Help us by suggesting a value. (AMD Ryzen 5 3600X)

Unknown. Help us by suggesting a value. (AMD Ryzen 7 5800X)

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

Benchmarks

1.PassMark result

This benchmark measures the performance of the CPU using multiple threads.

2.PassMark result (single)

This benchmark measures the performance of the CPU using a single thread.

3.Geekbench 5 result (multi)

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

4.Cinebench R20 (multi) result

Cinebench R20 is a benchmark tool that measures a CPU’s multi-core performance by rendering a 3D scene.

5.Cinebench R20 (single) result

Cinebench R20 is a benchmark tool that measures a CPU’s single-core performance by rendering a 3D scene.

6.Geekbench 5 result (single)

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

7.Blender (bmw27) result

235.9seconds

Unknown. Help us by suggesting a value. (AMD Ryzen 7 5800X)

The Blender (bmw27) benchmark measures the performance of a processor by rendering a 3D scene. More powerful processors can render the scene in less time.

8.Blender (classroom) result

709seconds

Unknown. Help us by suggesting a value. (AMD Ryzen 7 5800X)

The Blender (classroom) benchmark measures the performance of a processor by rendering a 3D scene. More powerful processors can render the scene in less time.

9.performance per watt

This means the CPU is more efficient, giving a greater amount of performance for each watt of power used.

Features

1. uses multithreading

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

Multithreading technology (such as Intel’s Hyperthreading or AMD’s Simultaneous Multithreading) provides increased performance by splitting each of the processor’s physical cores into virtual cores, also known as threads. This way, each core can run two instruction streams at once.

2.Has AES

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

AES is used to speed up encryption and decryption.

3.Has AVX

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

AVX is used to help speed up calculations in multimedia, scientific and financial apps, as well as improving Linux RAID software performance.

4.SSE version

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

5. Has F16C

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

F16C is used to speed up tasks such as adjusting the contrast of an image or adjusting volume.

6.bits executed at a time

Unknown. Help us by suggesting a value. (AMD Ryzen 5 3600X)

Unknown. Help us by suggesting a value. (AMD Ryzen 7 5800X)

NEON provides acceleration for media processing, such as listening to MP3s.

7.Has MMX

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

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

8.Has TrustZone

✖AMD Ryzen 5 3600X

✖AMD Ryzen 7 5800X

A technology integrated into the processor to secure the device for use with features such as mobile payments and streaming video using digital rights management (DRM).

9.front-end width

Unknown. Help us by suggesting a value. (AMD Ryzen 5 3600X)

Unknown. Help us by suggesting a value. (AMD Ryzen 7 5800X)

The CPU can decode more instructions per clock (IPC), meaning that the CPU performs better

Price comparison

Cancel

Which are the best CPUs?

Ryzen 7 5800X vs Ryzen 5 3600X

Availability

AMD Ryzen 7 5800X 8-core, 16-Thread Unlocked Desktop Processor

Buy on Amazon

$269.99

In Stock

Updated 55 minutes ago

Availability

AMD Ryzen 5 3600X 6-Core, 12-Thread Unlocked Desktop Processor with Wraith Spire Cooler

Buy on Amazon

$245

In Stock

Updated 54 minutes ago

Key Differences

In short, we have a clear winner — Ryzen 7 5800X outperforms the cheaper Ryzen 5 3600X on the selected game parameters, and is also a better bang for your buck! The better performing Ryzen 7 5800X is 487 days newer than the cheaper Ryzen 5 3600X.

Advantages of AMD Ryzen 7 5800X

Performs up to 16% better in Counter-Strike: Global Offensive than Ryzen 5 3600X — 774 vs 668 FPS
Up to 5% better value when playing Counter-Strike: Global Offensive than Ryzen 5 3600X — $0. 35 vs $0.37 per FPS
Can execute more multi-threaded tasks simultaneously than AMD Ryzen 5 3600X — 16 vs 12 threads

Advantages of AMD Ryzen 5 3600X

Counter-Strike: Global Offensive

Resolution

1920×1080

Game Graphics

High

Ryzen 7 5800X

Desktop • Nov 5th, 2020

FPS

774

100%

Value, $/FPS

$0.35/FPS

100%

Price, $

$269.99

90%

FPS and Value Winner

AMD Ryzen 7 5800X 8-core, 16-Thread Unlocked Desktop Processor

Buy for $269.99 on Amazon

In Stock

Updated 55 minutes ago

Ryzen 5 3600X

Desktop • Jul 7th, 2019

FPS

668

84%

Value, $/FPS

$0.37/FPS

95%

Price, $

$245

100%

AMD Ryzen 5 3600X 6-Core, 12-Thread Unlocked Desktop Processor with Wraith Spire Cooler

Buy for $245 on Amazon

In Stock

Updated 54 minutes ago

Resolution

1920×1080

Game Graphics

High

Ryzen 7 5800X

Desktop • Nov 5th, 2020

Ryzen 5 3600X

Desktop • Jul 7th, 2019

1252

FPS

1086

FPS

League of Legends

550

FPS

474

FPS

VALORANT

316

FPS

272

FPS

Grand Theft Auto V

349

FPS

298

FPS

Apex Legends

340

FPS

293

FPS

Fortnite

Geekbench 5 Benchmarks

Ryzen 7 5800X

Desktop • Nov 5th, 2020

Single-Core

1671

100%

Multi-Core

10337

100%

Ryzen 5 3600X

Desktop • Jul 7th, 2019

Single-Core

1246

66%

Multi-Core

6874

50%

AMD Ryzen 7 5800X	vs	AMD Ryzen 5 3600X
Nov 5th, 2020	Release Date	Jul 7th, 2019
Ryzen 7	Collection	Ryzen 5
Vermeer	Codename	Matisse
AMD Socket AM4	Socket	AMD Socket AM4
Desktop	Segment	Desktop
8	Cores	6
16	Threads	12
3. 8 GHz	Base Clock Speed	3.8 GHz
4.7 GHz	Turbo Clock Speed	4.4 GHz
105 W	TDP	95 W
7 nm	Process Size	7 nm
38.0x	Multiplier	38.0x
None	Integrated Graphics	None
Yes	Overclockable	Yes

Builds Using Ryzen 7 5800X or Ryzen 5 3600X

Counter-Strike: Global Offensive, 1080p, High

ASUS ROG Strix GL10 Gaming Desktop PC

344 FPS

$1.86/FPS

GeForce GTX 1660 Ti

Ryzen 5 3600X

8 GB, 256 GB SSD

Buy on Amazon

$639

In Stock

Updated 38 minutes ago

ASUS ROG Gaming Desktop PC

344 FPS

$1.89/FPS

GeForce GTX 1660 Ti

Ryzen 5 3600X

16 GB, N/A Storage

Buy on Amazon

$649

In Stock

Updated 39 minutes ago

HP OMEN Gaming Premium Desktop PC

738 FPS

$2. 03/FPS

GeForce RTX 3080

Ryzen 7 5800X

16 GB, 1 TB SSD

Buy on Amazon

$1,499.99

In Stock

Updated 37 minutes ago

HP OMEN Gaming Premium Desktop PC

738 FPS

$2.17/FPS

GeForce RTX 3080

Ryzen 7 5800X

32 GB, 1 TB SSD

Buy on Amazon

$1,599.99

In Stock

Updated 38 minutes ago

HP OMEN Gaming Premium Desktop PC

738 FPS

$2.2/FPS

GeForce RTX 3080

Ryzen 7 5800X

16 GB, 1 TB SSD + 2 TB HDD

Buy on Amazon

$1,619.99

In Stock

Updated 38 minutes ago

ASUS ROG Strix GL10 Gaming Desktop PC

344 FPS

$2.21/FPS

GeForce GTX 1660 Ti

Ryzen 5 3600X

16 GB, 512 GB SSD

Buy on Amazon

$759

In Stock

Updated 38 minutes ago

HP OMEN Gaming Premium Desktop PC

738 FPS

$2.25/FPS

GeForce RTX 3080

Ryzen 7 5800X

32 GB, 1 TB SSD + 2 TB HDD

Buy on Amazon

$1,659. 99

In Stock

Updated 38 minutes ago

2022 HP OMEN Gaming Desktop PC

738 FPS

$2.3/FPS

GeForce RTX 3080

Ryzen 7 5800X

16 GB, 1 TB SSD

Buy on Amazon

$1,699

In Stock

Updated 37 minutes ago

ROG Strix GA15DH Gaming Desktop PC

363 FPS

$2.31/FPS

GeForce GTX 1660 SUPER

Ryzen 5 3600X

8 GB, 256 GB SSD + 1 TB HDD

Buy on Amazon

$839.99

In Stock

Updated 48 minutes ago

2022 ASUS ROG Strix GL10 Gaming Desktop PC

344 FPS

$2.32/FPS

GeForce GTX 1660 Ti

Ryzen 5 3600X

16 GB, 256 GB SSD + 2 TB HDD

Buy on Amazon

$799

In Stock

Updated 38 minutes ago

Select from the most popular similar processor comparisons. Most compared processor combinations, including the currently selected ones, are at the top.

Ryzen 5 5600X

$193.99

Ryzen 7 5800X

$269. 99

$269.99

Ryzen 5 2600X

Ryzen 7 5800X

$269.99

Ryzen 7 2700X

Ryzen 7 5800X

$269.99

Ryzen 7 3800X

AMD Ryzen 7 5800X3D vs Ryzen 5 3600X: performance comparison

AMD Ryzen 7 5800X3D

AMD Ryzen 5 3600X

We compared two desktop CPUs: the 3.4 GHz AMD Ryzen 7 5800X3D with 8-cores against the 3.8 GHz Ryzen 5 3600X with 6-cores. On this page, you’ll find out which processor has better performance in benchmarks, games and other useful information.

Review
Differences
Performance
Specs
Comments

Review

General overview and comparison of the processors

Single-Core Performance

Performance in single-threaded apps and benchmarks

Ryzen 7 5800X3D

Ryzen 5 3600X

Performance

Measure performance when all cores are involved

Ryzen 7 5800X3D

Ryzen 5 3600X

Power Efficiency

The efficiency score of electricity consumption

Ryzen 7 5800X3D

Ryzen 5 3600X

NanoReview Final Score

Generic CPU rating

Ryzen 7 5800X3D

Ryzen 5 3600X

Key Differences

What are the key differences between 3600X and 5800X3D

Advantages of AMD Ryzen 7 5800X3D

Newer — released 2-years and 8-months later
Has 64 MB larger L3 cache size
Has 2 more physical cores
23% faster in a single-core Geekbench v5 test — 1633 vs 1325 points
2% higher Turbo Boost frequency (4. 5 GHz vs 4.4 GHz)

Advantages of AMD Ryzen 5 3600X

Unlocked multiplier
Consumes up to 10% less energy than the Ryzen 7 5800X3D – 95 vs 105 Watt

Benchmarks

Comparing the performance of CPUs in benchmarks

Cinebench R23 (Single-Core)

Ryzen 7 5800X3D
+10%

1452

Ryzen 5 3600X

1320

Cinebench R23 (Multi-Core)

Ryzen 7 5800X3D
+56%

14685

Ryzen 5 3600X

9437

Passmark CPU (Single-Core)

Ryzen 7 5800X3D
+16%

3100

Ryzen 5 3600X

2665

Passmark CPU (Multi-Core)

Ryzen 7 5800X3D
+50%

27319

Ryzen 5 3600X

18241

Geekbench 5 (Single-Core)

Ryzen 7 5800X3D
+23%

1625

Ryzen 5 3600X

1323

Geekbench 5 (Multi-Core)

Ryzen 7 5800X3D
+59%

11576

Ryzen 5 3600X

7288

▶️ Submit your Cinebench R23 result

By purchasing through links on this site, we may receive a commission from Amazon. This does not affect our assessment methodology.

Specifications

Full technical specification of AMD Ryzen 7 5800X3D and Ryzen 5 3600X

General

Vendor	AMD	AMD
Released	March 15, 2022	July 7, 2019
Type	Desktop	Desktop
instruction set	x86-64	x86-64
Codename	Vermeer	Zen 2
Socket	AM4	AM4
Integrated GPU	No	No

Performance

Cores	8	6
Threads	16	12
Base Frequency	3. 4 GHz	3.8 GHz
Turbo Boost Frequency	4.5 GHz	4.4 GHz
Bus frequency	100 MHz	100 MHz
Multiplier	34x	38x
L1 Cache	64K (per core)	64K (per core)
L2 Cache	512K (per core)	512K (per core)
L3 Cache	96MB (shared)	32MB (shared)
Unlocked Multiplier	No	Yes

Transistors	—	3.8 billions
Fabrication process	7 nm	7 nm
TDP	105 W	95 W
Max. temperature	90°C	95°C

Memory support

Memory types	DDR4-3200, DDR4-2933, DDR4-2667	DDR4-3200
Memory Size	128 GB	128 GB
Max. Memory Channels	2	2
Max. Memory Bandwidth	47.68 GB/s	47.68 GB/s
ECC Support	Yes	Yes

Official site	AMD Ryzen 7 5800X3D official page	AMD Ryzen 5 3600X official page
PCI Express Version	4.0	4.0
PCI Express Lanes	20	16

Cast your vote

Choose between two processors

Ryzen 7 5800X3D

5 (83.3%)

Ryzen 5 3600X

1 (16.7%)

Total votes: 6

ompetitors

1.
AMD Ryzen 5 5600X vs AMD Ryzen 7 5800X3D

2.
AMD Ryzen 7 3700X vs AMD Ryzen 7 5800X3D

3.
AMD Ryzen 9 5950X vs AMD Ryzen 7 5800X3D

4.
Intel Core i7 12700K vs AMD Ryzen 7 5800X3D

5.
AMD Ryzen 9 3900X vs AMD Ryzen 7 5800X3D

6.
AMD Ryzen 5 5600G vs AMD Ryzen 5 3600X

7.
Intel Core i5 12400F vs AMD Ryzen 5 3600X

8.
AMD Ryzen 5 5500 vs AMD Ryzen 5 3600X

So which CPU will you choose: AMD Ryzen 5 3600X or Ryzen 7 5800X3D?

