uarch-bench for Ryzen 3000

By: Adrian (a.delete@this.acm.org), September 16, 2019 8:19 am
Room: Moderated Discussions

For whom might be curious, here are the results of running uarch-bench on a 3700X.


I have not checked if any of the tests depends on the memory speed (also interchip speed for Ryzen), which is slow for this system, i.e. 2666 MHz, like for Xeon E. The reason for the slow memory is that for now it is almost impossible to find ECC memory faster than supported by Intel Xeon, and I value ECC more than maximum memory speed.


If anyone is curious to see the results for another test or benchmark on Zen 2, I can run it if requested and if the test can run under Linux.



Welcome to uarch-bench (b75e0b5)
Supported CPU features: SSE3 PCLMULQDQ SSSE3 FMA CX16 SSE4_1 SSE4_2 MOVBE POPCNT AES AVX RDRND BMI1 AVX2 BMI2 PQE RDSEED ADX CLFLUSHOPT CLWB SHA
Pinned to CPU 0
Median CPU speed: 4.329 GHz
Running benchmarks groups using timer clock

** Running group basic : Basic Benchmarks **
Benchmark Cycles Nanos
Dependent add chain 1.00 0.23
Independent add chain 0.25 0.06
Dependent imul 64->128 2.99 0.69
Dependent imul 64->64 2.97 0.69
Independent imul 64->128 0.99 0.23
Same location stores 1.00 0.23
Disjoint location stores 1.00 0.23
Dependent push/pop chain 1.00 0.23
Independent push/pop chain 1.00 0.23
64-bit dependent div 1/1 = 1 13.87 3.20
64-bit dependent idiv 1/1 = 1 13.86 3.20
Simple addressing pointer chase 3.96 0.91
Complex addressing pointer chase 4.95 1.14
Finished in 513 ms (basic)

** Running group memory/load-parallel : Parallel loads from fixed-size regions **
Benchmark Cycles Nanos
16-KiB parallel load 0.54 0.12
24-KiB parallel load 0.53 0.12
30-KiB parallel load 0.54 0.12
31-KiB parallel load 0.54 0.12
32-KiB parallel load 0.54 0.12
33-KiB parallel load 0.55 0.13
34-KiB parallel load 0.62 0.14
35-KiB parallel load 0.61 0.14
40-KiB parallel load 1.01 0.23
48-KiB parallel load 2.00 0.46
56-KiB parallel load 2.00 0.46
64-KiB parallel load 2.00 0.46
80-KiB parallel load 2.00 0.46
96-KiB parallel load 2.00 0.46
112-KiB parallel load 2.00 0.46
128-KiB parallel load 2.00 0.46
196-KiB parallel load 2.00 0.46
252-KiB parallel load 2.00 0.46
256-KiB parallel load 2.00 0.46
260-KiB parallel load 0.56 0.13
384-KiB parallel load 2.00 0.46
512-KiB parallel load 2.69 0.62
1024-KiB parallel load 2.96 0.68
2048-KiB parallel load 2.09 0.48
Finished in 2422 ms (memory/load-parallel)

** Running group memory/store-parallel : Parallel stores to fixed-size regions **
Benchmark Cycles Nanos
16-KiB parallel store 1.00 0.23
24-KiB parallel store 1.00 0.23
30-KiB parallel store 1.00 0.23
31-KiB parallel store 1.01 0.23
32-KiB parallel store 1.01 0.23
33-KiB parallel store 1.22 0.28
34-KiB parallel store 1.22 0.28
35-KiB parallel store 1.15 0.26
40-KiB parallel store 1.71 0.40
48-KiB parallel store 2.14 0.50
56-KiB parallel store 2.18 0.50
64-KiB parallel store 2.18 0.50
80-KiB parallel store 2.16 0.50
96-KiB parallel store 2.06 0.48
112-KiB parallel store 2.05 0.47
128-KiB parallel store 2.05 0.47
196-KiB parallel store 2.03 0.47
252-KiB parallel store 2.03 0.47
256-KiB parallel store 2.04 0.47
260-KiB parallel store 1.00 0.23
384-KiB parallel store 2.02 0.47
512-KiB parallel store 2.28 0.53
1024-KiB parallel store 2.36 0.54
2048-KiB parallel store 2.26 0.52
Finished in 2678 ms (memory/store-parallel)

