SiSoftware was one of the few companies to receive early samples of the 2nd Generation Ryzen processors. They tested both Ryzen 7 2700X and Ryzen 5 2600 processors, and inadvertently leaked their benchmark results and performance findings.

They have since removed their leaked performance evaluation of the Ryzen 7 2700X and Ryzen 5 2600, but the Internet never forgets. So we present to you – the leaked benchmark results and findings of the AMD Ryzen 7 2700X and Ryzen 5 2600 processors!

Note : The SiSoftware team only compared the Ryzen 7 2700X and Ryzen 5 2600 processors against the Ryzen 7 1700X processor, and the Intel Core i7-6700K processor, which is two generations old. Still, it gives us an advanced performance preview of the Ryzen 7 2700X and Ryzen 5 2600 processors.

• AMD Ryzen 7 2700X + Ryzen 5 2600 CPU Native Performance
• AMD Ryzen 7 2700X + Ryzen 5 2600 Software VM Performance
• AMD Ryzen 7 2700X + Ryzen 5 2600 Memory Performance
• Final Thoughts & Conclusion on the Ryzen 7 2700X + Ryzen 5 2600

AMD Ryzen 7 2700X and Ryzen 5 2600 CPU Native Performance

by SiSoftware

We are testing native arithmetic, SIMD and cryptography performance using the highest performing instruction sets (AVX2, AVX, etc.). Ryzen supports all modern instruction sets including AVX2, FMA3 and even more like SHA HWA (supported by Intel’s Atom only) but has dropped all AMD’s variations like FMA4 and XOP likely due to low usage.

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.

Dhrystone Integer : Right off Ryzen2 is 8% faster than Ryzen1, let’s hope it does better. Even 2600 beats the i7 easily.

Dhrystone Long : With a 64-bit integer workload – we finally get into gear, Ryzen2 is 12% faster than its old brother.

FP32 (Float) Whetstone : Even in this floating-point test, Ryzen2 is again 12% faster. All AMD CPUs beat the i7 into dust.

FP64 (Double) Whetstone : With FP64 nothing much changes, Ryzen2 is still 11% faster.

From integer workloads in Dhyrstone to floating-point workloads in Whestone, Ryzen2 is about 10% faster than Ryzen 1: this is exactly in line with the speed increase (9-11%) but if you were expecting more you may be a tiny bit disappointed.

Integer (Int32) Multi-Media : In this vectorised AVX2 integer test Ryzen2 starts to pull ahead and is 16% faster than Ryzen1; perhaps some of the arch improvements benefit SIMD vectorised workloads.

Long (Int64) Multi-Media : With a 64-bit AVX2 integer vectorised workload, Ryzen2 drops to just 10% but still in line with speed increase.

Quad-Int (Int128) Multi-Media : This is a tough test using Long integers to emulate Int128 without SIMD; here Ryzen2 drops to just 7% faster than Ryzen1 but still a decent improvement.

Float/FP32 Multi-Media : In this floating-point AVX/FMA vectorised test, Ryzen2 is the standard 11% faster than Ryzen1.

Double/FP64 Multi-Media : Switching to FP64 SIMD code, again Ryzen2 is just the standard 11% faster than Ryzen1.

Quad-Float/FP128 Multi-Media : In this heavy algorithm using FP64 to mantissa extend FP128 but not vectorised – Ryzen2 manages to pull ahead further and is 15% faster.

In vectorised AVX2/FMA code we see a similar story with 10% average improvement (7-15%). It seems the SIMD units are unchanged. In any case the i7 is left in the dust.

Crypto AES-256 : With AES HWA support all CPUs are memory bandwidth bound; as we’re testing Ryzen2 running at the same memory speed/timings there is still a very small improvement of 1%. But its advantage is that the memory controller is rated for 2933Mt/s operation (vs. 2533) thus with faster memory it could run considerably faster.

Crypto AES-128 : What we saw with AES-256 just repeats with AES-128; Ryzen2 is marginally faster but the improvement is there.

Crypto SHA2-256 : With SHA HWA Ryzen2 similarly powers through hashing tests leaving Intel in the dust; SHA is still memory bound but with just one (1) buffer it has larger headroom. Thus Ryzen2 can use its speed advantage and be 12% faster – impressive.