Name

Message

Promotion

Ryzen 7 5800X vs Ryzen 5 3600X — Call of Duty Modern Warfare with GTX 1660 Benchmarks 1080p, 1440p, Ultrawide, 4K Comparison

GTX 1660 with

AMD Ryzen 7 5800X

Call of Duty Modern Warfare

GTX 1660 with

AMD Ryzen 5 3600X

Ryzen 7 5800X
Ryzen 5 3600X

Multi-Thread Performance

24731 Pts

21536 Pts

Single-Thread Performance

3051 Pts

2910 Pts

Call of Duty Modern Warfare

Ryzen 7 5800X vs Ryzen 5 3600X in Call of Duty Modern Warfare using GTX 1660 — CPU Performance comparison at Ultra, High, Medium, and Low Quality Settings with 1080p, 1440p, Ultrawide, 4K resolutions

Ryzen 7 5800X
Ryzen 5 3600X

Ultra Quality

Resolution	Frames Per Second
1080p	116. 6 FPS
1080p	109.3 FPS
1440p	81.1 FPS
1440p	78.7 FPS
2160p	48.4 FPS
2160p	47.0 FPS
w1440p	69.3 FPS
w1440p	67.3 FPS

High Quality

Resolution	Frames Per Second
1080p	179.2 FPS
1080p	169.8 FPS
1440p	132.0 FPS
1440p	128.7 FPS
2160p	85.0 FPS
2160p	82.9 FPS
w1440p	115.5 FPS
w1440p	112.7 FPS

Medium Quality

Resolution	Frames Per Second
1080p	241. 9 FPS
1080p	230.3 FPS
1440p	183.0 FPS
1440p	178.7 FPS
2160p	121.7 FPS
2160p	118.9 FPS
w1440p	161.8 FPS
w1440p	158.1 FPS

Low Quality

Resolution	Frames Per Second
1080p	367.2 FPS
1080p	351.3 FPS
1440p	284.8 FPS
1440p	278.8 FPS
2160p	195.0 FPS
2160p	190.7 FPS
w1440p	254.3 FPS
w1440p	248.9 FPS

Ryzen 7 5800X

The Ryzen 7 5800X has more cores. The benefit of having more cores is that the system can handle more threads. Each core can handle a separate stream of data. This architecture greatly increases the performance of a system that is running concurrent applications.
The Ryzen 7 5800X has more threads. Larger programs are divided into threads (small sections) so that the processor can execute them simultaneously to get faster execution.
The Ryzen 7 5800X is more power efficient and generates less heat.

Ryzen 5 3600X

For some games, a cpu with a higher clock speed, or in a technical name IPC (Instructions per clock), has better results than other CPU’s with higher core count and lower core speed.

Compare Ryzen 7 5800X vs Ryzen 5 3600X specifications

Ryzen 7 5800X vs Ryzen 5 3600X Architecture

	Ryzen 7 5800X	Ryzen 5 3600X
Codename	Vermeer	Zen 2
Generation	Ryzen 7 (Zen 3 (Vermeer))	Ryzen 5 (Zen 2)
Market	Desktop	Desktop
Memory Support	DDR4	DDR4
Part#	unknown	unknown
Production Status	Active	Active
Released	Nov 5th, 2020	Jul 2019

Ryzen 7 5800X vs Ryzen 5 3600X Cache

	Ryzen 7 5800X	Ryzen 5 3600X
Cache L1	64K (per core)	96K (per core)
Cache L2	512K (per core)	512K (per core)
Cache L3	32MB	32MB (shared)

Ryzen 7 5800X vs Ryzen 5 3600X Cores

	Ryzen 7 5800X	Ryzen 5 3600X
# of Cores	8	6
# of Threads	16	12
Integrated Graphics	N/A	N/A
SMP # CPUs	1	1

Ryzen 7 5800X vs Ryzen 5 3600X Performance

	Ryzen 7 5800X	Ryzen 5 3600X
Base Clock	100 MHz	100 MHz
Frequency	3. 8 GHz	3.8 GHz
Multiplier	38.0x	38.0x
Multiplier Unlocked	Yes	Yes
TDP	105 W	95 W
Turbo Clock	up to 4.7 GHz	up to 4.4 GHz
Voltage	variable	variable

Ryzen 7 5800X vs Ryzen 5 3600X Physical

	Ryzen 7 5800X	Ryzen 5 3600X
Die Size	unknown	unknown
Foundry	TSMC	TSMC
Package
Process Size	7 nm	7 nm
Socket	AMD Socket AM4	AMD Socket AM4
Transistors	19,200 million	4,800 million
tCaseMax	unknown	unknown

Share Your Comments 0

2009

Minecraft

2009

AMD Ryzen 7 5800X vs.

AMD Ryzen 5 3600X

AMD Ryzen 7 5800X

The AMD Ryzen 7 5800X operates with 8 cores and 16 CPU threads. It run at 4.70 GHz base 4.40 GHz all cores while the TDP is set at 105 W.The processor is attached to the AM4 CPU socket. This version includes 32.00 MB of L3 cache on one chip, supports 2 memory channels to support DDR4-3200 RAM and features 4.0 PCIe Gen 20 lanes. Tjunction keeps below 95 °C degrees C. In particular, Vermeer (Zen 3) Architecture is enhanced with 7 nm technology and supports AMD-V, SVM. The product was launched on Q4/2020

AMD Ryzen 5 3600X

The AMD Ryzen 5 3600X operates with 6 cores and 16 CPU threads. It run at 4.40 GHz base 4.20 GHz all cores while the TDP is set at 95 W.The processor is attached to the AM4 CPU socket. This version includes 32.00 MB of L3 cache on one chip, supports 2 memory channels to support DDR4-3200 RAM and features 4.0 PCIe Gen 20 lanes. Tjunction keeps below 95 °C degrees C. In particular, Matisse (Zen 2) Architecture is enhanced with 7 nm technology and supports AMD-V, SVM. The product was launched on Q3/2019

AMD Ryzen 7 5800X

AMD Ryzen 5 3600X

Compare Detail

3.80 GHz	Frequency	3.80 GHz
8	Cores	6
4.70 GHz	Turbo (1 Core)	4.40 GHz
4.40 GHz	Turbo (All Cores)	4.20 GHz
Yes	Hyperthreading	Yes
Yes	Overclocking	Yes
normal	Core Architecture	normal
no iGPU	GPU	no iGPU
No turbo	GPU (Turbo)	No turbo
7 nm	Technology	7 nm
No turbo	GPU (Turbo)	No turbo
	DirectX Version
	Max. displays
DDR4-3200	Memory	DDR4-3200
2	Memory channels	2
	Max memory
Yes	ECC	Yes
4. 00 MB	L2 Cache	—
32.00 MB	L3 Cache	32.00 MB
4.0	PCIe version	4.0
20	PCIe lanes	20
7 nm	Technology	7 nm
AM4	Socket	AM4
105 W	TDP	95 W
AMD-V, SVM	Virtualization	AMD-V, SVM
Q4/2020	Release date	Q3/2019
Show more data		Show more data

Cinebench R23 (Single-Core)

Cinebench R23 is the successor of Cinebench R20 and is also based on the Cinema 4 Suite. Cinema 4 is a worldwide used software to create 3D forms. The single-core test only uses one CPU core, the amount of cores or hyperthreading ability doesn’t count.

Cinebench R23 (Multi-Core)

Cinebench R23 is the successor of Cinebench R20 and is also based on the Cinema 4 Suite. Cinema 4 is a worldwide used software to create 3D forms. The multi-core test involves all CPU cores and taks a big advantage of hyperthreading.

Cinebench R20 (Single-Core)

Cinebench R20 is the successor of Cinebench R15 and is also based on the Cinema 4 Suite. Cinema 4 is a worldwide used software to create 3D forms. The single-core test only uses one CPU core, the amount of cores or hyperthreading ability doesn’t count.

Cinebench R20 (Multi-Core)

Cinebench R20 is the successor of Cinebench R15 and is also based on the Cinema 4 Suite. Cinema 4 is a worldwide used software to create 3D forms. The multi-core test involves all CPU cores and taks a big advantage of hyperthreading.

Cinebench R15 (Single-Core)

Cinebench R15 is the successor of Cinebench 11.5 and is also based on the Cinema 4 Suite. Cinema 4 is a worldwide used software to create 3D forms. The single-core test only uses one CPU core, the amount of cores or hyperthreading ability doesn’t count.

Cinebench R15 (Multi-Core)

Cinebench R15 is the successor of Cinebench 11.5 and is also based on the Cinema 4 Suite. Cinema 4 is a worldwide used software to create 3D forms. The multi-core test involves all CPU cores and taks a big advantage of hyperthreading.

Geekbench 5, 64bit (Single-Core)

Geekbench 5 is a cross plattform benchmark that heavily uses the systems memory. A fast memory will push the result a lot. The single-core test only uses one CPU core, the amount of cores or hyperthreading ability doesn’t count.

Geekbench 5, 64bit (Multi-Core)

Geekbench 5 is a cross plattform benchmark that heavily uses the systems memory. A fast memory will push the result a lot. The multi-core test involves all CPU cores and taks a big advantage of hyperthreading.

Blender 2.81 (bmw27)

Blender is a free 3D graphics software for rendering (creating) 3D bodies, which can also be textured and animated in the software. The Blender benchmark creates predefined scenes and measures the time (s) required for the entire scene. The shorter the time required, the better. We selected bmw27 as the benchmark scene.

Estimated results for PassMark CPU Mark

Some of the CPUs listed below have been benchmarked by CPU-Comparison. However the majority of CPUs have not been tested and the results have been estimated by a CPU-Comparison’s secret proprietary formula. As such they do not accurately reflect the actual Passmark CPU mark values and are not endorsed by PassMark Software Pty Ltd.

Monero Hashrate kH/s

The crypto currency Monero has been using the RandomX algorithm since November 2019. This PoW (proof of work) algorithm can only efficiently be calculated using a processor (CPU) or a graphics card (GPU). The CryptoNight algorithm was used for Monero until November 2019, but it could be calculated using ASICs. RandomX benefits from a high number of CPU cores, cache and a fast connection of the memory via as many memory channels as possible

Electric Usage Estimate

Average hours of use per day

Average CPU Utilization (0-100%)

Power cost, dollar per kWh

Electric Usage Estimate

Average hours of use per day

Average CPU Utilization (0-100%)

Power cost, dollar per kWh

AMD Ryzen 7 5800X		AMD Ryzen 5 3600X
105 W	Max TDP	95 W
NA	Power consumption per day (kWh)	NA
NA	Running cost per day	NA
NA	Power consumption per year (kWh)	NA
NA	Running cost per year	NA

Popular Comparision

Comments

Ryzen 5 3600X vs Ryzen 7 5800X CPU Comparison

Compare AMD Ryzen 5 Desktop 6 core CPU vs AMD Ryzen 7 Desktop 8 core processor, specs and benchmark score. Which is the better CPU for gaming?

CPU Comparison

Quickly search and compare processors

Price

The cheapest price from our partner retailers

$ 244

$ 66.01
slightly less expensive

Overall Score

A combined score of all workloads

92 %

48%
significantly better overall score

Gaming Score

The raw gaming performance with a fast GPU

108 %

30%
slightly better gaming score

Multitasking Score

Performance in workloads using up to 8 cores

87 %

55%
significantly better multitasking score

Heavy Workload Score

Performance in workloads using up to 16 cores

85 %

57%
significantly better heavy workload score

Want to compare your processor against the Ryzen 5 3600X and the Ryzen 7 5800X? Download our free and quick PC Performance Test.

Download CPU Benchmark

Geekbench (multi) score

Multi threaded benchmark

10,319

50%
significantly higher Geekbench (multi) score

Geekbench (single) score

Single threaded benchmark

1,669

34%
slightly higher Geekbench (single) score

3DMark Physics score

Physics simulation benchmark

11,366

59%
significantly higher 3DMark Physics score

Cores

Number of physical processing units

33%
slightly more cores

Threads

Number of logical processing units

33%
slightly more threads

Clock Speed

Base frequency at which the chip operates

4 GHz

Identical

Turbo Speed

Higher frequency used for heavy workloads

5 GHz

6%
slightly higher turbo speed

TDP

Thermal Design Power: Measure of heat generated by the CPU

95 W

10%
slightly lower TDP

Memory Channels

Lanes for simultaneous memory access

Identical

L1 Cache

Smallest embedded memory, but the fastest

384 KiB

512 KiB

33%
slightly more L1 cache

L2 Cache

Secondary embedded memory, slower than L1 cache

4 MiB

33%
slightly more L2 cache

L3 Cache

General memory pool for all cores

32 MiB

Identical

32 MiB

Rank

Ranking in the hardwareDB database

229th
of 1,327

74th
of 1,327

Family

The product line

Ryzen 5 Desktop

Ryzen 7 Desktop

Release date

The official date of release of this chip

July 2019

May 2020

Memory Type

The type of memory used by this chip

DDR4

Is Unlocked

Can this CPU can be overclocked

Yes

Supports ECC memory

Does this CPU support error correcting memory

—

Yes

Ryzen 5 3600X vs Ryzen 7 5800X comparison

According to the hardwareDB Benchmark tool, the Ryzen 7 5800X is faster than the Ryzen 5 3600X. Furthermore, our gaming benchmark shows that it also outperforms the Ryzen 5 3600X in all gaming tests too.

When comparing core counts for these CPUs, we notice that the Ryzen 7 5800X has slightly more cores with 8 cores compared to the Ryzen 5 3600X that has 6 cores. It also has more threads than the Ryzen 5 3600X. Both the Ryzen 5 3600X and Ryzen 7 5800X have the same clock speed. Despite this, the Ryzen 7 5800X has a slightly higher turbo speed. A Ryzen 5 3600X CPU outputs less heat than a Ryzen 7 5800X CPU because of its slightly lower TDP. This measures the amount of heat they output and can be used to estimate power consumption. The info from our database shows that the Ryzen 7 5800X has more L2 cache than the Ryzen 5 3600X. However, both these chips have the same amount of L3 cache.

The more cores a CPU has, the better the overall performance will be in parallel workloads such as multitasking. Many CPUs have more threads than cores, this means that each physical core is split into multiple logical cores, making them more efficient. Indeed, the Ryzen 5 3600X has more threads than cores. Each physical core is split into multiple threads.

