AMD Ryzen 5 5600X (Zen3) Review & Benchmarks – CPU 6-core/12-thread Performance

What is “ZEN3” (Ryzen 5000)?

AMD’s Zen3 (“Vermeer”) is the 3rd generation ZEN core – aka the new 5000-series of CPUs from AMD, that introduces further refinements of the ZEN(2) core and layout. An APU version (with integrated “NaviX” graphics) is also scheduled to be launched later (as normal) but this time likely to keep the 5000-series moniker. The CPU/APUs remain socket AM4 compatible on desktop – thus allowing in-place upgrade (subject to BIOS upgrade as always) – but series 500-chipsets are recommended to enable all features (e.g. PCIe4, etc.). [Note this is the last CPU that will fit AM4 socket; future CPUs supporting DDR5 need a new socket]

Unlike ZEN2, the main changes are to the core/cache layout but they could still prove significant considering the cache/memory latencies issues that have impacted ZEN designs:

• Claim +19% IPC overall improvement vs. ZEN2
• Higher base and turbo clocks +3% [for 5600X vs. 3600XT]
• Still built around “chiplets” CCX (“core complexes”) but now of 8C/16T and 32MB L3 cache (still 7nm)
• Same central I/O hub with memory controller(s) and PCIe 4.0 bridges connected through IF (“Infinity Fabric”) (12nm)
• Still up to 2 chiplets on desktop platform thus up to 2x 8C (16C/32T 5950X)
• L3 is still the same 32MB but now unified (not 2x 16MB) still up to 64MB on 5950X
• 20 PCIe 4.0 lanes
• 2x DDR4 memory controllers up to 3200Mt/s official (4266Mt/s max)

1 chiplet (1 large CCX) + I/O hub

To upgrade from Zen2 (Ryzen 3000) or not?

Micro-architecturally there are more changes that should improve performance and security:

• VAES 256-bit (vs. AES HWA 128-bit) [note that VAES/AVX512 is 512-bit]
• Control Flow Integrity eXtensions (CFX) & Shadow Stacks (SSX)
• Multi-Key Memory Encryption, e.g. individually encrypted VM memory
• Inter-core latencies reduced through shared L3 (8C and less); no more trips to memory to share data
• AMD processors have thankfully not been affected by most of the vulnerabilities bar two (BTI/”Spectre”, SSB/”Spectre v4″) that have now been addressed in hardware.

You also need to watch out for the compatibility issues especially for older boards:

• X570, B550, A520 boards need AGESA 1.0.8.0 for Zen3 support
  • AGESA 1.1.0.0 Patch B/C or later recommended
• X470, B450 boards need at least AGESA 1.0.1.0 to boot Zen3 and won’t receive full support for some time
• X370, B350, A320 boards are not likely to be updated for Zen3

In this article we test CPU core performance; please see our other articles on:

• Zen3
  • AMD Ryzen 7 5800X (Zen3) Review & Benchmarks – Cache and Memory Performance
  • AMD Ryzen 9 5950X (Zen3) Review & Benchmarks – CPU 16-core/32-thread Performance
  • AMD Ryzen 7 5800X (Zen3) Review & Benchmarks – CPU 8-core/16-thread Performance
• Zen2
  • AMD Ryzen 9 3900X (Zen2) Review & Benchmarks – CPU 12-core/24-thread Performance
  • AMD Ryzen 7 3700X (Zen2) Review & Benchmarks – CPU 8-core/16-thread Performance
  • AMD Ryzen 7 3700X (Zen2) Review & Benchmarks – Cache and Memory Performance
  • AMD Ryzen Threadripper 3970X, 3960X Review & Benchmarks – CPU Performance

Hardware Specifications

We are comparing the middle-range Ryzen 5 5000-series (Zen3 6-core) with previous generation Ryzen 5 3000-series (Zen2 6-core) and competing architectures with a view to upgrading to a middle-range but still high performance design.

Disclaimer

This is an independent article that has not been endorsed nor sponsored by any entity (e.g AMD). All trademarks acknowledged and used for identification only under fair use. Errors and omissions excepted (E&OE).

The article contains only public information (available elsewhere on the Internet) and not provided under NDA nor embargoed. At publication time, the products have not been directly tested by SiSoftware and thus the accuracy of the benchmark scores cannot be verified; however, they appear consistent and do not appear to be false/fake.

Native Performance

We are testing native arithmetic, SIMD and cryptography performance using the highest performing instruction sets (AVX2, FMA3, AVX, etc.). Zen3 supports all modern instruction sets including AVX2, FMA3 and even more like SHA HWA but not AVX-512.