Crypto SHA1 : Ryzen also accelerates the soon-to-be-defunct SHA1 and here it is even faster – 14% faster than Ryzen1.

Crypto SHA2-512 : SHA2-512 is not accelerated by SHA HWA (version 1) thus Ryzen has to use the same vectorised AVX2 code path – it still is 12% faster than Ryzen1 but still loses to the i7. Those SIMD units are tough to beat.

In memory bandwidth bound algorithms, Ryzen2 will have to be used with faster memory (up to 2933 Mt/s officially) in order to significantly beat its older Ryzen 1 brother. Otherwise there is only a tiny 1% improvement.

Black-Scholes float/FP32 : In this non-vectorised test we see Ryzen2 is the standard 11% faster than Ryzen1.

Black-Scholes double/FP64 : Switching to FP64 code, nothing changes, Ryzen2 is still 11% faster.

Binomial float/FP32 : Binomial uses thread shared data thus stresses the cache & memory system; here the arch(itecture) improvements do show, Ryzen2 23% faster – 2x more than expected. Not to mention 3x (three times) faster than the i7.

Binomial double/FP64 : With FP64 code Ryzen2 is now even faster – 28% faster than Ryzen1 not to mention 2x faster than the i7. Indeed it seems there improvements to the cache and memory system.

Monte-Carlo float/FP32 : Monte-Carlo also uses thread shared data but read-only thus reducing modify pressure on the caches; Ryzen2 does not seem to be able to reproduce its previous gain and is just the standard 11% faster.

Monte-Carlo double/FP64 : Switching to FP64 nothing much changes, Ryzen2 is 10% faster.

Ryzen 1 does very well in these algorithms, but Ryzen2 does even better – especially when thread-local data is involved managing 23-28% improvement. For financial workloads Intel does not seem to have a chance anymore – Ryzen is impossible to beat.

SGEMM : In this tough vectorised AVX2/FMA algorithm Ryzen2 is still “just” the 10% faster than older Ryzen1 – but it finally manages to beat the i7.

DGEMM : With FP64 vectorised code, Ryzen2 only manages to be 4% faster. It seems the memory is holding it back thus faster memory would allow it to do much better.

SFFT : FFT is also heavily vectorised (x4 AVX/FMA) but stresses the memory sub-system more; Ryzen2 is just 4% faster again and is still 1/2x the speed of the i7. Again it seems faster memory would help.

DFFT : With FP64 code, Ryzen2’s improvement reduces to just 1% over Ryzen1 and again slower than the i7.

SNBODY : N-Body simulation is vectorised but many memory accesses to shared data and Ryzen2 gets back to 12% improvement over Ryzen1. This allows it to finally overtake the i7.

DNBODY : With FP64 code nothing much changes, Ryzen2 is still 13% faster.

With highly vectorised SIMD code Ryzen2 still improves by the standard 10-12% but in memory-heavy code it needs to run at higher memory speed to significantly overtake Ryzen 1. But it allows it to beat the i7 in more algorithms.

Blur (3×3) Filter : In this vectorised integer AVX2 workload Ryzen2 is 11% faster allowing it to soundly beat the i7.

Sharpen (5×5) Filter : Same algorithm but more shared data does not change things for Ryzen2. Only the i7 falls behind.

Motion-Blur (7×7) Filter : Again same algorithm but even more data shared does not change anything, but now the i7 is so far behind Ryzen2 is 50% faster. Incredible.

Edge Detection (2*5×5) Sobel Filter : Different algorithm but still AVX2 vectorised workload still changes nothing – Ryzen2 is 11% faster.

Noise Removal (5×5) Median Filter : Still AVX2 vectorised code and still nothing changes; the i7 falls even further behind with Ryzen2 2x (two times) as fast.

Oil Painting Quantise Filter : Again we see Ryzen2 11% faster than the older Ryzen1 and pulling away from the i7.

Diffusion Randomise (XorShift) Filter : Here Ryzen2 is just 8% faster than Ryzen1 but strangely it’s not enough to beat the i7. Those SIMD units are way fast.