Latest posts

How to choose a power supply for your gaming PC: wattage and efficiencyJuly 11, 2022

How to compare GPUs and GPUs: find the best components for gamingJuly 02, 2022

Best PC benchmarking tools 2022: how to performance test your gaming PCJuly 02, 2022

Popular CPU Comparisons

Ryzen 5 5600X vs Ryzen 7 5700G

Core i7-12700 vs Core i7-12700K

Ryzen 7 5800X vs Core i7-11700K

Ryzen 9 5900X vs Core i9-11900K

Core i5-10500T vs Core i5-10505

Core i7-11800H vs Core i7-12700H

Ryzen 5 5500U vs Core i5-10210U

Core i5-11600K vs Core i7-11700K

AMD Ryzen 7 5800X vs Ryzen 5 3600X:

performance comparison

AMD Ryzen 7 5800X

AMD Ryzen 5 3600X

Which is better: 8-core AMD Ryzen 7 5800X at 3. 8 GHz or Ryzen 5 3600X with 6 cores at 3.8 GHz? To find out, read our comparison test of these 7nm desktop processors in popular benchmarks, games and heavy applications.

Overview
Differences
Performance
Features
Comments

Overview

Overview and comparison of the main metrics from NanoReview

Single -flow performance

Rating in tests using one nucleus

Ryzen 7 5800x

Ryzen 5 3600x

Multi -flow performance

Tests in benchmarks where all nucleus

9 9,000 9,000 9,000

Ryzen 7 5800x

Ryzen 5 3600x

Energy Equality

Energy Effect Consum CHIP

Ryzen 7 5800x

Ryzen 5 3600x

9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58 9000 58

Ryzen 7 5800X

Ryzen 5 3600X

Key differences

What are the main differences between 3600X and 5800X

Reasons to choose AMD Ryzen 7 5800X

Launched 1 year and 3 months later than rival
Has 2 more physical cores
23% faster in single core Geekbench v5 — 1626 and 1325 points

higher frequency in Turbo Boost (4. 7 GHz vs 4.4 GHz)

Reasons to choose AMD Ryzen 5 3600X

10% lower than Ryzen 7 5800X peak power consumption — 95 vs 105 Watts

Benchmark tests

Compare the results of processor tests in benchmarks

Cinebench R23 (single core)

Ryzen 7 5800X
+21%

1594

Ryzen 5 3600X

1320

Cinebench R23 (multi-core)

Ryzen 7 5800X
+62%

15296

Ryzen 5 3600X

9437

Passmark CPU (Single Core)

Ryzen 7 5800X
+30%

3475

Ryzen 5 3600X

2665

Passmark CPU (multi-core)

Ryzen 7 5800X
+55%

28188

Ryzen 5 3600X

18241

Geekbench 5 (Single Core)

Ryzen 7 5800X
+22%

1613

Ryzen 5 3600X

1323

Geekbench 5 (multi-core)

Ryzen 7 5800X
+44%

10489

Ryzen 5 3600X

7288

▶️ Add your score to Cinebench R23

Specifications

AMD Ryzen 7 5800X and Ryzen 5 3600X full technical specifications list

General information

Manufacturer	AMD	AMD
Release date	October 8, 2020	July 7, 2019
Type	Desktop	Desktop
Instruction set architecture	x86-64	x86-64
Codename	Vermeer	Zen 2
Socket	AM4	AM4
Integrated graphics	No	No

Capacity

Number of cores	8	6
Number of threads	16	12
Frequency	3. 8 GHz	3.8 GHz
Max. frequency in Turbo Boost	4.7 GHz	4.4 GHz
Bus frequency	100 MHz	100 MHz
Multiplier	38x	38x
Level 1 cache	64KB (per core)	64KB (per core)
Level 2 cache	512KB (per core)	512KB (per core)
Level 3 cache	32MB (shared)	32MB (shared)
Unlocked multiplier	Yes	Yes

Power consumption

Number of transistors	—	3.8 billion
Process	7 nanometers	7 nanometers
Power consumption (TDP)	105 W	95 W
Critical temperature	90°C	95°C

Memory support

Memory type	DDR4-3200	DDR4-3200
Max. size	128 GB	128 GB
Number of channels	2	2
Max. bandwidth	47.68 GB/s	47.68 GB/s
ECC support	Yes	Yes

Other

website

Official site	AMD Ryzen 7 5800X	AMD Ryzen 5 3600X
PCI Express Version	4.0	4.0
Max. PCI Express lanes	20	16

Poll

What processor do you think is the best?

Ryzen 7 5800X

46 (70.8%)

Ryzen 5 3600X

19 (29.2%)

Total votes: 65

Competitors

1.
AMD Ryzen 7 5800X vs Apple M1 Max

2.
AMD Ryzen 7 5800X vs Apple M1 Pro

3.
AMD Ryzen 7 5800X vs AMD Ryzen 7 5700G

4.
AMD Ryzen 7 5800X vs Intel Core i5 12600K

5.
AMD Ryzen 7 5800X vs AMD Ryzen 7 5800

6.
AMD Ryzen 5 3600X vs AMD Ryzen 5 5600X

7.
AMD Ryzen 5 3600X vs Intel Core i5 12400

8.
AMD Ryzen 5 3600X vs AMD Ryzen 7 5700X

9.
AMD Ryzen 5 3600X vs AMD Ryzen 7 5800X3D

What will you choose: AMD Ryzen 5 3600X or Ryzen 7 5800X?

Name

Message

AMD Ryzen 7 5800H vs Ryzen 5 3600X:

performance comparison

AMD Ryzen 7 5800H

AMD Ryzen 5 3600X

Which is better: 8-core laptop AMD Ryzen 7 5800H at 3.2 GHz or desktop Ryzen 5 3600X with 6 cores at 3.8 GHz? To find out, read our comparison test of these 7nm processors in popular benchmarks, games and heavy applications.

Overview
Differences
Performance
Features
Comments

Overview

Overview and comparison of the main metrics from NanoReview

Single-threaded performance

Single-core benchmark rating

Ryzen 7 5800H

Ryzen 5 3600X

9002 Multi-threaded performance 30002 Tests in benchmarks, where all nuclei are involved in all the nuclei

Ryzen 7 5800h

Ryzen 5 3600x

Energy

Energy Effects Chip

5800h

70 9000

Ryzen 5 360029

NanoReview Rating

Final Processor Score

Ryzen 7 5800H

Ryzen 5 3600X

Key differences

What are the main differences between 3600X and 5800H

Reasons to choose AMD Ryzen 7 5800H

Launched 1 year and 6 months later than rival
43% lower than Ryzen 5 3600X peak power consumption — 54 vs 95 Watts
At 20. 59 GB/s higher maximum memory bandwidth
Has 2 more physical cores
Has an integrated Radeon RX Vega 8 graphics accelerator
More than 10° C higher than maximum operating temperature
10% faster in single core Geekbench v5 — 1454 and 1325 points

Reasons to choose AMD Ryzen 5 3600X

Has 16MB more L3 cache
Unlocked multiplier
Supports up to 128GB DDR4-3200
New PCI Express 90 4.0 standard

Benchmark tests

Compare the results of processor tests in benchmarks

Cinebench R23 (single core)

Ryzen 7 5800H
+7%

1410

Ryzen 5 3600X

1320

Cinebench R23 (multi-core)

Ryzen 7 5800H
+29%

12151

Ryzen 5 3600X

9437

Passmark CPU (Single Core)

Ryzen 7 5800H
+15%

3070

Ryzen 5 3600X

2665

Passmark CPU (multi-core)

Ryzen 7 5800H
+17%

21306

Ryzen 5 3600X

18241

Geekbench 5 (single core)

Ryzen 7 5800H
+9%

1444

Ryzen 5 3600X

1323

Geekbench 5 (multi-core)

Ryzen 7 5800H
+1%

7390

Ryzen 5 3600X

7288

▶️ Add your score to Cinebench R23

Specifications

AMD Ryzen 7 5800H and Ryzen 5 3600X full technical specifications list

General information

Manufacturer	AMD	AMD
Release date	January 12, 2021	July 7, 2019
Type	For laptop	Desktop
Instruction set architecture	x86-64	x86-64
Codename	Cezanne	Zen 2
Socket	FP6	AM4
Integrated graphics	Radeon RX Vega 8	No

Performance

Cores	8	6
Number of threads	16	12
Frequency	3. 2 GHz	3.8 GHz
Max. frequency in Turbo Boost	4.4 GHz	4.4 GHz
Bus frequency	100 MHz	100 MHz
Multiplier	32x	38x
Level 1 cache	64KB (per core)	64KB (per core)
Level 2 cache	512KB (per core)	512KB (per core)
Level 3 cache	16MB (shared)	32MB (shared)
Unlocked multiplier	No	Yes

Power consumption

Number of transistors	—	3.8 billion
Process	7 nanometers	7 nanometers
Power consumption (TDP)	35-54W	95W
Critical temperature	105°C	95°C

Integrated graphics	Radeon RX Vega 8	—
Boost GPU frequency	2000 MHz	—
Shader blocks	512	—
TMUs	32	—
ROPs	8	—
TGP	15W	—

Memory support

Memory type	DDR4-3200, LPDDR4-4266	DDR4-3200
Max. size	64 GB	128 GB
Number of channels	2	2
Max. bandwidth	68.27 GB/s	47.68 GB/s
ECC support	Yes	Yes

Other

website

Official site	AMD Ryzen 7 5800H	AMD Ryzen 5 3600X
PCI Express Version	3.0	4.0
Max. PCI Express lanes	16	16

Poll

What processor do you think is the best?

Ryzen 7 5800H

5 (83.3%)

Ryzen 5 3600X

1 (16.7%)

Total votes: 6

Competitors

1.
AMD Ryzen 7 5800H and Ryzen 5 5600H

2.
AMD Ryzen 7 5800H and Intel Core i7 12700H

3.
AMD Ryzen 7 5800H and Apple M1

4.
AMD Ryzen 7 5800H and Ryzen 7 6800H

5.
AMD Ryzen 7 5800H and Intel Core i7 12650H

6.
AMD Ryzen 5 3600X and Ryzen 5 5600X

7.
AMD Ryzen 5 3600X and Intel Core i5 12400

8.
AMD Ryzen 5 3600X and Ryzen 7 5700X

9.
AMD Ryzen 5 3600X and Ryzen 7 5800X3D

What will you choose: AMD Ryzen 5 3600X or Ryzen 7 5800H?

Name

Message

AMD Ryzen 5 3600X vs AMD Ryzen 7 5800X: What is the difference?

62 Ballla

AMD Ryzen 5 3600x

59 Ballla

AMD RYZEN 7 5800X

3600XT

64 Facts compared to

AMD RYZEN 5 3600X

AMD Ryzen 7 50023 Why is AMD Ryzen 5 3600X better than AMD Ryzen 7 5800X?

5°C higher than maximum operating temperature?
95°C vs 90°C
10W below TDP?
95W vs 105W
1. 33MB/core more L3 cache per core?
5.33MB/core vs 4MB/core
Supports memory debug code?
6.96% cheaper?
282.77 € vs 303.94 €

Why is AMD Ryzen 7 5800X better than AMD Ryzen 5 3600X?

33.33% higher CPU speed?
8 x 3.8GHz vs 6 x 3.8GHz
4 more CPU threads?
16 vs 12
1MB more L2 cache?
4MB vs 3MB
1.57x higher PassMark score?
28752 vs 18325
0.3GHz higher turbo clock speed?
4.7GHz vs 4.4GHz
128KB more L1 cache?
512KB vs 384KB
31.18% higher PassMark score (single)?
3513 vs 2678
49.54% higher Geekbench 5 multi-core result?
10330 vs 6908

Which comparisons are the most popular?

AMD Ryzen 5 3600X

AMD Ryzen 5 5600X

AMD Ryzen 7 5800X

Intel Core i7-12700k

AMD Ryzen 5 3600x

AMD Ryzen 5 5500

AMD Ryzen 7 5800x

AMD Ryzen 5 3600x

AMD AMD RYZEN 5 3600

AMD Ryzen 7 5800x

AMD Ryzen 7 5700x

AMD Ryzen 5 3600x

AMD Ryzen 5 5600g

9000 AMD Ryzen 7 5800x 9000 VS 7,000 VS 7000 5000 VS AMD RYZEN 7 5800X 9000 VS AMD

AMD Ryzen 5 3600X

AMD Ryzen 5 5500U

AMD Ryzen 7 5800X

AMD Ryzen 5 5600X

AMD Ryzen 5 3600X

AMD Ryzen 7 2700X

AMD Ryzen 7 5800x

Intel Core i5-12600K

AMD Ryzen 5 3600x

AMD Ryzen 7 3700X

AMD Ryzen 7 5800x

9000 AMD Ryzen 7 3700x

AMD Ryzen 7 3700x

AMD RYZEN 70003

AMD Ryzen 5 4500u

AMD Ryzen 7 5800x

Intel Core i7-11700K

AMD Ryzen 5 3600x

AMD Ry

Intel Core i7-10700k

AMD Ryzen 7 5800x

AMD Ryzen 7 3800x

Comparison0003

2 reviews of users

AMD Ryzen 5 3600x

9. 0 /10

2 Reviews of users

AMD Ryzen 7 5800x

3 Users

AMD Ryzen 7 5800x

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 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 Reviews of users

Functions

Price and quality ratio

10.0 /10

2 Votes

9.7 /10

3 Votes

Games

9.5 /10

2 Votes

9.7 /10

3 Votes

performance

9.0 /10

2 VOTES

9,000 9000

22

9.5 /10

2 Votes

9.7 /10

3 Votes

Energy

9.5 /10

9000 2 Votes

9000 /10 /10

3 votes

Performance

1.CPU speed

6 x 3.8GHz

8 x 3.8GHz

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 multitasking.

3.speed turbo clock

4.4GHz

4.7GHz

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

4. Unlocked

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

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

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

9.core L3

5.33MB/core

4MB/core

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

Memory

1.RAM speed

3200MHz

Can support faster memory which speeds up system performance.

2.max memory bandwidth

47.68GB/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 3600X)

Unknown. Help us offer a price. (AMD Ryzen 7 5800X)

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

7. Supports memory debug code

✔AMD Ryzen 5 3600X

✖AMD Ryzen 7 5800X

Memory debug 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 3600X)

Unknown. Help us offer a price. (AMD Ryzen 7 5800X)

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 3600X)

Unknown. Help us offer a price. (AMD Ryzen 7 5800X)

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

Geotagging

1. PassMark result

This benchmark measures CPU performance using multi-threading.