Results Interpretation: Higher values (GOPS, MB/s, etc.) mean better performance.

Environment: Windows 10 x64, latest AMD and Intel drivers. 2MB “large pages” were enabled and in use. Turbo / Boost was enabled on all configurations. All mitigations for vulnerabilities (Meltdown, Spectre, L1TF, MDS, etc.) were enabled as per Windows default where applicable.

Native Benchmarks		AMD Ryzen 5 5600X 6C/12T (Vermeer)	AMD Ryzen 5 3600XT 6C/12T (Matisse)	Intel i9 9900K 8C/16T (Coffeelake-R)	Intel i7 8700K 6C/12T (Coffeelake)	Comments
	Native Dhrystone Integer (GIPS)	341 [+15%]	297	366	273	Zen3 starts strongly with 15% faster than Zen2 in this legacy integer benchmark.
	Native Dhrystone Long (GIPS)	347 [+16%]	300	347	255	With a 64-bit integer workload still 16% improvement.
	Native FP32 (Float) Whetstone (GFLOPS)	221 [+26%]	176	243	173	Floating-point performance is even better, 25% better than Zen2.
	Native FP64 (Double) Whetstone (GFLOPS)	183 [21%]	151	196	143	With FP64 nothing much changes again.
The 6-core Zen3 improves by 15-26% compared to the old Zen2-XT which is still decent considering almost the same Turbo clock. This means it is not quite enough to beat the 8-core CFL-R though it soundly outperforms the 6-core CFL.
	Native Integer (Int32) Multi-Media (Mpix/s)	1,113 [+38%]	804	1,000	781	Zen3 is still 38% over Zen2 despite same width SIMD units.
	Native Long (Int64) Multi-Media (Mpix/s)	446 [+56%]	286	435	320	With a 64-bit AVX2 integer vectorised workload, Zen3 is 56% faster.
	Native Quad-Int (Int128) Multi-Media (Mpix/s)	87.4 [+31%]	66.5	83	62.2	This is a tough test using Long integers to emulate Int128 (now vectorised), Zen3 is still 31% faster.
	Native Float/FP32 Multi-Media (Mpix/s)	1,131 [+38%]	820	998	746	In this floating-point AVX/FMA vectorised test, Zen3 is again 38% faster than Zen2.
	Native Double/FP64 Multi-Media (Mpix/s)	670 [+39%]	481	573	427	Switching to FP64 SIMD code, Zen3 is 39% faster.
	Native Quad-Float/FP128 Multi-Media (Mpix/s)	26 [+27%]	20.54	27	20.5	In this heavy algorithm using FP64 to mantissa extend FP128, Zen3 still manages to be 27% faster.
While Zen2 with its new 256-bit wide SIMD units was almost 2x faster (+100%) than Zen1/+, Zen 3 still manages to improve anywhere between 25-56%, even better than what we’ve seen in the lagacy benchmarks. As a result it even beats the 8-core CFL-R is just about all tests while it ties one. It shows that even with 2 more cores (+33%), the old Intel cores just cannot compete these days.
	Crypto AES-256 (GB/s)	26*** [+13%]	23	16.7	15.8	Zen3 support VAES (256-bit) but all CPUs are memory bandwidth bound – Zen3 is 13% faster.
	Crypto AES-128 (GB/s)			16.7	15.8	What we saw with AES-256 just repeats with AES-128.
	Crypto SHA2-256 (GB/s)	19.8** [+24%]	16**	12.3	9.3	With SHA/HWA Zen3 similarly powers through hashing tests leaving Intel in the dust – and is still 24% faster than Zen2!
	Crypto SHA1 (GB/s)			22.7	17.6	The less compute-intensive SHA1 does not change things due to acceleration.
	Crypto SHA2-512 (GB/s)			9	6.8	–
While streaming tests (crypto/hashing) are memory bound, Zen3 still manages a decent 13-24% improvement over Zen2, again enough to beat both Intel CPUs, even with more cores. As Zen3 can use even faster memory – performance can improve even further. Note***: using VAES 256-bit (AVX2) Note**: using SHA HWA not SIMD (e.g. AVX512, AVX2, AVX, etc.)
	Black-Scholes float/FP32 (MOPT/s)			285	212	The stadard financial algorithm.
	Black-Scholes double/FP64 (MOPT/s)	244 [+24%]	196	254	183	Switching to FP64 code, Zen3 is 24% faster than Zen2.
	Binomial float/FP32 (kOPT/s)			60.5	43.8	Binomial uses thread shared data thus stresses the cache & memory system;
	Binomial double/FP64 (kOPT/s)	71.27 [+46%]	48.9	63.13	46.8	With FP64 code Zen3 is now 46% faster.
	Monte-Carlo float/FP32 (kOPT/s)			218	162	Monte-Carlo also uses thread shared data but read-only thus reducing modify pressure on the caches;
	Monte-Carlo double/FP64 (kOPT/s)	99.21 [+16%]	85.82	88	66.8	The cache improvements show here, Zen3 is 16% faster.
Ryzen always did well on non-SIMD floating-point algorithms and here it further cements its dominance: 6-core Zen3 anywhere between 16-46% faster than Zen2 and again beats the 8-core CFL-R in most tests (or at least ties). Intel will need more cores to match its performance and that is not easy to do.
	SGEMM (GFLOPS) float/FP32			411	370	In this tough vectorised algorithm that is widely used (e.g. AI/ML).