** Running group memory/prefetch-parallel : Parallel prefetches from fixed-size regions **
Benchmark Cycles Nanos
16-KiB parallel prefetcht0 0.50 0.12
16-KiB parallel prefetcht1 0.51 0.12
16-KiB parallel prefetcht2 0.51 0.12
16-KiB parallel prefetchnta 0.51 0.12
32-KiB parallel prefetcht0 0.51 0.12
32-KiB parallel prefetcht1 0.51 0.12
32-KiB parallel prefetcht2 0.51 0.12
32-KiB parallel prefetchnta 0.51 0.12
64-KiB parallel prefetcht0 0.51 0.12
64-KiB parallel prefetcht1 0.51 0.12
64-KiB parallel prefetcht2 0.51 0.12
64-KiB parallel prefetchnta 0.51 0.12
128-KiB parallel prefetcht0 0.51 0.12
128-KiB parallel prefetcht1 0.51 0.12
128-KiB parallel prefetcht2 0.51 0.12
128-KiB parallel prefetchnta 0.51 0.12
256-KiB parallel prefetcht0 0.51 0.12
256-KiB parallel prefetcht1 0.51 0.12
256-KiB parallel prefetcht2 0.51 0.12
256-KiB parallel prefetchnta 0.51 0.12
512-KiB parallel prefetcht0 1.01 0.23
512-KiB parallel prefetcht1 1.01 0.23
512-KiB parallel prefetcht2 1.01 0.23
512-KiB parallel prefetchnta 1.01 0.23
2048-KiB parallel prefetcht0 1.01 0.23
2048-KiB parallel prefetcht1 1.01 0.23
2048-KiB parallel prefetcht2 1.01 0.23
2048-KiB parallel prefetchnta 1.00 0.23
4096-KiB parallel prefetcht0 1.00 0.23
4096-KiB parallel prefetcht1 1.00 0.23
4096-KiB parallel prefetcht2 1.00 0.23
4096-KiB parallel prefetchnta 1.00 0.23
8192-KiB parallel prefetcht0 1.00 0.23
8192-KiB parallel prefetcht1 1.01 0.23
8192-KiB parallel prefetcht2 1.01 0.23
8192-KiB parallel prefetchnta 1.01 0.23
32768-KiB parallel prefetcht0 2.81 0.65
32768-KiB parallel prefetcht1 2.82 0.65
32768-KiB parallel prefetcht2 2.81 0.65
32768-KiB parallel prefetchnta 3.06 0.71
Finished in 515 ms (memory/prefetch-parallel)

** Running group memory/pointer-chase : Pointer-chasing **
Benchmark Cycles Nanos
Simple addressing chase, half diffpage 3.96 0.91
Simple addressing chase, different pages 3.97 0.92
Simple addressing chase with ALU op 4.97 1.15
load5 -> load4 -> alu 8.91 2.06
load4 -> load5 -> alu 8.91 2.06
8 parallel simple pointer chases 3.96 0.91
10 parallel complex pointer chases 4.98 1.15
10 parallel mixed pointer chases 5.00 1.16
Finished in 335 ms (memory/pointer-chase)

** Running group memory/load-serial : Serial loads from fixed-size regions **
Benchmark Cycles Nanos
16-KiB serial loads 4.00 0.92
24-KiB serial loads 4.00 0.92
30-KiB serial loads 3.98 0.92
31-KiB serial loads 3.98 0.92
32-KiB serial loads 3.98 0.92
33-KiB serial loads 5.54 1.28
34-KiB serial loads 5.67 1.31
35-KiB serial loads 6.06 1.40
40-KiB serial loads 7.39 1.71
48-KiB serial loads 8.45 1.95
56-KiB serial loads 9.20 2.13
64-KiB serial loads 9.56 2.21
80-KiB serial loads 10.16 2.35
96-KiB serial loads 10.62 2.45
112-KiB serial loads 10.89 2.52
128-KiB serial loads 11.07 2.56
196-KiB serial loads 11.47 2.65
252-KiB serial loads 11.61 2.68
256-KiB serial loads 11.68 2.70
260-KiB serial loads 11.74 2.71
384-KiB serial loads 14.09 3.25
512-KiB serial loads 23.42 5.41
1024-KiB serial loads 35.73 8.25
2048-KiB serial loads 40.33 9.32
Finished in 230 ms (memory/load-serial)