2. PassMark result (single)

This test measures processor performance using a thread of execution.

3.Geekbench 5 result (multi-core)

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)

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)

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)

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

7. Blender test result (bmw27)

235.9seconds

Unknown. Help us offer a price. (AMD Ryzen 7 5800X)

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)

709seconds

Unknown. Help us offer a price. (AMD Ryzen 7 5800X)

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

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

Features

1.uses multithreading

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

processor 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 3600X

✔AMD Ryzen 7 5800X

AES is used to speed up encryption and decryption.

3. Has AVX

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

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

4.Version SSE

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

5. Has F16C

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

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 3600X)

Unknown. Help us offer a price. (AMD Ryzen 7 5800X)

NEON provides faster media processing such as MP3 listening.

7. Has MMX

✔AMD Ryzen 5 3600X

✔AMD Ryzen 7 5800X

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

8.Has TrustZone

✖AMD Ryzen 5 3600X

✖AMD Ryzen 7 5800X

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 3600X)

Unknown. Help us offer a price. (AMD Ryzen 7 5800X)

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

Price comparison

Cancel

Which CPUs are better?

Testing AMD Ryzen 7 5800X and Ryzen 9 5950X vs. Ryzen 5 5600X, Ryzen 9 5900X and other current AMD and Intel 9 processors0001
Testing AMD Ryzen 5 5600X and Ryzen 9 5900X processors: the new Zen3 microarchitecture and the same AM4

platform We recently got acquainted with the first pair of AMD processors built on the basis of the new Zen3 microarchitecture, and now two more models have reached us. More interesting — the Ryzen 7 5800X and Ryzen 9 5950X use «full» chiplets with eight working cores, and not six, as in the previously reviewed 5600X and 5900X. Considering that one of the significant changes in the new generation is the transition from a quad-core CCX (the minimal «brick» on which processors are built) to an octa-core one, it is the 5800X that is the best example of why all this was started — and the best AMD monolithic processor. Conditionally monolithic — the chiplet composition is not used only in APUs (i.e., processors with integrated graphics), so in this respect the 5000 line is identical to the 3000. Moreover, the I / O chip (IOD) has remained generally the same, so peripheral capabilities, and the memory controller remain old. But in practice, this did not interfere too much, but the architecture of the processor chiplet itself can now be simplified and accelerated. Eight processor cores in it work with a single array of third-level cache memory, and not divided, as before. In six-core models — also, but six is not eight. And in general, often the disposal of marriage (especially in the early stages), so that Ryzen 7 can objectively surpass Ryzen 5 not only quantitatively, but also qualitatively. And in Ryzen 9There are two processor chipsets as before, but the 5950X is «just» the company’s top model — and the fastest processor for the AM4 platform in combination. The Ryzen 9 5900X is also good, of course, and cheaper — but the is not as perfect as the . Including at the production level — similar to the relationship in the younger families.

Ideally, we would prefer to test all four at once — but the peculiarities of logistics «forced» the first test to be divided into two parts. Therefore, today we will simply bring all the models together. This will end the novelties for a while (no new models are expected in the coming months), so it will be possible to study this family widen and deepen . Including, and by making extended game testing — the need for which is long overdue (and overripe). In the meantime, a fairly short and simple material that will allow you to evaluate the entire 5000 line in comparison with its predecessors and some competitors.

Test participants

	AMD Ryzen 5 5600X	AMD Ryzen 7 5800X	AMD Ryzen 9 5900X	AMD Ryzen 9 5950X
Kernel name	Vermeer	Vermeer	Vermeer	Vermeer
Production technology	7/12 nm	7/12 nm	7/12 nm	7/12 nm
Core frequency, GHz	3. 7/4.6	3.8/4.7	3.7/4.8	3.4/4.9
Number of cores/threads	6/12	8/16	12/24	16/32
L1 cache (total), I/D, KB	192/192	256/256	384/384	512/512
L2 cache, KB	6×512	8×512	12×512	16×512
L3 cache, MiB	32	32	64	64
RAM	2×DDR4-3200	2×DDR4-3200	2×DDR4-3200	2×DDR4-3200
TDP, W	65	105	105	105
Number of PCIe 4.0 lanes	20	20	20	20
Integrated GPU	no	no	no	no

Now we have all four processors based on the Zen3 microarchitecture. Unfortunately, there are really only four of them, with suggested retail prices starting at $299. Zen2-based processors debuted prettier — there were six of them, and prices started at $199. Perhaps this is the main drawback of the new family — more disturbing, of course, not to us, but to the most impatient customers. Therefore, first of all, this is still an addition to the AMD range, and not a replacement for old models that continue to remain relevant. Especially the younger models of those — for which there are no replacements yet and are not expected in the near future.

	AMD Ryzen 5 3600X	AMD Ryzen 7 3800X	AMD Ryzen 9 3900X	AMD Ryzen 9 3950X
Kernel name	Matisse	Matisse	Matisse	Matisse
Production technology	7/12 nm	7/12 nm	7/12 nm	7/12 nm
Core frequency, GHz	3.8/4.4	3.9/4.5	3.8/4.6	3.5/4.7
Number of cores/threads	6/12	8/16	12/24	16/32
L1 cache (total), I/D, KB	192/192	256/256	384/384	512/512
L2 cache, KB	6×512	8×512	12×512	16×512
L3 cache, MiB	32	32	64	64
RAM	2×DDR4-3200	2×DDR4-3200	2×DDR4-3200	2×DDR4-3200
TDP, W	95	105	105	105
Number of PCIe 4. 0 lanes	20	20	20	20
Integrated GPU	no	no	no	no

Ryzen 5 3600 and Ryzen 7 3700X participated in the first article. Some did not like this choice, so today there will be an alternative Ryzen 5 3600X and Ryzen 7 3800X. In fact, the difference between the 3600/3600X and 3700X/3800X pairs is very small — both in terms of performance and power consumption. The existing legend about the greater voracity of the 3800X is a legend: in practice, under the same favorable conditions, the 3700X operates at comparable frequencies, consumes a little less energy and releases a little less heat. It’s just that the 3600 and 3700X have a TDP of 65 W, i.e. the company officially allows you to cool them worse while maintaining warranty obligations — but the 3600X and 3800X are not required to work in such conditions. But they can. When power and cooling capabilities are sufficient and are redundant, all AMD processors are «creatively» subject to formal restrictions. It’s time to get used to it, since the joke about “105 W” (and a maximum of 142) has acquired a fair beard back in the days of the existence of the 2000 line based on Zen +. But this «creativity» has long been characteristic of both AMD and Intel, so it’s just not worth trying to extract more from the specifications than what they actually contain.

C Ryzen 9 and there are no special options at all — just 3900X and 3950X once again. Moreover, the second one did not even have a formal update in the XT line this year.

	AMD Ryzen 5 2600	AMD Ryzen 7 2700X
Kernel name	Pinnacle Ridge	Pinnacle Ridge
Production technology	12 nm	12 nm
Core frequency, GHz	3.4/3.9	3.7/4.3
Number of cores/threads	6/12	8/16
L1 cache (total), I/D, KB	384/192	512/256
L2 cache, KB	6×512	8×512
L3 cache, MiB	16	16
RAM	2×DDR4-2933	2×DDR4-2933
TDP, W	65	105
Number of PCIe 3. 0 lanes	20	20
Integrated GPU	no	no

Also , at the request of workers , we added a couple of «old» Ryzens to the list — Ryzen 5 2600 and Ryzen 7 2700X. It is clear that none of the main participants in testing and the best of them is not a competitor in terms of performance, but sometimes it is useful to evaluate the latter on a long interval of . Just a year and a half ago, the Ryzen 7 2700X was still the fastest processor under AM4 and one of the best in its price niche. Modern octa-cores are more expensive (although… not more expensive than the Ryzen 7 1800X at the time of launch), but they also work faster. And how much faster — you can compare directly. The Ryzen 5 2600 has recently become something of a «people’s minimum»: it’s already six cores, but still is very cheap . Much cheaper than the 3600X and generally not very comparable to the 5600X — but also is not fast . But you also need to understand why pay more — and for this you can just take and compare performance. Thankfully it’s possible.

	Intel Core i5-10600K	Intel Core i7-10700K	Intel Core i9-10900K
Kernel name	Comet Lake	Comet Lake	Comet Lake
Production technology	14 nm	14 nm	14 nm
Core frequency, GHz	4.1/4.8	3.8/5.1	3.7/5.3
Number of cores/threads	6/12	8/16	10/20
L1 cache (total), I/D, KB	192/192	256/256	320/320
L2 cache, KB	6×256	8×256	10×256
L3 cache, MiB	12	16	20
RAM	2×DDR4-2933	2×DDR4-2933	2×DDR4-2933
TDP, W	125	125	125
Number of PCIe 3. 0 lanes	16	16	16
Integrated GPU	UHD Graphics 630	UHD Graphics 630	UHD Graphics 630

As for Intel processors, we simply and voluntarily took the top three models in the Core i5, i7 and i9 lines. In fact, any would do — since they are needed today only as guidelines. At first we even thought about taking the «under-top» Core i9-10850K, however, after looking at the results of the Ryzen 7 5800X, we left 109 in place00K. In general, for lovers of the notorious philosophical combination «price / performance» (and in a degenerate form — when the price of the processor and the performance of the system on it are taken), models for LGA1200, as well as LGA1151, are of no interest. But they are still sold in large quantities, because there are other criteria — both objective and even more subjective. Therefore, one cannot do without comparison at all — but it is pointless to try to somehow drive its participants into some kind of framework. So let it be the way it is.

Other environments traditional: AMD Radeon Vega 56 graphics card, SATA SSD and 16 GB DDR4 memory. The memory clock speed is the maximum according to the processor specification. Intel Multi-Core Enhance and AMD Precision Boost Overdrive technologies are disabled — for the latter, this is typical by default, but many motherboards strive to turn on the former quietly. Here, along with the memory frequency, they already affect performance, and their use is made more specific by the requirements for the board and chipset, but in normal mode there are no problems. Yes, and in itself, the inclusion of MCE, but without overclocking, increases the performance of the Core i9-10900K by only 3% with an increase in power consumption by 5% — as we have already seen. Therefore, in our opinion, such technologies still have no practical meaning. Another thing is manual overclocking, but here everything is individual. And it depends on both technique and personal luck.

Test Method

Methodology for testing computer systems of the 2020 sample

The testing methodology is described in detail in a separate article, and the results of all tests are available in a separate table in Microsoft Excel format. Directly in the articles, we use the processed results: normalized with respect to the reference system (Intel Core i5-9600K with 16 GB of memory, AMD Radeon Vega 56 graphics card and SATA SSD) and grouped by computer application. Accordingly, on all diagrams related to applications, dimensionless points, so here “more is better” everywhere. And starting from this year, we are finally transferring game tests to an optional status (the reasons for which are discussed in detail in the description of the test methodology), so that only specialized materials will be available for them. In the main lineup there are only a couple of «processor-dependent» games in low resolution and medium quality — synthetic, of course, but conditions close to reality for testing processors are not suitable, since nothing depends on them in such conditions.

iXBT Application Benchmark 2020

The introduction of Zen2 last year was a more significant update to AM4 than the advent of Zen3 — as it allowed for a “repricing” of families: new models started at the level where the old ones ended: Ryzen 5 no slower than Ryzen 7 previous series. Now it’s still slower. But a significant increase in performance is also visible. And first of all, it is important not for intra-company competition (AMD still cannot completely replace Zen2 with Zen3, so you have to maneuver prices), but for cross-platform competition — the new Ryzen 7 5800X is so much faster than the previous eight-core models from AMD and Intel that we can already say and about competition with ten-core processors from the second company. Moreover, Intel has the «previous» generations of the microarchitecture still concurrently and the last — a serious update of the range of processors for LGA1200 will occur only next year. And hovering over it all Ryzen 9- alternatives to which can only be found in the HEDT camp, and not on mass platforms. And it’s finally clear why the Ryzen 9 3950XT didn’t appear in the summer: its life would be too short.

The layout is the same. With one exception — here the number of cores has a slightly larger relative value, so last year’s Ryzen 5 has not yet caught up with the best the year before last Ryzen 7. And now the problem has been solved completely and with a margin. The architecture of Intel processors is equivalent to last year’s . More precisely, on the contrary — Core with such cores (albeit in smaller quantities) appeared back in 2015, and it took AMD «extra» four years to make no worse than . But the company did not stop there — and did better. We will see the answer from Intel only in half a year — and only then it will be possible to evaluate its «persuasiveness». And now — like this.

The last bastion has fallen — here AMD processors still needed a head start in the amount of hardware resources, and other things being equal, Core remained faster. Now it’s gone. Moreover, microarchitectural changes turned out to be very significant in terms of improving scalability from Ryzen 5 to Ryzen 9. But inside the last line, everything is still not going smoothly with her. Although this is understandable: when we hang two chiplets on the same memory channels, etc., then it is no longer possible to gain a lot by simply increasing the number of cores in them in some tasks. But if we evaluate the result globally, then it is more significant that earlier there was no point in buying Ryzen 9 for these tasks, since the performance increase was not adequate to the price increase, and now … Now, perhaps, you can seriously think about Ryzen 9 5900X.

As has been noted more than once, for modern multi-core processors this is purely a task for the microarchitecture — they all have “too” a lot of cores for this to continue to matter. At the same time, two years ago, Ryzen was still significantly behind Core, a year ago they overtook, and now they simply improved their positions. As a result, the Ryzen 5 5600X became the ideal processor for working with graphics — it is faster than all earlier models, and inferior to its counterparts insignificantly; is much cheaper. On the other hand, scalability in the lineup has grown a little here as well. But this, of course, still does not concern the older pair: there are no differences in it except for the number of cores.

On the other hand, here the quantity is trying to «steer» the quality, but it also cannot always. Here’s what such loads highlight well: Zen2 is the equivalent of Core in terms of efficiency, and Zen3 is a step up from both. AMD already made it. Intel in the desktop segment is just getting ready. It has been warming up for five years already — but mayo mayo mayo .

