AMD Ryzen 7 5800X (Zen3) Review & Benchmarks – CPU 8-core/16-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:

  • (AMD) Claims +19% IPC (instructions per clock) overall improvement vs. ZEN2
  • Higher base and turbo clocks +7% [for 5800X vs. 3700X]
  • 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) [future AM5 socket for DDR5 support]

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
    • X570 recommended due to better VRMs especially if you overclock
  • X470, B450 boards need at least AGESA 1.0.1.0 to boot Zen3 and won’t receive full support for some time
    • X470 recommended due to better VRMs especially if you overclock
  • 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:

Hardware Specifications

We are comparing the top-range Ryzen 7 5000-series (Zen3 8-core) with previous generation Ryzen 7 3000-series (Zen2 8-core) and competing architectures with a view to upgrading to a top-range, high performance design.

CPU Specifications AMD Ryzen 7 5800X 8C/16T (Vermeer)
AMD Ryzen 7 3700X 8C/16T (Matisse) AMD Ryzen 7 2700X 8C/16T (Pinnacle Ridge) Intel i9 9900K 8C/16T (Coffeelake-R) Intel i9 7900X 10C/20T (Skylake-X) Comments
Cores (CU) / Threads (SP) 8C / 16T 8C / 16T 8C / 16T 8C / 16T 10C / 20T Core counts remain the same.
Topology 1 chiplet, 1 CCX, each 8 core (8C) + I/O hub 1 chiplet, 2 CCX, each 4 cores (8C) + I/O hub 2 CCX, each 4 cores (8C) Monolithic die Monolithic die Large CCX with 8 cores not 4
Speed (Min / Max / Turbo) (GHz)
3.8 [+6%] / 4.7GHz [+7%] 3.6 / 4.4GHz 3.7 / 4.2GHz 3.6 / 5GHz 3.3 / 4.3GHz Both base and turbo are up 5-8%.
Power (TDP / Turbo) (W)
105 / 135W (PL2) 105 / 135W (PL2) 105 / 135W (PL2) 95 / 135W (PL2) 140 / 308W (PL2) Same TDP
L1D / L1I Caches (kB)
8x 32kB 8-way / 8x 32kB 8-way 8x 32kB 8-way / 8x 32kB 8-way 8x 32kB 8-way / 8x 64kB 4-way 8x 32kB 8-way / 8x 32kB 8-way 10x 32kB 8-way / 10x 32kB 8-way No changes to L1
L2 Caches (MB)
8x 512kB (4MB) 8-way inclusive 8x 512kB (4MB) 8-way inclusive 8x 512kB (4MB) 8-way 8x 256kB (2MB) 16-way 10x 1MB (10MB) 16-way No changes to L2
L3 Caches (MB)
32MB 16-way exclusive 2x 16MB (32MB) 16-way exclusive 2x 8MB (16MB) 16-way 16MB 16-way 13.75MB 11-way Unified L3 but not larger
Mitigations for Vulnerabilities BTI/”Spectre”, SSB/”Spectre v4″ hardware BTI/”Spectre”, SSB/”Spectre v4″ hardware BTI/”Spectre”, SSB/”Spectre v4″ software/firmware RDCL/”Meltdown”, L1TF hardware, BTI/”Spectre”, MDS/”Zombieload”, software/firmware RDCL/”Meltdown” , L1TF, BTI/”Spectre”, MDS/”Zombieload”, all software/firmware No new fixes required… yet!
Microcode (MU)
MU-xxx MU-8F7100-11 MU-8F0802-04 MU-069E0C-9E MU-065504-49 The latest microcodes have been loaded.
SIMD Units 256-bit AVX/FMA3/AVX2 256-bit AVX/FMA3/AVX2 128bit AVX/FMA3/AVX2 256-bit AVX/FMA3/AVX2 512-bit AVX512 Same SIMD widths
Price/RRP (USD)
$450  [+36%] $330 [higher now due to demand] $330 [much cheaper now] $480 $990 Almost 40% price increase, cheaper than Intel