Marbling Perlin Noise 2D Filter : In this final test, Ryzen2 returns to being 11% faster and again strangely not enough to beat the i7.

With all the modern instruction sets supported (AVX2, FMA, AES and SHA HWA) Ryzen2 does extremely well in all workloads – but it generally improves only by the 11% as per clock speed increase, except in some cases which seem to show improvements in the cache and memory system (which we have not tested yet).

Next Page > AMD Ryzen 7 2700X + Ryzen 5 2600 Software VM Performance

Support Tech ARP!

If you like our work, you can help support our work by visiting our sponsors, participating in the Tech ARP Forums, or even donating to our fund. Any help you can render is greatly appreciated!

AMD Ryzen 7 2700X and Ryzen 5 2600 Software VM Performance

We are testing arithmetic and vectorised performance of software virtual machines (SVM), i.e. Java and .Net. With operating systems – like Windows 10 – favouring SVM applications over “legacy” native, the performance of .Net CLR (and Java JVM) has become far more important.

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

Environment: Windows 10 x64, latest drivers. .Net 4.7.x (RyuJit), Java 1.9.x. Turbo / Boost was enabled on all configurations.

.Net Dhrystone Integer : .Net CLR integer performance starts off OK with Ryzen2 just 8% faster than Ryzen1 but now almost 3x (three times) faster than i7.

.Net Dhrystone Long : Ryzen seems to favour 64-bit integer workloads, with Ryzen2 20% faster a lot higher than expected.

.Net Whetstone float/FP32 : Floating-Point CLR performance was pretty spectacular with Ryzen already, but Ryzen2 is 15% than Ryzen1 still.

.Net Whetstone double/FP64 : FP64 performance is also great (CLR seems to promote FP32 to FP64 anyway) with Ryzen2 even faster by 20%.

Ryzen1’s performance in .Net was pretty incredible but Ryzen2 is even faster – even faster than expected by mere clock speed increase. There is only one game in town now for .Net applications.

.Net Integer Vectorised/Multi-Media : Just as we saw with Dhrystone, this integer workload sees a 9% improvement for Ryzen2 which makes it 2x faster than the i7.

.Net Long Vectorised/Multi-Media : With 64-bit integer workload we see a similar story – Ryzen2 is 8% faster and again 2x faster than the i7.

.Net Float/FP32 Vectorised/Multi-Media : Here we make use of RyuJit’s support for SIMD vectors thus running AVX/FMA code; Ryzen2 is 11% faster but still almost 2x faster than i7 despite its fast SIMD units.

.Net Double/FP64 Vectorised/Multi-Media : Switching to FP64 SIMD vector code – still running AVX/FMA – Ryzen2 is still 12% faster. i7 is truly left in the dust 1/4x the speed.

Ryzen2 is the usual 9-12% faster than Ryzen 1 here but it means that even RyuJit’s SIMD support cannot save Intel’s i7 – it would take 2x as many cores (not 50%) to beat Ryzen2.

Java Dhrystone Integer : We start JVM integer performance with the usual 12% gain over Ryzen1.

Java Dhrystone Long : Nothing much changes with 64-bit integer workload, we have Ryzen2 12% faster.

Java Whetstone float/FP32 : With a floating-point workload Ryzen2 performance improvement is 13%.

Java Whetstone double/FP64 : With FP64 workload Ryzen2 is just 7% faster but still welcome

Java performance improves by the expected amount 7-13% on Ryzen2 and allows it to completely dominate the i7.

Java Integer Vectorised/Multi-Media : Oracle’s JVM does not yet support native vector to SIMD translation like .Net’s CLR but here Ryzen2 manages a 15% lead over Ryzen1.

Java Long Vectorised/Multi-Media : With 64-bit vectorised workload Ryzen2 (similar to .Net) increases its lead by 24%.

Java Float/FP32 Vectorised/Multi-Media : Switching to floating-point we return to the usual 14% speed improvement.

Java Double/FP64 Vectorised/Multi-Media : With FP64 workload Ryzen2’s lead somewhat unexplicably drops to 1%.

Java’s lack of vectorised primitives to allow the JVM to use SIMD instruction sets (aka SSE2, AVX/FMA) gives Ryzen2 free reign to dominate all the tests, be they integer or floating-point. It is pretty incredible that neither Intel CPU can come close to its performance.