	DGEMM (GFLOPS) double/FP64	147 [-3%]	152	231	179	With FP64 vectorised code, we have an outlier here.
	SFFT (GFLOPS) float/FP32			20.5	20.26	FFT is also heavily vectorised but stresses the memory sub-system more.
	DFFT (GFLOPS) double/FP64	8.79 [+33%]	6.59	9.41	9.89	With FP64 code, Zen3 is 33% slower likely memory related.
	SNBODY (GFLOPS) float/FP32			483	368	N-Body simulation is vectorised but fewer memory accesses.
	DNBODY (GFLOPS) double/FP64	177 [+18%]	150	172	126	With FP64 precision Zen3 is 18% faster.
With highly vectorised SIMD code Zen3 still improves by a decent amount, although memory-access latency sensitive algorithms (not streaming) like FFT/N-Body are still problematic. We saw in our other reviews that GEMM is faster on Zen3 and here we have an outlier for some reason that needs to be investigated.
	Blur (3×3) Filter (MPix/s)	2,709 [+43%]	1,892	2,590	2,000	In this vectorised integer workload Zen3 starts 43% faster than Zen2!
	Sharpen (5×5) Filter (MPix/s)	1,074 [+46%]	735	1,000	766	Same algorithm but more shared data makes Zen3 46% faster.
	Motion-Blur (7×7) Filter (MPix/s)	533 [+41%]	378	524	390	Again same algorithm but even more data shared still 41% faster
	Edge Detection (2*5×5) Sobel Filter (MPix/s)	858 [+33%]	646	879	657	Different algorithm but still vectorised workload Zen3 is 33% faster.
	Noise Removal (5×5) Median Filter (MPix/s)	88.5 [+29%]	68.7	77.82	58.33	Still vectorised code but Zen3 is “only” 29% faster.
	Oil Painting Quantise Filter (MPix/s)	28.5 [+50%]	19	43	31.92	This test has always been tough for Ryzen but Zen3 still manages 50% improvement!
	Diffusion Randomise (XorShift) Filter (MPix/s)	2,656 [+2.18x]	1,220	4,000	3,330	With integer workload, Intel CPUs seem to do much better but Zen3 is over 2x faster than Zen2.
	Marbling Perlin Noise 2D Filter (MPix/s)	341 [+52%]	224	610	455	In this final test again with integer workload Zen3 is 39% faster.
While Zen2 brought almost 2x improvement due to its 256-bit wide SIMD units, Zen3 still manages anywhere between 29-52% improvement, generally enough to beat Intel’s 8-core CFL-R or at least be competitive with it. This is something Intel always used to win by a big margin and now seems to be firmly AMD’s favourite.
	SCNN Inference (Samp/s) float/FP32	88.45 [+13%]	78.46	46.79	46.33	Zen3 is 13% than Zen2 in inference.
	SCNN Training (Samp/s) float/FP32	17.1 [+3%]	16.57	6.46	5.64	Traning only improves by a measly 3%.
	SRNN Inference (Samp/s) float/FP32	126.24 [+58%]	80.08	75.03	55.51	Zen3 is 58% faster than Zen2 and much faster than Intel.
	SRNN Training (Samp/s) float/FP32	3.83 [+31%]	2.93	3.48	2.69	Training also improves by a generous 31%.
Zen3 does extremely well overall with neural networks, with large speed increases (13-58%) depending on the task. The unified L3 caches helps inter-core data transfers greatly which is widely used in neural networks. Note*: using AVX512 not AVX2/FMA3.
	Aggregate Score (Points)	8,240 [+16%]	7,130	8,150	6,520	Across all benchmarks, Zen3 is 16% faster than Zen2!
Overall the 6-core Zen3 (X) is only about 16% faster than the old Zen2 (XT), as it can cannot rely on much faster Turbo as we’ve seen with its other siblings. While we may have hoped for more, the XT Zen2 have too good performance rather than Zen3 not improving enough.
	Price/RRP (USD)	$300 [+25%]	$240 [higher now due to demand]	$480	$380 [lower now due to age]	Somewhat large price increase, larger than improvement over Zen2.
	Price Efficiency (Perf. vs. Cost) (Points/USD)	27.46 [-8%]	29.70	16.97	17.15	Zen3 is 8% less cost efficient vs. the old Zen2.
Due to the (modest) performance gains, the 5600X price efficiency is lower (-8%) than the old 3800XT due to the relatively large price increase (+25%). On the other hand the current prices of 3000-series is up by $50+ which makes the new part better value all things considered. While Intel does not seem price efficient, the cost is lower today and likely to be reduced further soon.
	Power/TDP (W)	65 (135 PL2 turbo) [=]	65 (135 PL2 turbo)	95 (135 PL2 turbo)	95 (135 PL2 turbo)	Zen3 keeps the same TDP as Zen2.
	Power Efficiency (Perf. vs. Power) (Points/W)	126.77 [+16%]	109.69	85.79	68.63	Zen3 is 16% more power efficient than Zen2.
Despite the (modest) performance gains, Zen3 is 16% more power efficient as the TDP remains the same as Zen2. Still, due to better Turbo (Boost) and higher clocks, Zen3 will spend more time closer to the power limit than the old Zen2.