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 7 5800X  8C/16T (Vermeer) AMD Ryzen 7 3700X 8C/16T (Matisse) Intel i9 9900K 8C/16T (Coffeelake-R)
Intel i9 7900X 10C/20T (Skylake-X)
Comments
CPU Arithmetic Benchmark Native Dhrystone Integer (GIPS) 453 [+35%] 336 366 397 Zen3 starts strongly with 35% faster than Zen2 in this legacy integer benchmark.
CPU Arithmetic Benchmark Native Dhrystone Long (GIPS) 463 [+37%] 339 347 392 With a 64-bit integer workload still 37% improvement.
CPU Arithmetic Benchmark Native FP32 (Float) Whetstone (GFLOPS) 292 [+45%] 202 243 264 Floating-point performance is even better, 45% better than Zen2!
CPU Arithmetic Benchmark Native FP64 (Double) Whetstone (GFLOPS) 241 [+42%] 170 199 221 With FP64 nothing much changes again.
Zen3 improves by a pretty large 35-45% over Zen2 in legacy integer/floating-point benchmarks, a pretty remarkable improvement! Indeed now AMD has a comfortable performance lead over Intel that even 10-cores (+25%) cannot match.
BenchCpuMM Native Integer (Int32) Multi-Media (Mpix/s) 1,504 [+46%] 1,029 1,000 1,590* Zen3 is still 46% over Zen2 despite same width SIMD units.
BenchCpuMM Native Long (Int64) Multi-Media (Mpix/s) 597 [+63%] 366 435 548* With a 64-bit AVX2 integer vectorised workload, Zen3 is 63% faster.
BenchCpuMM Native Quad-Int (Int128) Multi-Media (Mpix/s) 117 [+39%] 84 83 125* This is a tough test using Long integers to emulate Int128 (now vectorised), Zen3 is still 39% faster.
BenchCpuMM Native Float/FP32 Multi-Media (Mpix/s) 1,547 [+45%] 1,065 998 1,930* In this floating-point AVX/FMA vectorised test, Zen3 is again 45% faster than Zen2.
BenchCpuMM Native Double/FP64 Multi-Media (Mpix/s) 915 [+48%] 618 573 1,120* Switching to FP64 SIMD code, Zen3 is 48% faster.
BenchCpuMM Native Quad-Float/FP128 Multi-Media (Mpix/s) 34.62 [+25%] 27.64 27 50* In this heavy algorithm using FP64 to mantissa extend FP128, Zen3 still manages to be 25% 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 25-60%, similar to what we’ve seen in legacy benchmarks. While SIMD workloads were Intel’s strenghts, Zen3 manages to beat even AVX512 10-core SKL-X with its 2x 512-bit wide SIMD units! It seems that there’s nothing stopping Zen3!

Note*: using AVX512 instead of AVX2/FMA.

BenchCrypt Crypto AES-256 (GB/s) 22*** [+22%] 18 16.7 34*** Zen3 (and SKL-X) support VAES but all CPUs are memory bandwidth bound – Zen3 is 22% faster.
BenchCrypt Crypto AES-128 (GB/s) *** 18.76 16.7 34*** What we saw with AES-256 just repeats with AES-128.
BenchCrypt Crypto SHA2-256 (GB/s) 26** [+29%] 20.21** 12.3 26* With SHA/HWA Zen3 similarly powers through hashing tests leaving Intel in the dust – and is still 29% faster than Zen2!
BenchCrypt Crypto SHA1 (GB/s) ** 20.41** 22.7 39* The less compute-intensive SHA1 does not change things due to acceleration.
BenchCrypt Crypto SHA2-512 (GB/s) ** ** 9 21*
While streaming tests (crypto/hashing) are memory bound, Zen3 still manages a decent 22-29% improvement over Zen2, again enough to beat even AVX512 10-core SKL-X with its 4-memory controllers! Not to mention that Zen3 can likely use even faster DDR4 memory (up to 4266Mt/s) that should further improve performance.

Note***: using VAES 256-bit (AVX2) or 512-bit (AVX512)

Note**: using SHA HWA not SIMD (e.g. AVX512, AVX2, AVX, etc.)

Note*: using AVX512 not AVX2.