Software VM Performance Summary

Ryzen1 dominated the .Net and Java benchmarks – but now Ryzen2 extends that dominance out-of-reach. It would take a very much improved run-time or Intel CPU to get anywhere close. For .Net and Java code, Ryzen is the CPU to get!

Next Page > AMD Ryzen 7 2700X + Ryzen 5 2600 Memory Performance

Support Tech ARP!

If you like our work, you can help support our work by visiting our sponsors, participating in the Tech ARP Forums, or even donating to our fund. Any help you can render is greatly appreciated!

AMD Ryzen 7 2700X and Ryzen 5 2600 Memory Performance

by SiSoftware

Core Topology & Testing

Cores on Ryzen are grouped in blocks (CCX or compute units) each with its own 8MB L3 cache – but connected via a 256-bit bus running at memory controller clock. This is better than older designs like Intel Core 2 Quad or Pentium D which were effectively 2 CPU dies on the same socket – but not as good as a unified design where all cores are part of the same unit.

Running algorithms that require data to be shared between threads – e.g. producer/consumer – scheduling those threads on the same CCX would ensure lower latencies and higher bandwidth which we will test with presently.

We have thus modified Sandra’s ‘CPU Multi-Core Efficiency Benchmark‘ to report the latencies of each producer/consumer unit combination (e.g. same core, same CCX, different CCX) as well as providing different matching algorithms when selecting the producer/consumer units: best match (lowest latency), worst match (highest latency) thus allowing us to test inter-CCX bandwidth also. We hope users and reviewers alike will find the new features useful!

Native Performance

We are testing native arithmetic, SIMD and cryptography performance using the highest performing instruction sets (AVX2, AVX, etc.). Ryzen supports all modern instruction sets including AVX2, FMA3 and even more.

Results Interpretation: Higher rate values (GOPS, MB/s, etc.) mean better performance. Lower latencies (ns, ms, 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.

Total Inter-Core Bandwidth – Best : Ryzen2 manages 15% higher bandwidth between its cores, slightly better than just 11% clock increase – signalling some improvements under the hood.

Total Inter-Core Bandwidth – Worst : In worst-case pairs on Ryzen must go across CCXes – and with this link running at the same clock (1200 MHz) on Ryzen2 we can only manage a 2% increase in bandwidth. This is why faster memory is needed.

Inter-Unit Latency – Same Core : Within the same core (sharing L1D/L2), Ryzen2 manages a 13% reduction in latency, again better than just clock speed increase.

Inter-Unit Latency – Same Compute Unit : Within the same compute unit (sharing L3), the latency decreased by 7% on Ryzen2 thus L3 seems to have improved also.

Inter-Unit Latency – Different Compute Unit : Going inter-CCX we still see a 6% reduction in latency on Ryzen2 – with the CCX link at the same speed – a welcome surprise.

The multiple CCX design still presents some challenges to programmers requiring threads to be carefully scheduled – but we see a decent 6-7% reduction in L3/CCX latencies on Ryzen2 even when running at the same clock as Ryzen1.

Aggregated L1D Bandwidth : Right off we see a 18% bandwidth increase – almost 2x higher (than the 11% clock increase) – thus some improvements have been made to the cache system. It allows Ryzen2 to finally beat the i7 with its wide L1 data paths (512-bit) though with twice as many (8 vs 4).

Aggregated L2 Bandwidth : We see a huge 32% increase in L2 cache bandwidth – almost 3x clock increase (the 11%) suggesting the L2 caches have been improved also. Ryzen2 has thus 2x the L2 bandwidth of i7 though with 2x more caches (8 vs 4).

Aggregated L3 Bandwidth : The bandwidth of the L3 caches has also increased by 19% (2x clock increase) though we see the 6-core 2600 doing better (398 vs 339) likely due to less threads competing for the same L3 caches. Ryzen2 L3 caches are not just 2x bigger than Intel but also 2x more bandwidth.

Aggregated Memory : With the same memory clock, Ryzen2 does still manage a small 2% improvement – signalling memory controller improvements. We also see Ryzen’s memory controller at 2400 having better bandwidth than Intel at 2533MHz.