What you should pay attention to — Ryzen 7 «accelerated» more than Ryzen 5. Exactly with a priori assumptions: for eight-core processors, the effect of a monolithic design (albeit not a processor as a whole, but a chiplet) is more pronounced. But in general, the main bottlenecks (from the point of view of such scenarios) have been preserved. First of all — the same memory controller, and not integrated. L3 volume jumps when moving to Ryzen 9- also bounces and speed. However, «unsuccessful» (in some of the perspectives — do not forget about it) technical solutions can be compensated for by brute force. In Zen / Zen +, the memory controller was integrated — so what? In archivers, these processors were radically behind Core. Zen2 — almost did not lag behind. And Zen3 is just faster. So, if something seems “wrong”, you just need to remember that the creation of any real devices is always a road of compromise. Could you just upgrade a monolithic eight-core? Yes. That’s just 16 cores from it in a «normal» way. Just a simple doubling — what happened in the first Ryzen Threadripper with all their troubles. And the chiplet layout as a whole turned out to be much more efficient — both in desktop solutions and in HEDT. What compensates for some of its shortcomings is the final result.

The «honest» eight-core looks very interesting — which here is already simply faster than the Core i9 and not too slower than the old Ryzen 9. However, a little lower we will see one of the explanations for this fact — but it is indicative in itself. But in general, these programs are the best argument for the 5000 series. And given their “multidirectional” optimization, this is a serious argument.

Because, on average, everything is very similar. And the effect of the spring update of the Intel range is practically leveled. With titanic efforts, the company «returned» the ability for the Core i7 to compete with Ryzen 7, not Ryzen 5 — and AMD came in and did everything as it was. Almost as it was — after all, the recommended prices have grown. And there are not enough processors at all — so real ones can behave unpredictably. But from a technical point of view, that’s it. It turns out that it is possible. Since this is not the first time — and the second time it is not necessary to “throw cores”. Intel holds blows, generally speaking, not bad — but it is necessary to respond fully sometime, and not just by lowering prices.

Power Consumption and Energy Efficiency

The power consumption of almost all members of the new line has been reduced, except for the Ryzen 7 5800X. On the other hand, there is nothing new here — almost the same as the Core i7-10700K or … Ryzen 7 2700X. And the reasons are probably the same as in the last case: the 2700X needed to overtake the Core i7-8700K and return the nominal leadership among desktop processors, so they didn’t stand on ceremony with it. And the 5800X was tasked to reach the level of ten-core Core in terms of performance, but “eat” no more than eight-core ones. In fact, he dealt with it. Why was everything wrong in the previous family? And it wasn’t required — Intel had only eight-core processors, and quite voracious, but already slower than Ryzen 7. In the spring, the new Core i7 managed to overtake the old Core i9, and to keep up with from Ryzen 7 — so such a small tuning was needed. Moreover, within the acceptable range — there were models with the same power consumption earlier in this family, and the appetites of Ryzen 9 generally had to be limited. Although after the release of the LGA1200 they stopped embarrassing anyone, but since such an opportunity has appeared, it would be foolish not to use it.

The energy efficiency of the Ryzen 7 5800X suffered somewhat, of course — but only in the sense that it is no better than the previous generation. And Intel processors are still significantly superior. In terms of performance, too, so the company left itself more freedom of maneuver: the potential Ryzen 7 5700X may turn out to be a much more economical solution in practice. Everything will depend on the operating clock frequencies — which have not yet been determined. And, very likely, they will be selected with an eye to the preliminary successes of the next generation of Intel processors. In any case, the power management schemes in modern processors are so debugged that devices of various purposes can be obtained on the same base with easy adjustment. It will be necessary — and at 45 W the eight-core will be «crammed» 🙂

Games

As already mentioned in the description of the methodology, it makes no sense to keep the «classic approach» to testing gaming performance — since video cards have long been determining not only it, but also significantly affect the cost of the system, you need to «dance» exclusively from them. And from the games themselves — too: in modern conditions, fixing a game set for a long time does not make sense, since literally everything can change with the next update. But we will carry out a brief test in (albeit) relatively synthetic conditions — using a couple of games in the «processor-dependent» mode.

However, there is already a feeling that for modern processors, even with the video card used, they cease to be such. But, as mentioned above, this issue will be investigated separately — and in detail. Moreover, speaking of the increased “gaming performance”, the company itself honestly gives examples of games where Ryzen is still lagging behind Core. This means that we need a broad test “base”, etc., since questions arise. Although from a practical point of view — they are not so interesting. And this was well shown by adding a pair of “old” Ryzen to the number of subjects. There was a difference there — and it was easy to get a noticeable difference in the results. In fact, you didn’t even have to try. But catching fleas in the form of FPS units, and only in synthetic (to be honest) initially conditions … A kind of sport. But it will have to be dealt with in the near future, of course.

Total

Four years ago (when the release of processors based on the Zen microarchitecture was once again postponed), many were sure that AMD would not rise from the ashes. The company’s products continued to be used in budget computers and laptops, occupying a little more than 10% of the market (and this is more than, for example, all «alternative» manufacturers of x86 processors 25 years ago — although there were much more of them then), but nothing interesting She could not offer a demanding buyer. In 2017, she was able to, although the assessments of the “offer” turned out to be very different. Optimists paid attention to cheaper multi-core — for many years mass processors were limited to four cores, and then suddenly eight for the same money or six — is very cheap . Pessimists rightly blamed the companies for the fact that the quantitative superiority was not accompanied by the corresponding quality — the company’s processors were a significant step forward compared to its earlier products, but they no longer reached the Core sample of at least 2015. However, as it turned out, the Zen’s upgradability margin turned out to be no worse than that of the Core. So in the process of introducing Zen2, the latter managed to catch up with almost all points, and the transition to Zen3 allowed them to overtake. It is clear that a significant part of AMD’s success is due to Intel’s failures in mastering new technical processes, which also slow down the introduction of new microarchitectures, so the company, in a strategic sense, has been treading water for a long time since that very 2015, carrying out only tactical measures. However … And what does this matter to the buyer? He chooses not some futures, but ready-made products that are on store shelves. And, if we understand processors as such, then we can state a simple fact — the market leader has completely changed. Intel has nothing like the new Ryzen processors of the 5000 family — now this company will have to provide its devices with a head start in the number of cores (under equal conditions, AMD lagged behind two or three years ago — but everything has changed since last year) or look for opportunities to reduce prices. Although the last question is not easy for AMD either: it is not enough to come up with new processors — they still need to be produced in the right quantities, sufficient for everyone. And with this, everything is not easy — the deficit is unlikely to be completely overcome in the near future, so the real retail prices are noticeably higher than the recommended ones. But the latter, it should be noted, not everyone liked it — and when the same Ryzen 7 5800X is offered more expensive than Ryzen 93900X or Core i9-10900KF, its brilliance is noticeably dimmed. Moreover, it (suddenly) turned out that the decision to “pull” AM4 as a universal solution for several years has its drawbacks, and attempts to restore order sometimes only increase entropy.

But these shortcomings (some of which are obviously temporary) do not negate the fact that today, from a technical point of view, Zen3 is the best microarchitecture of x86 processors on the market. In any case, if we talk about desktop processors. Who would like to see is bigger and cheaper — Zen2 debuted with six models starting at $199, and so far only four processors and starting at $299. For many buyers, this makes the Ryzen update only theoretical — in practice, they still continue to choose between the ancient Ryzen 7 2700 and the already “obsolete” Ryzen 5 3600. Although the processors themselves are not to blame. They are, perhaps, simply the best on the market today. And tomorrow will be tomorrow.

ClockTuner for Ryzen 2.1 RC5 by 1usmus. We are studying the second version of CTR for automatic overclocking of AMD Ryzen.

— i2HARD

Articles

Anem

April 17, 2021

This time let’s take a look at the new version of 1usmus software for Ryzen — ClockTuner for Ryzen. At the moment, the program is developing, since version 1.0 it has received several updates, including a rather large-scale update — 2.0 and the current version — 2.1. 1usmus doesn’t plan to stop there.

The author posted a roadmap for future updates so that you and I know what to expect in the future.

This article will be updated, so stay tuned. The main innovations will be described a little further, before the main block of the program description.

The program does not work as well for everyone as we would like, there are reasons for this, and before proceeding to the description of the program, we will consider the main ones.

First of all, consider the necessary requirements for ClockTuner for Ryzen (CTR):

Zen 3: Ryzen 9 5950X, Ryzen 95900X, Ryzen 7 5800X, Ryzen 5 5600X.
Zen 2: Threadripper 3970x, Threadripper 3960x, Ryzen 9 3950x, Ryzen 9 3900X, Ryzen 9 3900XT, RYZEN 9 3900, Ryzen 7 3800XT, RYZEN 7 3800X, RYZEN 7 3700X, RYZEN 5 3600XEN 5, Ryzen 5 3600XT, RYENE 5 Ryzen 5 3500X, Ryzen 5 3500, Ryzen 3 3300X, Ryzen 3 3100.
APU: Ryzen 7 PRO 4750G , Ryzen 7 PRO 4650G, Ryzen 3 PRO 4350G.
Motherboard must be UEFI version with AGESA Combo AM4 1.0.0.4 or newer for Zen 2.
Motherboard must be UEFI version with AGESA Combo v2 AM4 AGESA 1.1.0.0D or newer for Zen 3.
Motherboard must be UEFI version with AGESA Combo v2 AM4 1.2.0.0 or later for APU.
BIOS settings (some settings may not be available):
- Performance Enhancer—Disabled.
- CPU Virtualization — any value.
- CPPC-Enabled.
- CPPC Preferred Cores — Enabled.
- Global C-State — Enabled.
Leave CPU Voltage, Multiplayer CPU multiplier set to Auto. The RAM should work either on the XMP profile, or stable on overclocking, which you have already tested several times. All these settings are necessary to avoid problems during operation, to accurately determine frequency and voltage thresholds.
Operating system Windows 10 x64, preferably with all updates.
.NET Framework 4.7.2 or higher.
If all the points above are met, then download the ClockTuner for Ryzen

software itself

After unpacking the program, go to the cb20 folder and unpack Cinebench R20 there. After that, launch the unpacked Cinebench, accept the license agreement and close Cinebench R20. This benchmark is needed to compare performance before and after.

General information

In general information, I would like to dwell on motherboards, namely their VRM. I would like to start with recommendations from 1usmus on setting LLC (Load-Line Calibration). For now, he says that the LLC setting can be left at Auto on any motherboard, but I would still stick to the original advice, namely:

ASUS — LLC 3 or LLC 4;
MSI — LLC 3;
Gigabyte — Turbo, but can be left in Auto;
ASRock — Auto or LLC 2, but with some caveats;
Biostar — Level 4+.

Your LLC will determine the frequencies you end up with. So, it turns out we set the maximum level of LLC, but why then are there numbers in the recommendation above that correspond to the average level of LLC? If everything were so simple, then we would not stop at the VRM of motherboards. Let’s start with what an LLC is.

Load-Line Calibration (LLC) is a mechanism designed to compensate for large voltage drops (in our case, on the CPU) at the time of increased load. The load can increase gradually or instantly, depending on the task. The mechanism tries to compensate for the sudden voltage drop (Vdroop) by supplying additional voltage, thereby acting in advance. This mechanism was introduced to smoothly apply voltage during states, in our case CPU, load and idle. Thus, we eliminate the instability of the system during overclocking. When the processor is not overclocked, the LLC level is usually set to Auto, because voltage drops for normal modes are a common occurrence when the CPU is running.

Let’s look at the processor in normal mode.

Iout — output current, Vout — output voltage, Vmax — maximum voltage, Vmin — minimum voltage.

As you can see, with an increase in the current supplied to the processor in the LLC case, the voltage Vout drops until the current decreases, after which the voltage returns to its previous level, close to the maximum level before the load appeared. In the case without LLC, we would most likely see that the standard voltage would be equal to half the difference between the maximum and minimum voltage and after each load increase, the voltage would drop back to the original level, and in the case of a load drop, we would see a significant increase in voltage to return to the previous level.

What is the problem then?

The problem lies in the states under the numbers 1 and 2, which appear when the load is increased and decreased, respectively. In the case under the number 1 — affects the stability of the system under load, under the number 2 — affects the operation of transistors in your processor. Why is this happening? Any VRM has a controller that monitors the load and tries to compensate for the voltage drop by applying even more voltage to achieve the required level under load. However, when the processor returns to idle mode, the controller cannot predict the future and thus a power surge occurs at number 2. This surge is very often called an overshoot (overshoot). For example, the specifications from Intel indicate the maximum value of the excess and the time that it can last. The voltage drop in our case serves to prevent an excessive voltage surge, which can be harmful to the processor. The easiest way to solve the problem is to install the controller with a high frequency of operation, thus the controller can compensate for any kind of power surges faster. Unfortunately, this also affects the strapping of the VRM zone, and ultimately the overall cost of the motherboard, so manufacturers often have to come to compromises. If we talk about LLC levels, then the picture below schematically shows the change in voltage depending on the LLC level.

As you can see, as the LLC level increases, the voltage drop becomes lower, but the overvoltage after the load drop becomes, on the contrary, higher. Of course, this may not be observed in all motherboards, but it is rather an exception that you will not have this, because some expensive motherboards can also have this behavior. You can say: «I don’t see anything like that in the program monitoring graphs» — and you will be right, you won’t notice it programmatically. Voltage drops and surges can only be determined with an oscilloscope, the picture below used a Siglent SDS1104X-E.

If you draw a parallel between software monitoring and what the oscilloscope shows, then you can see the following (see the image below), where the graph that is displayed in the monitoring programs is shown in red. Everything is beautiful and beautiful, isn’t it?

Of course, overvoltage is a problem not only when the processor is out of load, but also when the processor enters the load, as you can see below (Undershoot).

Naturally, the names of LLC on different boards may differ, but if you see the name LLC — Extreme, then it is better to refrain from setting it, or set it only for the duration of the tests. We are more focused on overclocking for constant use.

Changes in CTR

Most of the changes in CTR compared to 1.1 beta 7 are described in the roadmap, besides this, we will touch on what the author notes and the latest changes in the software. Further you can meet RC5 and RC4. It is strongly recommended to use the minimum version 2.0 of RC4 due to bugs found in RC3.