BenchFinance Black-Scholes float/FP32 (MOPT/s) 285 340 The stadard financial algorithm.
BenchFinance Black-Scholes double/FP64 (MOPT/s) 326 [+42%] 229 254 289 Switching to FP64 code, Zen3 is 42% faster than Zen2.
BenchFinance Binomial float/FP32 (kOPT/s) 60.5 71 Binomial uses thread shared data thus stresses the cache & memory system;
BenchFinance Binomial double/FP64 (kOPT/s) 96 [+66%] 57.98 63.13 68 With FP64 code Zen3 is now 66% faster.
BenchFinance Monte-Carlo float/FP32 (kOPT/s) 218 249 Monte-Carlo also uses thread shared data but read-only thus reducing modify pressure on the caches;
BenchFinance Monte-Carlo double/FP64 (kOPT/s)  133 [+2.89x]
46.34 88 104 The cache improvements show here, Zen3 is almost 3x faster.
Ryzen always did well on non-SIMD floating-point algorithms and here it further cements its dominance: it is anywhere between 40 to 290% faster than Zen2 and absolutely demolishes Intel’s CPUs, even the 10-core SKL-X is left in the dust. It just shows that the good-old Skylake core is now pretty much obsolete and the slightly improved versions (e.g. CoffeeLake, CometLake, etc.) are just not enough to catch up with AMD. We really need IceLake/TigerLake here.
BenchScience SGEMM (GFLOPS) float/FP32 263 411 685* In this tough vectorised algorithm that is widely used (e.g. AI/ML).
BenchScience DGEMM (GFLOPS) double/FP64 360 [+42%] 193 231 259* With FP64 vectorised code, Zen3 is still 42% faster.
BenchScience SFFT (GFLOPS) float/FP32 22.78 20.5 39* FFT is also heavily vectorised but stresses the memory sub-system more.
BenchScience DFFT (GFLOPS) double/FP64 8.88 [-20%] 11.16 9.41 19.6* With FP64 code, Zen3 is 20% slower likely memory related.
BenchScience SNBODY (GFLOPS) float/FP32 612 483 592* N-Body simulation is vectorised but fewer memory accesses.
BenchScience DNBODY (GFLOPS) double/FP64 235 [+7%] 220 172 179* With FP64 precision ZEN2 is only 7% 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. GEMM is widely used in convolution (e.g. neural-networks AI/ML, image processing) and here Zen3 is much faster.

Note*: using AVX512 not AVX2/FMA3.

CPU Image Processing Blur (3×3) Filter (MPix/s) 3,528 [+72%] 2,049 2,590 4,440* In this vectorised integer workload Zen3 starts 70% faster than Zen2!
CPU Image Processing Sharpen (5×5) Filter (MPix/s) 1,423 [+50%] 950 1,000 2,000* Same algorithm but more shared data makes Zen3 50% faster.
CPU Image Processing Motion-Blur (7×7) Filter (MPix/s) 705 [+42%] 495 524 1,000* Again same algorithm but even more data shared still 50% faster
CPU Image Processing Edge Detection (2*5×5) Sobel Filter (MPix/s) 1,133 [+37%] 826 879 1,560* Different algorithm but still vectorised workload Zen3 is 37% faster.
CPU Image Processing Noise Removal (5×5) Median Filter (MPix/s) 121 [+35%] 89.68 77.8 217* Still vectorised code but Zen3 is “only” 35% faster.
CPU Image Processing Oil Painting Quantise Filter (MPix/s) 37.22 [+49%] 25.05 43 68 This test has always been tough for Ryzen but Zen3 still manages 50% improvement!
CPU Image Processing Diffusion Randomise (XorShift) Filter (MPix/s) 2,519 [43%] 1,763 4,000 3,440* With integer workload, Intel CPUs seem to do much better but Zen3 is still 43% faster than Zen2.
CPU Image Processing Marbling Perlin Noise 2D Filter (MPix/s) 446 [+39%] 321 610 777* 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 35-70% improvement, again more than enough to beat Intel’s competition though the AVX512 equipped SKL-X is still faster (and has more cores). In any case, it is unlikely Intel is happy to see Zen3 soundly beating CFL-R into dust.

Note*: using AVX512 not AVX2/FMA3.