SiSoftware Official Ranker Scores

• AMD Ryzen 9 5950X 16-Core/32-Threads
• AMD Ryzen 7 5800X 8-Core/16-Threads
• AMD Ryzen 5 5600X 6-Core/12-Threads

Final Thoughts / Conclusions

Executive Summary: Zen3 (5600X, 6-core) is ~16% faster than Zen2 (3600XT, 6-core) across all kinds of algorithms but 25% more expensive. We’ll give it 8/10.

Despite no major architectural changes (except larger 8-core single CCX layout (2 cores disabled here) and thus unified L3 cache) over Zen2 – Zen3 manages to be quite a bit faster across legacy and heavily vectorised SIMD algorithms, though in the case of the 6-core (aka 5600X) due to its almost identical Turbo compared to old Zen2 (3600XT) – it cannot beat it so soundly as the other siblings have as we’ve seen in the other reviews. The Zen2 XTs performance is just too good!

Still, it is enough to overtake the Intel competition, even the old top-end 8-core i9 9900K which is a massive win – and naturally leaves the older 6-core i7 8700K in the dust.

Again, in effect it is like getting a 8-core CPU for the price of the 6-core and despite of the higher cost vs. older CPUs; it is a big upgrade especially for older AM4 boards likely still using Zen1/Zen+ CPUs (e.g. 1600, 2600 or so). At this level, PCIe3 performance (on 400, 300-series boards) is just fine and you are getting a relatively cheap system that can beat older top-of-the-range systems of the past. [Note: as long as there is a BIOS update available or you can mod one]

As before – if you want one of these CPUs, make sure you buy or even pre-order one now to avoid disappointment, though watch out for the relatively large price increase!

Please see the other reviews on how the 8-core (5800X) and the top-end 16-core (5950X) variants perform:

• Zen3
  • AMD Ryzen 7 5800X (Zen3) Review & Benchmarks – Cache and Memory Performance
  • AMD Ryzen 9 5950X (Zen3) Review & Benchmarks – CPU 16-core/32-thread Performance
  • AMD Ryzen 7 5800X (Zen3) Review & Benchmarks – CPU 8-core/16-thread Performance
• Zen2
  • AMD Ryzen 9 3900X (Zen2) Review & Benchmarks – CPU 12-core/24-thread Performance
  • AMD Ryzen 7 3700X (Zen2) Review & Benchmarks – CPU 8-core/16-thread Performance
  • AMD Ryzen 7 3700X (Zen2) Review & Benchmarks – Cache and Memory Performance

Disclaimer

This is an independent article that has not been endorsed nor sponsored by any entity (e.g AMD). All trademarks acknowledged and used for identification only under fair use. Errors and omissions excepted (E&OE).

The article contains only public information (available elsewhere on the Internet) and not provided under NDA nor embargoed. At publication time, the products have not been directly tested by SiSoftware and thus the accuracy of the benchmark scores cannot be verified; however, they appear consistent and do not appear to be false/fake.