In the article itself, you will find images of RC4 and RC5. CTR received a new RC5 update, which did not affect the main elements of the program.

Interface

The layout of some elements has changed in the new interface, and a dark theme has become the main one, which should help when working with the program for color blind people.

CTR HYBRID OC (example: AMD Ryzen 5 5600x)

In this mode, there is no need to choose between standard boost and manual overclocking. HYBRID OC is a combination between two user profiles and a bus.

The P1 profile is designed for maximum load on all cores, while the P2 profile is designed for maximum performance for the best CCX (or several CCXs). In other words, the P2 profile is a gaming profile that allows you to achieve higher frequencies for several cores than a standard single-core boost. Both profiles can be flexibly configured, including setting the transition time between profiles.

Activation of profiles depends on the load on the CCX or CPU. This means that this control method does not include all the shortcomings of technologies based on the loading of TDC or EDC parameters.

Benefits of HYBRID OC:

Best performance for various work scenarios.
CB20 results:
12 streams 4289 / 4616 +8% growth
6 streams 2842 / 3144 + 10% increase
1 thread 599 / 597 unchanged
Easy to use, CTR controls all parameters in a minimized state
Special approach to profile personalization
Activation of profiles in less than 35 ms
Energy efficiency. Standard TDP for P1 and P2 profiles.
Enabled energy saving features for all profiles
Only the best cores or CCX are used for the P2 profile. Managed by the operating system.
Zen 2 and Zen 3 support

Summing up the changes, it is worth noting that the performance increases by an average of 7.5 percent with the same power consumption for all cores. Naturally, the result is an average and it depends on the processor instance, cooling system and motherboard.

Changes in version 2.0 RC4 and RC5

Driver updated. Full compatibility with all anti-cheats. Full compatibility with Ryzen Master.
Profile switching response speed increased by 5 times. Now the time of one cycle is 50 ms.
New design HYBRID OC. More stable and smooth profile switching.
Automatic Cinebench R20 license verification (no need to verify license before running CTR).
Improved CTR security and interaction with the processor.
Improved protection against dangerous user actions (attempts to influence the CTR process).
Improved support for unique dual CCD processors (5600X, 5800X, 3600, 3600X).
Updated PROFILE MANAGEMENT. Multifunction buttons and simpler interface (PROFILES tab).
CPU rankings (Gold, Silver, etc.) are now protected from «Cheater LLCs».
New LLC tips.
Added tips for incorrect data entry in profiles.
Profile status. Now in the PROFILES and TUNER tabs you can see which profile is currently active.
Fixed bug with high VID or temperature display.
Fixed random program crash when using HYBRID OC.
Fixed other minor bugs.
Lower frequency references for Zen3 (additional security when pressing the TUNE button)
* The number of attempts to connect to the SMU has been increased from 5 to 10
* Testing Mode — AVX Light only (fixed a bug where it was possible to unlock the choice of test modes)

Changes in version 2.

1 RC5 (information from the author)

PX profiles for all processors. Get the best performance in low-threaded applications and games!
Fast and accurate creation of P1, P2 and PX profiles during diagnostics (Zen 3 and Renoir only). The whole process takes about 7 minutes.
Redesigned CTR HYBRID OC. Smooth and safe profile switching. The speed of switching profiles reaches 62 times per second (4 times faster than in previous versions).
Preload profiles. Increases the speed of CTR response to load changes.
Efficient CTR HYBRID OC throttling system allows your processor to efficiently handle heavy load (base frequency at 1.1V).
Automatic recommendation of the lowest possible stresses for each profile. Allows you to achieve maximum energy efficiency without compromising performance. In most cases, profiles are able to outperform PBO.
Improved idle state.
Read CPPC information via the Advanced Configuration and Power Interface (ACPI).
Improved security of CTR communication with the system control unit (SMU).
CTR BOOST TESTER — Whether you’re using CTR or Curve Optimizer (CO), you can compare the results of each booster with extended information (temperature, PPT, etc.).
CLEAR CONFIG & CLOSE — button that allows you to reset all CTR settings to default values.
Real-time profile switching statistics (PROFILES tab).
Automatic recovery of profile(s) after sleep.
CALCULATE button for PX, P2 and P1. If you need to recalculate a profile, you don’t need to restart the entire diagnostic process (Zen 3 and Renoir).
Windows Hardware Error Architecture (WHEA) check for tuning process.
Improved protection against dangerous and unintentional user actions.
Automatic recovery of configuration files after a crash.
Improved Phoenix mode (fixed «PLATINUM sample», etc.).
Independent profile system. If one profile is broken, the other profiles are safe.
Improved support for the unique dual CCD processors Ryzen 5 5600X and Ryzen 7 5800X.
Fully supported by AGESA 1.2.0.1 and later.
TO TRAY is a separate button. The user can choose how to minimize the CTR.
Improved logging for all modes and functions.
Processor quality categories (gold, silver, etc.) have been rebalanced.
CTR automatically adjusts the «Holding time». No longer available for manual control.
Increased CTR startup speed (up to 4 times).
CTR has driver priority. Allows it to work more efficiently when the load on the system is at its maximum.
Reduced CPU usage.
Fix Alt-Tab when CTR is in system tray.
Easier way to activate profiles (no save required).
Numerous minor bug fixes and improvements.

What is the CTR software and how does it work

ClockTuner for Ryzen is a CPU overclocker. On this, of course, I would like to end the section, but we will continue. Yes, the software automatically overclocks your processor on the Zen 2 and Zen 3 architecture. Of course, you can set the processor to work at higher frequencies and voltages, or vice versa at lower frequencies and voltages, or if you are lucky at higher frequencies and low voltages , But more on that later. Since we are talking about any kind of overclocking, then any actions with frequencies and voltages can damage the motherboard or processor. Cases are different, to my regret, no one is immune from this.

CTR currently contains third-party modules:

Ryzen Master SDK is a monitoring module, the main reason that you need to have Ryzen Master software installed on your PC.
LibreHardwareMonitorLib — monitoring cpu svi2 and soc svi2.

Maxon’s

Cinebench R20 is a benchmark for comparing your performance, the installation of which is mentioned at the beginning of the article.
Prime95 by George Woltman is a comprehensive CPU stress test.
Reverse engineering version of Shamino’s CCX Work Tool to access SMU.

How does the software work? The program evaluates the quality of each CCX separately from the evaluation that was made at the factory and sets the required frequencies.

Prime95, which is the basis of the program, with the help of special settings allows you to detect the instability of each CCX during overclocking. The program uses a step-by-step algorithm with many rules, during which the frequencies are selected for the voltage you set. Unfortunately, not everything always goes smoothly, as we would like, we will talk about this below.

Existing issues

Yes, there are a few issues that are interfering with the program at this stage. Is the author to blame in this case? I think not. It is rather difficult to take into account all the nuances at the stage of program development, and we will move on to considering two main points within the framework of one problem.

VRM Vdroop

And again we are back to VRM, just above there was some preparation for this section. If you remember, at the beginning of the article, when I mentioned LLC levels in Asrock, I indicated that there are some caveats. According to the company’s technical support, their VRM is configured according to the Intel specification, and, accordingly, most likely has extremely low overvoltage values. Unfortunately, I have not been able to verify this. What do we have in practice? In the case of using processors up to 3600x, you will probably not see any problems, because. The VRM will provide a minimum undervoltage and overvoltage value. If we talk about 3900x and 3950x, the situation here is somewhat different, in my case with the x570 Taichi and, judging by the reviews, on other boards of the same segment on Asrock. Level 1 LLC is optimal in terms of voltage drop, although this is the maximum LLC value. By the way, it is worth clarifying that ClockTuner for Ryzen displays the current voltage drop as a percentage and, according to information from 1usmus, the best option would be a drop from 1 to 2. 5%. In the case of 3900x and x570 Taichi on LLC 1, the drop is 2-2.5% according to the information recorded by the monitoring, but judging by the periodic behavior of the motherboard during program operation, the drop is more. Of course, it is impossible to say unequivocally that the problem is only in the motherboard, but the fact that worse phenomena are observed on such motherboards already makes you think. Below you can see the graphs that 1usmus has provided to understand several situations that you may have while running the program, namely reboots or BSODs.

This is how the processor voltage behavior looks like in two states of idle and load. As you can see, there is a certain virtual “error detection” line and a reboot line, or BSOD. In the case of a weak VRM, we do not have time to identify errors during the operation of the program and the PC immediately goes into reboot. Below you can see the state of the lines when the processor temperature is at the borderline. The lines get even closer and the quality of VRM becomes even more critical.

In the case of a higher quality VRM, the situation looks somewhat different: the overshoot and drop values are much lower and we do not go beyond the reboot line or BSOD, thus the program works stably throughout its entire cycle.

“error detection” line? For each voltage and frequency, there is a conditional temperature limit at which the processor becomes unstable. Each processor instance will have a different value. For example, on one processor, the frequency of 4300 and the voltage of 1300 mV are stable up to 70 degrees, and on the other processor it is similar, but only up to 65 degrees. It’s the same with the reboot line or BSOD.

VRM: voltage step

One of the interesting problems that you probably won’t run into if your motherboard is in the above-average segment. Others, who have a motherboard from the segment below the average, can only hope that the manufacturer did not limit us. I hope you know — the voltage setting step on AMD processors is 6.25 mV, the program sets your voltage during the stability check, and then reduces it by 1 step, i.e. at 6.25 mV. This is where the problem lies. If you cannot set the voltage yourself with an accuracy of 6.25, but you set it in increments of 12.5, or while testing the program you see that the voltage does not change or is significantly lower, then most likely you will receive instant reboots at some of the steps of the program. How to act in such a case? You just have to remember at what step you had a problem and take the frequencies that were one step before it, and then use these frequencies for voltage during overclocking.

Windows Defender

Sometimes Windows Defender sees a program as a Trojan — this is a bug. Latest VirusTotal report.

Configuration file

If you cannot start the application or some functions do not work, then delete the following folder along the path C:\Users\{your username}\AppData\Local.

In addition, there may be a Nemesis_Ulv2 folder, which must also be deleted.

Autorun

Some versions of the Windows 10 operating system may have corrupted autorun features or may not support them at all. If the program cannot start, then this means that Windows for some reason does not allow it to do so. You can check the autostart status in Task Scheduler. To launch the scheduler, press the keyboard shortcut Windows + R and in the window that appears, enter taskschd.msc, then click OK.

SMU problems

Missing frequency and temperature or CMD_REJECTED_PREREQUISITE or FAILED — Incorrect BIOS version. Not every BIOS version is released without bugs. The solution to this problem is to install a different BIOS version. Also, the problem may lie in running third-party monitoring applications.

CTR does not work

If the program itself displays incorrect values or you see empty fields instead of values (as shown below), you can use the CTR Config Cleaner program, which is available at the main ClockTuner for Ryzen 9 link0003

Description of the program

Now let’s move on to the description of the program. On the left are tabs with the main sections of the program, and on the right — the program itself. The very first tab HOMEPAGE will redirect you directly to the twitter of the developer of the 1usmus program. Next comes the TUNER tab.

TUNER

We are met by a variety of fields, numbers, buttons, switches. We will explore all this further.

Let’s start with the monitoring data, which is located at the top.

CCX number, in this case the first one. In all programs, the designation CCX is indicated in different ways. For example, in Ryzen Master, the counting starts from 0 and it has CCX 0 and CCX 1 for CCD 0 and similarly CCX 0 and CCX 1 for CCD 1. In CTR, the designation is sequential.
Temperature CCX. Here we are greeted by one of the changes in version 2.0. Now the temperature is displayed separately for each CCX for Zen 2, unlike other programs for monitoring parameters. According to the author, AMD processors do indeed measure the temperature of each CCX, although in most programs the temperature is displayed only for the CCD. For Zen 3 CCD = CCX, unlike Zen 2 (Zen 2 has 1 CCD = 2 CCX), so the temperatures will be displayed the same.
Number of cores, some programs like HWINFO count from 0 instead of 1.
Current effective processor frequency. Yes, it is effective, it is considered a little different from what we are used to, so in some monitoring programs you will have a constant frequency during overclocking, and in others, such as HWINFO (effective clock) or ryzen master, you will be able to see the effective frequency.
Marking kernels in Windows by their “hit rating” (CPPC labels). The very best cores usually reach higher frequencies, but this is not always true. So, in my case, CCX1 and CCX2 can operate at the same frequencies, although their core markings are different.

Just below you can see the parameters, among which there were also changes in version 2.0: CPU Usage (%), CPU VID (V), PPT, EDC, TDC remained, CPU SVI2 TFN (V) is now called CPU TEL (V) and new parameters: CPU TEL (A) and CPU TEL (W).

CPU Usage (%) — CPU usage in percent,
CPU TEL (V) — voltage applied to the processor, in volts. According to the author, the most accurate measurement,
CPU VID (V) — voltage requested by the processor, in volts,
CPU PPT (W) (Package Power Tracking) — processor power consumption in watts,
CPU EDC (A) (Electrical Design Current) — the maximum short-term value of current supplied to the processor, in amperes, some boards have a microcode error that displays this value constant at 140 A,
CPU TDC (A ) (Thermal Design Current) — the current value supplied to the processor, in amperes, at elevated temperatures (for example, the characteristics of mosfets often indicate a temperature of 25 degrees).
CPU TEL (A) and CPU TEL (W) — CPU monitoring telemetry elements: current and power consumption values

Next, let’s go a little non-standard way to leave all the most interesting for last, namely from right to left.

ACTIVE PROFILE — currently activated profile.
Info — contains information about the CTR version, your processor, motherboard, BIOS version, SMU version, table version, RAM frequency, current time, and also keeps a log of all program actions after pressing the DIAGNOSTIC, TUNE, STOP buttons , CHECK STABILITY, BOOST TESTER. In addition, in version 2.1, information about the profiles P1, P2, PX appeared, namely, frequencies, voltages and the percentage of load at which they are activated.
— button for copying information to the clipboard, which is displayed in Info.

Now let’s move on to the main block with which we will work.

Settings Mode is used to switch the settings modes to Default (default) and Advanced (extended), which you can see below, but first things first.

On the right is the button RESET SETTINGS — reset the settings to the original ones.