BenchCpuAI SCNN Inference (Samp/s) float/FP32 97.61 [+11%] 88.2 46.79 54.19* Zen3 is 11% faster here, much faster than even SKL-X.
BenchCpuAI SCNN Training (Samp/s) float/FP32 20.28 [+82%] 11.12 6.46 9.16 Training just flies on Zen3, it is 80% faster.
BenchCpuAI SRNN Inference (Samp/s) float/FP32 149 [+2.34x] 63.85 75.03 71.81* Zen3 is over 2x faster than Zen2 and much faster than Intel.
BenchCpuAI SRNN Training (Samp/s) float/FP32 3.45 [-11%] 3.87 3.48 6.08* We have an outlier here as Zen3 is 11% slower.
Zen3 does extremely well overall with neural networks, with large speed increases (11-234%) 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) 11,400 [+39%] 8,200 8,150 13,090* Across all benchmarks, Zen3 is 40% faster than Zen2!
We previously saw Zen2 40% faster than Zen+ and now Zen3 is a similar 40% faster than Zen2: in effect for the same number of cores (8C/16T), Zen3 is 2x faster than Zen+. Simply astonishing!

Note*: using AVX512 note AVX2/FMA3.

Price/RRP (USD) $450 [+36%] $330 [higher now due to demand] $480 $990 Price increase is almost in-line with performance increase!
Price Efficiency (Perf. vs. Cost)  (Points/USD) 25.33 [+2%]
24.84 16.97 13.22 Small 2% efficiency increase due to large cost increase.
Due to the cost increase, price efficiency of 5800X has not increased much vs. old 3700X – but still much better value than competition. Due to lack of availability, the real price of the 3700X is much higher (+$50 to +$100) today making the new 5800X even more cost effective.

While Intel seems very price inefficient based on RRP, the CPU cost is generally less than RRP today and likely to be reduced further soon.

Power/TDP (W) 105 (135 PL2 turbo) [=]
105 (135 PL2 turbo) 95 (135 PL2 turbo) 140 (308 PL2 turbo) TDP has remained the same for Zen3.
Power Efficiency (Perf. vs. Power) (Points/W) 108.57 [+39%]
78.10 85.79 93.50 As TDP is the same, Zen3 is 39% more power efficient.
As TDP has remained constant, Zen3 has greatly improved in power efficiency (+39%) and has now overtaken Intel competition in this segment also. It is interesting that despite the great performance Zen2 (at least the 3700X) was not as power efficient as expected.

SiSoftware Official Ranker Scores

Final Thoughts / Conclusions

Executive Summary: Zen3 (5800X, 8-core) is ~25-40% faster than Zen2 (3700X, 8-core) across all kinds of algorithms but 36% more expensive. We have to give it 10/10 overall!

Despite no major architectural changes over Zen2 (except larger 8-core single CCX layout and thus unified L3 cache), Zen3 manages to be quite a bit faster across legacy and heavily vectorised SIMD algorithms: it beats the competition even with AVX512 and more cores (e.g. 10-core SKL-X). Even streaming algorithms (memory-bound) improve over 20%. We certainly did not expect performance to be this good.

In effect, it is like getting 50% more cores – 8-core Zen3 performs like a 12-core Zen2 (e.g. 3900X) – and thus even a 10-core 10900K cannot compete. Considering you can just “pop it” into an existing AM4 mainboard (requires a BIOS update to support it) it is a massive upgrade from say, original Zen1/Zen+.

If you can afford it – especially in these unprecedented times – it is a “no brainer” upgrade allowing older AM4-based computers to live many, many more years. You don’t really need PCIe4 and its modest improvement (and thus a 500-series board) – that would anyway require costly PCIe4 SSDs and costly GP-GPU upgrade.

Considering Zen2 is ~40% faster than Zen+ (never mind original Ryzen), Zen3 is in effect 2x faster than Zen+ – a 2x (96%) improvement over just 2 generations, while core counts remained the same (unlike Intel that just increased core counts). Also consider you can now get a 16-core/32-threads AM4 CPU (which originally only had 6-core option), it is like having a 32-core/64-thread Ryzen in the same AM4 slot – a 5.3x increase in overall performance.

About the only issue we can think of is the increase in cost; not only Ryzen 3000-series have gone up in price since launch, the RRP of series 5000 is $50 (or so) higher; it is likely once performance numbers come out and considering the reduced stock and increased demand for IT equipment – price is likely to sky-rocket. Better buy or pre-order one now to avoid disappointment!

Please see the other reviews on how the 6-core (5600X) and the top-end 16-core (5950X) variants perform:

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.

Tagged , , , , , . Bookmark the permalink.

Comments are closed.