We see big improvements on Ryzen2 for all caches L1D/L2/L3 of 20-30% – more than just raw clock increase (11%) – so AMD has indeed made improvements – which to be fair needed to be done. The memory controller is also a bit more efficient (2%) though it can run at higher clocks – hopefully fast DDR4 memory will become more affordable.

Data In-Page Random Latency : In-page latency has decreased by a noticeable 6% on Ryzen2 – we see 5 clocks reduction for L2 and 4 for L3 a welcome improvement. But still a way to go to catch Intel which has 1/3x (three times less) latency.

Data Full Random Latency : Out-of-page latencies have also been reduced by 8% on Ryzen2 (same memory) and we see the same 5 and 4 clock reduction for L2 and L3 (on both 2700X and 2600 thus no fluke). Again these are welcome but still have a way to go to catch Intel.

Data Sequential Latency : Ryzen’s prefetchers are working well with sequential access pattern latency and we see a 8% latency drop for Ryzen2.

Ryzen1’s issue was high memory latencies (in-page/full random) and Ryzen2 has reduced them all by 6-8%. While it is a good improvement, they are still pretty high compared to Intel’s thus more work needs to be done here.

Code In-Page Random Latency : Code latencies were not a problem on Ryzen1 but we still see a welcome reduction of 9% on Ryzen2.

Code Full Random Latency : Out-of-page latency also sees a 9% decrease on Ryzen2 but somewhat surprisingly a 1-2 clock increase.

Code Sequential Latency : Ryzen’s prefetchers are working well with sequential access pattern latency and we see a 8% reduction on Ryzen2.

While code access latencies were not a problem on Ryzen1 and they also see a 8% improvement on Ryzen2 which is welcome.

Memory Update Transactional : Ryzen2 is 10% faster than Ryzen1 but naturally without HLE support it cannot match the i7. But with Intel disabling HLE on all but top-end CPUs AMD does not have much to worry.

Memory Update Record Only : With only record updates we still see an 11% increase.

Ryzen2 brings nice updates – good bandwidth increases to all caches L1D/L2/L3 and also well-needed latency reduction for data (and code) accesses. Yes, there is still work to be done to bring the latencies down further – but it may be just enough to beat Intel to 2nd place for a good while.

At the high-end, ThreadRipper will likely benefit most as it’s going against many-core SKL-X AVX512-enabled competitor and we cannot wait to test those.

Final Thoughts & Conclusion On The Ryzen 7 2700X and Ryzen 5 2600

Suggested Reading

• Everything On The 2nd Gen Ryzen (Pinnacle Ridge) CPUs!
• AMD Ryzen 7 2700X Octa-Core Processor Preview
• AMD Ryzen 5 2600X Hexa-Core Processor Preview
• 2nd Gen Ryzen Price List + Availability in Malaysia and US!
• The AMD Ryzen Gen 2 Reviewer’s Kit Revealed!
• Cheaper Ryzen CPUs When Ryzen 2 Launches?
• The 2018 AMD Ryzen Price Cut Details Examined!
• The AMD Raven Ridge Desktop APUs – Everything You Need To Know!
• AMD Ryzen 5 2400G with Radeon RX Vega 11 Graphics
• AMD Ryzen 3 2200G with Radeon Vega 8 Graphics[adrotate group=”2″]
• Thank The Ryzen Effect For Better Intel Processors!
• The 8th Gen Intel Core Desktop CPU Tech Report
• The Intel Core i7-8700K Hexa-Core Processor Review
• Everything You Need To Know About The Intel Coffee Lake CPUs!
• All You Need To Know About AMD Ryzen Threadripper!
• The AMD Ryzen PRO Processor Tech Report
• The AMD Ryzen 7 1800X Octa-Core Processor Review
• The AMD Ryzen 5 1500X Quad-Core Processor Review
• The AMD Ryzen 3 1300X Quad-Core Processor Review

Go Back To > First Page | Reviews | Home

Support Tech ARP!

If you like our work, you can help support our work by visiting our sponsors, participating in the Tech ARP Forums, or even donating to our fund. Any help you can render is greatly appreciated!