Next, let’s go through Settings Mode — Default. The following is available to us:

Testing mode — testing mode. At the moment, only one is available — AVX Light, which is suitable for almost any operating mode, both rendering and games, with the exception of tasks related only to AVX instructions and only in highly specialized applications.
Cycle time (s) — cycle time for each test step in seconds, a larger value helps to get a more stable result. The recommended value is 360 s.
CCX delta (MHz) — the difference between the best and worst CCX in MHz at which the frequency matching algorithm stops. For Ryzen 3, 5, 7, the recommended value is 25 MHz, for 9 — 150-175 MHz, for XT versions of processors — 100-150 MHz, for Threadripper — 75-100.
Polling period (ms) — polling period for sensors (temperature, voltage, frequency, etc.)

Now let’s switch Settings Mode — Advanced immediately, so as not to return later.

As you can see, the first column remains, but 2 more are added.

Reference voltage (mV) — voltage value at which the processor frequency will be selected.

Reference frequency (MHz) 4100, 4125, 4150, 4175 etc.

Max frequency (MHz) — the maximum frequency value for any of the CCX, at which the program execution will stop. By adjusting this value, you can avoid unnecessary reboots, provided you know approximately at what frequency they occur.

Diagnostic Voltage (mV) — voltage value at which processor diagnostics will be performed. It is recommended to leave it at the current value, if at the same time you have a reboot, then you need to increase this value. For example, my value is 1231 mV, I reduced it to 1150 mV and the PC hung up, rebooted at a value of 1100 mV, although if I started it at a value of 1231 mV, it reached a voltage of 1100 mV without any problems. If this happens to you, then you can set the value to 1300 mV or 1350 mV. Yes, this will increase the number of steps, but it will allow diagnostics.

Max PPT (W), Max EDC (A), Max TDC (A) — consumption and current values that are required for the protection system built into the program. As soon as one of these parameters is reached, the program execution will stop. These values do not apply to PBO.

Max temperature (oC) — temperature value required for the protection system built into the program. As soon as the temperature is equal to the one you specified, the program will stop running.

Switches:

IFSO 1.0 / IFSO 2.0 (Initial frequency smart offset) is a time-saving technology for Ryzen 9 39xx and 59xx and Threadripper owners by allowing frequencies to be set based on CPPC marks. If you have a different processor or suddenly your CCX3 and CCX4 have CPPC label values higher than CCX 1 and CCX 2, then you can disable this switch. What happens if I disable it on Ryzen 9 39xx and Threadripper? All your CCXs will start at a starting frequency of Reference frequency (MHz).

Enhance accuracy — increase the accuracy of finding the required frequency at a given voltage. When testing, it uses 3 stages of 3 steps with the selection of voltages for each frequency value. Significantly increases time. We recommend disabling this option, in earlier versions it helped a lot to get a stable result.

CB20 testing — Cinebench 20 test is used to measure performance before and after program execution. If you are not interested in the comparison between the original and the final result, and also you do not want to compare your result with other users, then this switch can be turned off.

To tray — minimize the program to tray on startup.

Autoshare stats — Automatically load your results into a spreadsheet after the program is finished. It only works if you set the switch before pressing the START button.

Now about the buttons:

TUNE — start the program execution. In the settings mode Settings Mode — Default starts execution at the standard frequency and voltage settings. I have it — 4125 MHz and 1250 mV.

DIAGNOSTIC — launch of the processor diagnostics in order to classify it and issue recommendations on setting voltages and frequencies for overclocking and undervolting. The default voltage for my 3900x processor is 1231mV. Let’s take a closer look in the section on working with the program;

STOP — stop the program execution.

Button — immediate reboot of the system, which becomes active when the CTR protection system for some reason cannot solve the problem.

CHECK STABILITY — check the stability of the system at the current values of frequency and voltage.

BOOST TESTER — boost test for Zen 3 and Renoir processors, as well as for Zen 2 processors, but with some nuances, about which a little later.

PROFILES

A new tab that used to be in the TUNER section and was just a small button.

In this tab, we will look at the main elements from the bottom up.

PROFILE SETTINGS — basic profile settings, including:

Autoload profile(s) — profile autoload along with the operating system. This is especially true for someone who does not have CCX frequencies set in the BIOS.

HYBRID OC is a new feature in CTR 2.0, an alternative boost from the author of the program.

P1 PROFILE — profile for high loads, by default the program suggests recording 1250 mV values and tuning the processor in the TUNER tab with the calculated starting frequency.

VID (mV), CCX (MHz) — the requested processor voltage and frequency of each CCX, which you can set yourself by entering your values and pressing the SAVE P1 PROFILE button, and then turn it on using ACTIVATE P1 PROFILE.

P2 PROFILE — profile for low loads, by default the program suggests recording 1350 mV values and tuning the processor in the TUNER tab with the calculated starting frequency.

Other settings are the same as profile P1.

In addition, there is an interesting setting for the minimum CPU usage or CPU usage min (%), many of you could already guess why it is needed, but more on that later.

PX PROFILE — maximum boost profile, which will be available from version 2.1.

RESULTS

The next tab after PROFILES is RESULTS , which allows you to view the results obtained in CB20 before and after running the program.

By default, this tab is empty, but after error-free execution, you will get something like this: CPU Voltage , as well as the difference between the consumption of CPU PPT . Also on the bottom right you can see the system information section System Information , which displays the frequency of each CCX in the default mode Frequency per CCX — DEFAULT and in the mode you received Frequency per CCX — TUNING , your system information and your PC username are also displayed. With the button SEND STATS you can send your results to a common table with processors similar to yours. The PC username is used for a reason, it will also go into this table as @Anem, so if you want to commit your result, make sure you are not another @Admin. By pressing the corresponding buttons STATS , which are almost in the center at the bottom, you can see the results of other users and how your result is better or worse than others.

ABOUT & HELP

In this section you can see information about the program, see guides for the program from the author and not only, as well as about the modules used. There is also a separate line of gratitude to the people who helped with testing this program.

DONATE & UPGRADE

If you are satisfied with the result, then, I think, it will not be superfluous to support the developer in this not very rewarding business.

Tab MINIMIZE minimizes the program to the taskbar, EXIT closes the program and the main button SCREENSHOT allows you to take a screenshot of the program on the tab that you have open. On this description came to an end, let’s move on to working with CTR.

Working with ClockTuner for Ryzen

Before you start working with the program, take the following as a rule: while the program is running, disable third-party programs that you usually run. This is necessary to obtain higher frequencies, because. the program works in unstable modes and any background activity can give you a noticeably worse result than you would like to see.

As you already know, the program provides 2 operating modes Default and Advanced . We’ll start by looking at working in Default mode. Let’s go straight to the button DIAGNOSTIC . Below you can see the result of passing diagnostics in relation to my processor.

Above shows that 23 steps have passed (Step# 23), the diagnostic ends after one of the threads falls (Thread# 22 fall down!) and the program returns to the previous voltage value (1111 mV). As a result, we get the following Diagnostic Results (from top to bottom): your processor name, CPU VID in mV, CPU TEL in mV, maximum temperature 57. 4 degrees, energy efficiency 3.72, definition of the type of your processor instance, in my case — silver, then let’s go changes in version 2.0: recommended delta values for CCX — 175, theoretical maximum delta for CCX — 200, profile P1 for overclocking — 4250 MHz and 1250 mV in order to reduce the operating voltage and maintain current performance, profile P2 4400 MHz and 1350 mV for maximum “conditionally” safe overclocking, as well as values for the undervolt of 4050 MHz and 1150 mV.

As for the instances of processors, the program uses several levels according to the quality of your processor: bronze — bronze, silver — silver, gold — gold, platinum — platinum. What does it ultimately give? Platinum and Gold are select CPUs that will most likely give you better single-threaded, multi-threaded performance than stock, while lowering the voltages your processor runs at. In the case of silver, everything is not so clear: the result in multi-thread is higher at a slightly lower voltage, while the performance in single-thread is slightly lower. In the case of bronze samples, overclocking may not give you much performance in general, but in any case, I would try. Of course, if you use water cooling, the results will be slightly different in favor of overclocking through CTR, but in the case of silver and bronze, you will most likely get a big difference in multi-threading, but the difference between single-threaded performance will still remain in favor of auto mode .

Why is diagnostic mode needed? It’s simple: it depends on what frequency you start your movement at when using CTR to overclock in the Default mode. Now press the button TUNE . As you can see, IFSO 1.0 / IFSO 2.0 and CB20 testing are enabled in my case, but you can disable the first option, based on the description above in the article. In addition, I would like to note that diagnostics in Zen 3 and Renoir follow a different scenario, unlike Zen 2, but more on that later.

In version 2.1, after pressing the TUNE button in the Info window , information appeared on the performance coefficients for each CCX, as well as the «success» of each core according to information CPPC 43 1294 according to factory information FIT , namely the performance of each core based on the calculation of frequencies at certain voltages and the energy released.

The result was the following: 4275 at 1250 mV, the program set 4375 4400 4275 4275. Why then 4400? How it works IFSO 1.0 / IFSO 2.0 for older processors. Each core of Zen 2 and Zen 3 processors has its own label that characterizes its success. CTR defines all kernel labels in CCX. As a result, we get the label CCX, which is equal to the worst label in CCX. For less successful CCX, the frequency obtained during diagnostics is set, and for more successful ones, a slightly higher frequency. After that, the program started its work. While the program runs several stress tests, you can see at the beginning of each test the CPU Vdroop and the temperature of the hottest CCX pair.

At step 1 (Step# 1), stream number 9 immediately fell, thus the program determined that in CCX2 it could not take the frequency 4400 and it should roll back one step, which is 25 MHz — 4375, and continued the test with the resulting result at the same voltage. The program runs 4 small stress tests at a voltage of 1250 mV, after which it repeats 4 stress tests, but at a voltage lower by 6 mV, namely 1244 mV. This is done specifically to get a more stable result and reduce the chances of a reboot or BSOD on most motherboards and processors. You may have noticed that there is a new line WHEA — counter — WHEA error counter. With a value of up to 3-4, you will not see any negative changes in the operation of your system, in addition, the operating system itself does not display non-critical errors.

Below you can see the output of the program. 4350 4325 4225 4250 at 1250 mV, the result in CB 20 is 7661, the power consumption is 134.9 W at 65.8 degrees. I would like to continue after that, but that’s not all.

The result above is ideal, but when the program is running, you may encounter a reboot, after which the CTR will automatically load (or you will have to start it yourself) and you will see the following:

What does this mean?

Phoenix — hello there 🙂
CTR completed profile creation in alternative mode

or

Phoenix — hello 🙂

CTR finished profiling in alternate mode.

Where are all the steps? This is a logical question and there is an explanation for this. One of the innovations since version 2.0 has been one of the systems called Phoenix. If your PC reboots while the program is running, this will be a signal for the program to stop selecting frequencies for this voltage and save the last option. Again you can see the last result:

CCX1 4375 MHz VID 1250 mV

CCX2 4375 MHz VID 1250 mV

CCX3 4275 MHz VID 1250 mV

CCX4 4275 MHz VID 1250 mV

Why is the frequency lower? CTR uses a system of so-called fines.

If we look at the previous step, we see the following:

Step#2

WHEA-counter: 0

CCX1 (145): 4375 MHz, 1250 mV OC=
CCX2 (156): 4375 MHz, 1250 mV OC=
CCX3 (134): 4275 MHz, 1250 mV OC+
CCX4 (122): 4275 MHz, 1250 mV OC+

Stress test #1 started…

CPU Vdroop: 2% temperature: 51. 4°

Stress test stopped.

Stress test #2 started…

CPU Vdroop: 1.8% temperature: 51.7°

Stress test stopped.

Stress test #3 started…

CPU Vdroop: 1.8% temperature: 51.6°

Stress test stopped.

Stress test #4 started…

CPU Vdroop: 1.8% temperature: 51.8°

Stress test stopped.

CCX1 and CCX2 have an = sign, while CCX3 and CCX4 have a + sign. Those. the program shows that on CCX1 and CCX2 it is necessary to stop in the selection of a higher frequency, and CCX3 and CCX4 can be increased by another 25 MHz and after that the PC is rebooted. The processor cannot run at 4300 MHz for CCX3 and CCX4 at 1250 mV. So the last two CCXs are to blame. They receive a penalty of 25 MHz, in addition to this, the voltage on the processor rises by 12.5 mV (rounded up to 13 mV in monitoring). In addition, the program recalculates the frequency for CB20 and thus takes another 25 MHz from each CCX, and we get our cherished 4350 4350 4225 4225.

Now back to our test in version 2.1. Now we can graphically view the result on the tab RESULTS . Yes, the result for a silver sample is quite impressive.

When the processor was running, the default frequencies were 4005 MHz for all CCX, voltage — 1.297 V, heat dissipation — 138.8 W, and the result is 7190. Now 4350 4325 4225 4250 at a voltage of 1.25 V and heat dissipation 134.9 W, and the result — 7661. For those who want to see the percentage: voltage decreased by 3.62 percent, heat dissipation by 2.81 percent, and multi-thread performance increased by 6.15 percent. Of course, in this case, the performance in a single thread is reduced, not for everyone it will be critical. As I already noted, in the case of more successful processors, the result in a single thread can be even higher. Everything is individual.

Now switch to Advanced mode. In this mode, more extensive possibilities open up before us. For example, if you have problems on Default , then in this mode you can still try to get the result, of course, if your board has a voltage setting step of 6.25 mV, which I wrote about above.

Let’s see a different result at a different voltage so that there are no repetitions.

As you can see, I set Reference Voltage (mV) to to a value of 1350 and this is the maximum value given by 1usmus.

If you have the LLC level set correctly, you will always have a voltage drop when the amount of current supplied to the processor increases. The logic of operation is ultimately simple: more current — less voltage and low current — high voltage. Notice how the processor works in its normal state: at the minimum current value, often in single-threaded loads, the voltage can go up to 1.47 V, and in multi-threaded loads with a large current value, it can drop to 1. 25 V. The AVX Light test mode does not imply a large AVX load, so your PC will most likely just reboot if you still have such a task, but these are not renders or games, but, as I already noted, highly specialized tasks. In this case, you will need to choose the frequencies and voltages yourself or leave the state of the processor in the stock state.