** Running group bmi : BMI false-dependency tests **
Benchmark Cycles Nanos
dest-dependent tzcnt 0.50 0.12
dest-dependent lzcnt 0.25 0.06
dest-dependent popcnt 0.25 0.06
Finished in 10 ms (bmi)

** Running group studies/vzeroall : VZEROALL weirdness **
Benchmark Cycles Nanos
vpaddq zmm0, zmm0, zmm0 Skipped because hardware doesn't support required features: [AVX512F]
vpaddq zmm0, zmm1, zmm0 Skipped because hardware doesn't support required features: [AVX512F]
vpaddq zmm0, zmm16, zmm0 Skipped because hardware doesn't support required features: [AVX512F]
vpxor zmm16; vpaddq zmm0, zmm16, zmm0 Skipped because hardware doesn't support required features: [AVX512F]
vpaddq ymm0, ymm0, ymm0 1.00 0.23
vpaddq ymm0, ymm1, ymm0 1.00 0.23
vpaddq xmm0, xmm0, xmm0 1.00 0.23
vpaddq xmm0, xmm1, xmm0 1.00 0.23
paddq xmm0, xmm0 1.00 0.23
paddq xmm0, xmm1 1.00 0.23
Finished in 48 ms (studies/vzeroall)


 Next Post in Thread >
TopicPosted ByDate
uarch-bench for Ryzen 3000Adrian2019/09/16 08:19 AM
  uarch-bench for Ryzen 3000anonymous22019/09/16 09:08 AM
    uarch-bench for Ryzen 3000Adrian2019/09/16 11:29 AM
      X570 FanDavid Hess2019/09/18 08:10 AM
        X570 FanPeter E. Fry2019/09/19 07:30 AM
          X570 FanDavid Hess2019/09/19 03:52 PM
      idle poweranonymous22019/09/18 04:29 PM
        idle powerAdrian2019/09/19 02:17 PM
  uarch-bench for Ryzen 3000Gabriele Svelto2019/09/16 11:16 AM
    overclocking ECC RAMhobold2019/09/16 11:55 AM
      overclocking ECC RAMGabriele Svelto2019/09/16 10:10 PM
        overclocking ECC RAMMontaray Jack2019/09/17 10:34 AM
      overclocking ECC RAMDavid Hess2019/09/18 08:14 AM
        overclocking ECC RAMGabriele Svelto2019/09/18 09:27 AM
          overclocking ECC RAMDavid Hess2019/09/18 12:03 PM
            overclocking ECC RAMEmil Briggs2019/09/19 04:24 AM
              overclocking ECC RAMDavid Hess2019/09/19 03:58 PM
                overclocking ECC RAMHoward Chu2019/09/20 01:59 PM
              overclocking ECC RAMMontaray Jack2019/09/21 01:33 AM
                overclocking ECC RAMMontaray Jack2019/09/21 01:59 AM
                overclocking ECC RAMMontaray Jack2019/09/21 03:43 AM
                  overclocking ECC RAMMichael S2019/09/21 10:20 AM
                    overclocking ECC RAMMontaray Jack2019/09/22 02:09 AM
  uarch-bench for Ryzen 3000Travis2019/09/17 07:14 PM
    uarch-bench for Ryzen 3000Adrian2019/09/18 05:54 AM
      uarch-bench for Ryzen 3000Adrian2019/09/18 06:10 AM
        uarch-bench for Ryzen 3000Travis2019/09/18 04:10 PM
          complete results for Ryzen 3000Adrian2019/09/22 11:00 AM
        uarch-bench for Ryzen 3000Travis Downs2019/10/01 07:34 AM
          uarch-bench for Ryzen 3000Travis Downs2019/10/01 08:32 AM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell purple?