AMD Naples S2 M4 C24
- Processor: AMD EPYC 7451 24-Core Processor
- Base frequency: 2.3 GHz
- Number of sockets: 2
- Number of memory domains per socket: 4
- Memory domain specs: 2-channel DDR4-2666
- Number of cores per socket: 24
- Number of HWThreads per core: 2
- MachineState output: NA
+----------+---------------------------------+
| Compiler | icc (ICC) |
|----------|---------------------------------|
| Version | icc (ICC) 19.0.5.281 20190815 |
+----------+---------------------------------+
Optimizing flags: -fast -xHost -qopt-streaming-stores=always -std=c99 -ffreestanding -qopenmp
All results are in GB/s.
Summary results:
+---------------------------------------------+
| Single core | 38.65 (Init) |
| Memory domain | 40.45 (Init with 6 cores) |
| Socket | 160.63 (Init with 6 cores) |
| Node | 317.49 (Init with 6 cores) |
+---------------------------------------------+
Results for scaling within a memory domain:
#nt Init Sum Copy Update Triad Daxpy STriad SDaxpy
1 38.65 23.48 26.34 27.30 29.46 27.77 29.20 29.09
2 38.49 35.49 34.12 34.78 31.29 30.27 30.40 29.98
3 38.38 35.86 34.51 34.58 30.12 29.76 29.59 28.96
4 39.56 37.98 35.38 35.60 33.10 32.89 32.30 31.92
5 40.13 38.41 35.23 35.49 34.01 34.06 33.12 32.96
6 40.45 38.98 35.03 35.26 34.64 34.70 33.43 33.22
Results for scaling across memory domains. Shown are the results for the number of memory domains used (nm) with columns number of cores used per memory domain.
Init:
#nm 1 2 3 4 5 6 7 8
1 38.65 76.89 114.54 150.96 185.07 220.67 260.64 295.78
2 38.49 76.79 114.77 152.12 185.33 221.06 264.42 299.13
3 38.38 76.62 114.76 152.51 186.67 222.99 264.94 299.36
4 39.56 78.93 118.05 157.02 196.24 233.92 272.55 310.88
5 40.13 80.15 119.79 159.25 198.60 236.80 276.37 315.82
6 40.45 80.64 120.68 160.63 199.16 238.21 278.39 317.49
Sum:
#nm 1 2 3 4 5 6 7 8
1 23.48 46.88 70.11 93.42 115.71 138.80 158.95 185.44
2 35.49 70.86 106.01 141.76 174.50 208.23 237.83 279.27
3 35.86 71.60 107.24 142.71 176.12 211.02 240.63 280.89
4 37.98 75.89 113.76 151.38 188.63 225.98 263.86 300.01
5 38.41 76.74 114.77 153.24 190.11 228.10 266.18 302.58
6 38.98 77.89 116.55 155.48 192.77 229.02 267.71 308.38
Copy
#nm 1 2 3 4 5 6 7 8
1 26.34 52.44 78.62 104.71 129.79 154.46 180.11 206.80
2 34.12 67.82 102.17 136.43 168.68 201.78 233.11 269.69
3 34.51 68.98 103.17 137.74 170.48 204.02 234.90 270.76
4 35.38 70.60 105.42 140.83 174.83 209.18 244.67 278.18
5 35.23 70.32 104.97 139.65 173.55 206.91 241.62 274.40
6 35.03 70.17 104.66 139.72 173.10 206.49 239.91 272.06
Update
#nm 1 2 3 4 5 6 7 8
1 27.30 54.74 82.10 109.78 137.24 165.05 190.73 221.04
2 34.78 69.76 104.65 139.92 174.97 210.49 242.77 282.38
3 34.58 69.34 103.95 139.03 173.68 208.83 241.53 280.22
4 35.60 71.51 107.36 143.61 180.04 216.34 252.43 290.19
5 35.49 71.27 106.98 143.43 179.47 216.03 252.87 289.36
6 35.26 71.15 107.73 144.84 181.30 218.66 256.23 296.29
Triad
#nm 1 2 3 4 5 6 7 8
1 29.46 56.00 88.26 111.51 139.45 164.66 194.05 222.24
2 31.29 62.82 94.25 125.40 155.95 186.92 219.33 249.71
3 30.12 59.92 90.01 119.81 148.84 178.97 209.21 240.05
4 33.10 66.18 99.10 131.81 164.29 197.51 230.27 263.82
5 34.01 68.09 101.98 136.13 168.57 202.88 236.57 270.75
6 34.64 69.36 103.79 138.29 171.83 206.59 241.00 275.65
Memory bandwidth scaling within one memory domain:
The following plots illustrate the the performance scaling over multiple memory domains using different number of cores per memory domain.
Memory bandwidth scaling across memory domains for init:
Memory bandwidth scaling across memory domains for sum
Memory bandwidth scaling across memory domains for copy
Memory bandwidth scaling across memory domains for Triad