So, next I set Reference frequency (MHz) equal to 4400, which were obtained as P2 value during diagnostics. I would like to recommend setting the value slightly less than that obtained during diagnostics. Yes, it will take more time, but the result in terms of stability will be better. I also have IFSO 1.0 / IFSO 2.0 and CB20 testing disabled to minimize rebooting while testing on Asrock X570 Taichi and 3900x motherboards and getting comparative results. Enhance accuracy turn on if necessary, increases the accuracy when searching for the required frequency, but increases the time to pass the tests. Autoshare stats is off because my overclock data is already in the table. The following result was obtained: 4475 4450 4350 4350 at a voltage of 1363 mV with a heat dissipation of 167.6 W, a temperature of 69.4 degrees and 7880 points in CB20.

As you can see, the frequencies are higher, but the voltage is also higher. For comparison, at 1250 mV, the frequency is 4350 4325 4225 4250, the result in CB 20 is 7661, the power consumption is 134.9W at 65.8 degrees. As a result, the increase relative to the previous result was 8.29 percent in voltage, 21.25 percent in consumption, and 10.84 percent in temperature. And now let’s compare the same thing, but with the usual mode of operation of the processor.

As a result, it was 4005 4005 4005 4005 at a voltage of 1.297 V, consumption is 138.8 W, the result is 7190. It became 4425 4425 4325 4275 at a voltage of 1.363 V, consumption is 171.3 W, the result is 71.3 W32 then voltage increased by 4.84 percent, consumption — by 18. 97 percent, the result is 8.2 percent. With regard to voltages, the increase is not bad, but if you look at the received consumption, the result is rather sad. At the moment, a voltage of 1.356 V is used, which is slightly different from the obtained values \u200b\u200band was calculated empirically, but why not 1.25 V? Higher frequency in single-threaded performance, slightly higher result in multithreading and slightly lower latency when working with memory. On the other hand, why not, if the temperatures allow you to work in this mode.

With the results everything became clear, and now let’s move on to the interesting tab PROFILES . The program still remembers our result, so immediately click on the button FILL & SAVE P1 PROFILE .

Do not forget that by default profile P1 is used for frequencies and voltages under high load. You will then see the following:

A successful save will display filled and saved! Now we have recorded frequency and voltage for high loads. On this, of course, one could finish, press button ACTIVATE P1 PROFILE , put the switch Autoload profile with OS on, go to the tab TUNER , press To tray to minimize the program to the system tray. After that, press EXIT . The program writes a task for autorun in Windows and that’s it, our overclocking is ready. But that’s not all…

Similarly, we perform tuning in the tab TUNER at a higher voltage of 1350 mV and write down our values.

Yes, you can separately select the frequency for different voltages and then fill in all the necessary lines, and press SAVE PROFILE . Please note that the number of CCXs for processors is different. From version 2.1, only existing CCXs are displayed, so filling errors will be minimized. After that you will get the following:

Both profiles are filled. Then you have two options for working with the program. You can manually turn on the profile at the right time, after turning it on you will see the inscription activated and the rocket icon next to the profile.

Or you can use the “magic” of CTR. To do this, turn on the CTR HYBRID OC switch and activate both profiles. Yes, this is not a mistake. Below you can see the result. The frequencies are equal to the P2 profile and next to the rocket icon. The value of VID (mV) in CTR is equal to VID in HWINFO. I would like to remind you once again that VID is the requested voltage, and the current voltage is highlighted in blue in HWINFO.

Why include two profiles then? Let’s go back to the previous image.

In two profiles there is a line CPU usage min (%) and it is responsible for launching a certain profile with the appropriate CPU load. Profile P2 turns on at 18 percent CPU load and runs smoothly until 82 percent load and switches to profile P1. It remains to find out what frequency the processor will have at a load of less than 18 percent. It will be automatic. Of course, you can move the inclusion boundaries of each of the profiles. However, if you have the PX profile enabled in version 2. 1, then the author does not recommend changing the default values.

To understand what is happening, I recommend watching the video below. Pay attention to the change in frequencies under different types of loads, following the change in the position of the rocket icon in the CTR program. In addition, the video was filmed according to version 2.0, in version 2.1 the logic of work is identical.

Update 2.1

And now let’s move on to the innovation of version 2.1 — this is an improved diagnostic mode and PX PROFILE.

You can see the improved diagnostics, as well as the principle of operation of some functions for Zen 3 and Renoir, in the author’s video below.

What is a PX profile? Let’s turn to the video.

As you can see in idle, the frequency of the most successful cores is at the maximum value, as soon as the CPU utilization reaches 29 percent, we switch to the P2 profile, and at a load above 81 percent to the P1 profile.

You could see the multistream result above for 1250mV, the single stream result is:

Is this a good or bad result? 527 for this instance of the processor was the ceiling in bus mode, taking into account all the conditions that had to be met. 521-523 — there was an almost constant value, i.e. the increase was from a constant value of 3.34 percent on Zen2. On Zen 3, the result will be more significant due to changes in the layout of the crystals.

Let’s dwell a little on the PX profile values, High state, Mid state, Low state — new settings. As you can see, I currently have 4625MHz, 4575MHz, 4550MHz for the respective states.

High state — frequency of processor cores with a load of 1-2 threads,

Mid state — frequency of processor cores with a load of 3-4 threads,

Low state — frequency of processor cores with a load of 5 -8 streams.

The boost of cores is selected from the luck of each individual core. The video above shows all modes of operation. The more background processes you have running, the less likely it is that your processor will run on a small number of threads. In the video, you may have noticed that you had to change the number of threads for the test due to the fact that the screen recorder also used processor power.

BOOST TESTER

Finally, let’s dwell on this button in the Tuner menu to talk about the «fly in the ointment» for Zen 2 processors and, in general, about the principle of operation

For Zen 3 and Renoir, everything works correctly due to the presence of the necessary sensors in the processors themselves. As a result, you will get a table with core numbers, frequency, requested VID voltage, heat dissipation of each POWER core, and processor temperature.

In the case of Zen 2 you will also get this table as you can see above, but the author states that this mode works quite rough for Zen 2, so you could see that the frequencies are set to 4625 4575 and 4550 in the states, and not , for example, 4525 4475 and 4450, which are received in this mode.

Of course, it would be possible to set up a boost tester on Zen 2, but with a high degree of probability we would have received PC reboots at this stage, but is this necessary? I think no. Picking up 3 values, if desired, is not difficult, in addition to this, you will already know some starting points.

Results

Already enough people have posted their results in the public domain, and therefore there is something to see.

Below you can see the results for Ryzen 5 3600, 3600X and 3600XT.

Results for Ryzen 7 3700X, 3800X, 3800XT.

Results for Ryzen 5 5600x.

Results for Ryzen 7 5800x.

First, let’s look at the number of results. 5492 results for 8-core and 6623 6-core Zen2 processors. Among Zen 3, 181 8-core processors and 89 6-core processors have been tested so far. As you can see, the situation is reversed: 8 Zen 3 cores are more than 6.

Now let’s talk about the results for Zen 2 and Zen 3. For the Ryzen 5 3600, 3600X and 3600XT, the minimum frequency bar without taking into account rare values is 3600 MHz. For Ryzen 7 3700X, 3800X, 3800XT — 3700 MHz. For Zen 3, the sample is not so large, so we count almost from the lowest frequency. For Ryzen 5 5600x — 4200 MHz, and for Ryzen 7 5800x — also 4200 MHz. As a result, for Zen 3 it is 500-600 MHz higher than for Zen 2. If we talk about the average frequency for all processors, then for Ryzen 5 3600, 3600X and 3600XT — 4184 MHz, for 3700X, 3800X, 3800XT — 4209MHz, for Ryzen 5 5600x — 4437 MHz, for Ryzen 7 5800x — 4580 MHz.

If we digress a little and consider the choice between the Ryzen 5 3600, 3600X and 3600XT and the Ryzen 7 3700X, 3800X, 3800XT, then the Ryzen 7 3700X, 3800X, 3800XT is needed more for those who want to have 8 cores, not 6. In addition, if you carefully analyze the table, then the Ryzen 5 3600, 3600X and 3600XT frequency spread is higher and therefore the probability of getting a less successful instance is also higher.

Regarding the choice of Ryzen 5 5600x and Ryzen 7 5800x, it’s already harder to say, because. the number of copies is smaller, but if you look at the Ryzen 7 5800x, there is practically no difference between a little more than 1.25 V and a little more than 1.325 V. With the Ryzen 5 5600x, most of the runs were done at a little over 1.15V and a little over 1.3V. Users are likely to take the Ryzen 5 5600x in combination with a not the most powerful cooling system, so they undervolt their processor, while the frequency remains at 4300 — 4500 MHz , which is generally more than Zen 2 by 100 — 500 MHz, depending on the luck of Zen 2 and Zen 3.

In the end, you can see that there is nothing complicated in the operation of the program, you just need time for the full cycle of the program, because when working it is desirable not to perform any operations on your PC and know where you need to click and what values select frequencies, voltages, and other parameters, although in the case of the Default mode, everything is quite simple. Thus, you can flexibly customize the program to your requirements and get the result you need.

Unfortunately, during tests on the Asrock x570 Taichi motherboard, there are problems when using CTR on older processor lines. As you have seen, both results are obtained after triggering the new Phoenix system, i.e. each tuning was completed with a reboot of the PC. In addition, the results floated from tuning to tuning and were sometimes unstable. Will you have something similar? Most likely not, because when choosing the right LLC level, there should ideally be no reboots (if you have a non-Asrock motherboard and one of the 39 processors is installed on it00x, 3950x, 5900x and 5950x). As I already noted, there are too many nuances during overclocking, in this case the program does everything by itself and tries to take everything into account for you. To use it or not? I took advantage, got new experience, learned about great testing methods, got great frequencies. If you like to touch everything yourself and want to try overclocking or undervolting yourself, then I would still postpone this program to test your strengths and as an additional experience with new software.

Thanks to 1usmus for ClockTuner for Ryzen and making us happy with this software.

Good luck to everyone in reaching new heights of overclocking your processors!

PC builders complain: AMD flooded the market with processors that don’t work at all

Technics

15 February 2021 11:34 15 Feb 2021 11:34

|

Share

The latest AMD processors fail several times more often than Intel chips. This is the conclusion of the PowerGPU custom PC assembly company — it claims that the defect rate in new AMD processors reaches 6%, and motherboards based on fresh AMD 500 chipsets are not reliable. The flagship 16-core Ryzen 9 5950X processor caused the most complaints, and among the Intel PowerGPU chips, it found only one defective chip in 22 years.

Marriage put on stream

AMD has been accused of having an excessively high percentage of defective copies of its latest Ryzen 5000 series processors. It is blamed by PowerGPU, a company that builds custom desktop PCs. According to them, a huge number of new AMD processors do not work out of the box, that is, they are initially defective, and some working instances begin to function incorrectly after some time in assemblies.

PowerGPU employees calculated that out of a batch of top AMD Ryzen 9 5950X processors, consisting of 50 chips, eight CPUs refused to work, out of 50 Ryzen 9 5900X four chips turned out to be defective, and the same number was in a Ryzen 7 5800X batch of 100 pieces. Noticeably less marriage turned out to be among the Ryzen 5 5600X — the company ordered 120 of these processors, and only three of them did not start after the PC was assembled.

Statistics not in favor of AMD

In total, out of 320 AMD processors purchased by PowerGPU, 19turned out to be non-working, and this is about 5.9% of their total number. The company’s specialists also noticed that the more productive the processor, the higher the percentage of defects.

The most productive processor in the new AMD line turned out, according to PowerGPU, to be also the most poor quality

PowerGPU emphasized that they assemble their PCs not only on AMD, but also on Intel, but in the case of the latter, they came across only one non-working processor in the entire history of the company. How close this statement is to the truth remains unknown, but the company has been assembling computers since 1999 «Before the release of the Ryzen 5000 line, we assembled 80% of computers on Intel and 20% on AMD,» the company said.

According to PowerGPU employees, the only defective Intel processor was the Core i7-9700K, which went on sale in the fourth quarter of 2018 and is produced using a 14-nanometer process technology. This is an octa-core processor with eight threads (i.e. without Hyper Threading), running at a frequency of 3.6 to 4.9 GHz. In comparison, the AMD Ryzen 9 5950X, which received the most complaints from PowerGPU experts, was released in October 2020 and offers users 16 cores, 32 threads, clock speeds from 3.4 to 4.9GHz and 7nm process technology.

Cause not established

PowerGPU claims that they are facing defects not only in AMD processors, but also in third-party motherboards. They mention board models based on AMD 500 chipsets, in particular, based on B550 and X570 — according to their calculations, every week they come across from three to five non-working boards on these chipsets.

Post PowerGPU on Twitter

What exactly influences such a high, according to PowerGPU, percentage of defects among AMD products, remains unknown. It should be noted that, unlike Intel, it currently does not have its own factories, and it orders the production of all its 7-nm semiconductor products from the Taiwanese TSMC, one of the first in the world to master the 5-nm process technology and is preparing to switch to 3 nm.

Vadim Glushchenko, Center for Global IT Cooperation: Focusing on the Eastern Direction

Digitalization

Intel, in turn, in August 2019 mastered the 10-nanometer production standards at its own factories, but, as CNews reported, it still clings to 14 nanometers, justifying this with financial benefits. However, for the foreseeable future, it may also turn to TSMC for help in releasing its chips using modern technology. Such a forecast was voiced by TrendForce analysts in January 2021 — they believe that the first Intel CPUs to leave the TSMC assembly line will be 5-nanometer Core i5.

Other confirmations

Other assemblers also talk about the numerous defective AMD chips. Insider harukaze5719 drew attention to the PowerGPU complaint — on his Twitter he wrote that he had contacted a Korean PC assembly company (its name was not disclosed), and that company also confirmed the fact that there were problems with the new Ryzen and motherboards on the B550 and X570. At the same time, its representatives called the level of marriage indicated by PowerGPU «ridiculous», saying that such conclusions should not be drawn from a sample of three hundred processors.

PowerGPU’s arguments didn’t convince everyone

AMD itself did not react to PowerGPU reports about a high percentage of defects in its products at the time of publication of the material, but this is not the first time when its devices are spoken of in this way. For example, in August 2020, Mindfactory, a major German hardware retailer, compiled statistics on the return of video cards based on GeForce (Nvidia) and Radeon (AMD).

According to this report, Mindfactory sold 44.