AMD Genoa S2 M4 C96
- Processor: AMD EPYC 9654 96-Core Processor
- Base frequency: 2.4 GHz
- Number of sockets: 2
- Number of memory domains per socket: 4
- Number of cores per socket: 96
- Number of HWThreads per core: 2
- MachineState output: [html] [json]
+----------+-----------+
| Compiler | icc (ICC) |
|----------|-----------|
| Version | icc (ICC) 2021.9.0 20230302 |
+----------+-----------+
Optimizing flags: -fast -xCORE-AVX512 -qopt-streaming-stores=always -std=c99 -ffreestanding -qopenmp
All results are in GB/s.
Summary results:
+---------------------------------+
| Single core | 48.89 (Update) |
| Memory domain | 96.95 (Sum with 24 cores) |
| Socket | 383.03 (Sum with 24 cores) |
| Node | 757.13 (Sum with 24 cores) |
+---------------------------------+
Results for scaling within a memory domain:
#nt Init Sum Copy Update Triad Daxpy STriad SDaxpy
1 28.30 35.06 47.77 48.89 45.94 44.72 43.05 39.91
2 28.74 37.81 53.06 53.27 47.21 46.54 44.13 42.58
3 28.74 37.13 52.26 52.58 47.49 46.58 45.16 43.50
4 28.74 38.04 52.54 53.00 48.62 46.56 44.94 43.35
5 28.75 38.14 52.79 53.64 48.30 46.60 44.86 43.66
6 28.74 38.00 52.20 53.46 49.32 47.24 44.99 44.12
7 28.75 38.25 52.88 53.27 48.81 47.86 44.96 43.88
8 28.75 38.45 52.46 53.41 49.16 47.42 45.76 44.78
9 32.34 43.18 58.54 58.25 54.38 52.93 50.38 48.74
10 35.93 46.97 62.50 62.35 59.56 57.59 56.38 53.89
11 39.52 52.73 67.53 66.63 63.30 61.38 61.67 59.18
12 43.12 57.51 71.33 70.55 68.02 65.95 65.22 62.83
13 46.71 60.88 75.74 74.21 73.36 71.02 71.12 67.98
14 50.30 67.17 79.38 77.36 76.84 74.47 74.41 72.28
15 53.90 72.02 83.31 80.66 81.10 78.65 78.99 76.48
16 57.48 76.96 86.74 84.03 85.28 82.72 82.78 80.53
17 61.02 80.24 87.29 84.98 86.48 83.91 83.83 82.14
18 64.55 82.57 87.76 85.99 87.44 84.85 85.20 82.86
19 68.12 83.97 88.14 87.08 87.86 85.32 85.95 83.45
20 71.70 86.22 88.23 87.81 88.70 86.13 86.78 84.32
21 75.26 88.85 88.15 88.29 89.55 87.00 87.86 85.44
22 78.82 92.26 88.38 89.02 90.22 87.87 88.40 86.22
23 82.35 94.45 88.40 89.83 90.75 88.74 89.37 86.88
24 85.82 96.95 89.82 90.88 91.79 89.77 90.46 88.00
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 28.30 57.43 86.16 114.89 143.60 172.31 201.06 229.78
2 28.74 57.46 86.19 114.92 143.63 172.32 201.12 229.77
3 28.74 57.47 86.21 114.91 143.65 172.30 201.05 229.79
4 28.74 57.48 86.21 114.95 143.66 172.35 201.09 229.73
5 28.75 57.48 86.22 114.92 143.65 172.40 201.06 229.74
6 28.74 57.49 86.21 114.93 143.63 172.38 201.13 229.81
7 28.75 57.49 86.22 114.95 143.68 172.40 201.12 229.80
8 28.75 57.48 86.22 114.95 143.65 172.34 201.08 229.84
9 32.34 64.66 96.99 129.31 161.60 193.92 226.26 258.54
10 35.93 71.85 107.76 143.67 179.59 215.49 251.32 287.26
11 39.52 79.02 118.53 158.02 197.45 236.91 276.46 315.90
12 43.12 86.22 129.32 172.41 215.46 258.52 301.60 344.60
13 46.71 93.40 140.10 186.69 233.38 280.09 326.72 373.32
14 50.30 100.58 150.87 201.11 251.37 301.60 351.77 402.03
15 53.90 107.77 161.64 215.49 269.34 323.14 376.82 430.78
16 57.48 114.96 172.42 229.79 287.17 344.70 402.00 459.41
17 61.02 122.04 182.94 243.89 304.87 365.79 426.79 487.48
18 64.55 129.12 193.49 258.12 322.45 386.96 451.28 515.85
19 68.12 136.17 204.23 272.28 340.32 408.18 476.20 544.10
20 71.70 143.33 214.98 286.46 357.92 429.50 501.07 572.63
21 75.26 150.38 225.59 300.70 375.74 450.80 525.76 600.73
22 78.82 157.51 236.15 314.83 393.39 471.54 550.01 628.71
23 82.35 164.64 246.84 329.04 411.19 493.10 575.07 656.94
24 85.82 171.70 257.57 342.96 428.70 514.07 599.70 685.35
Sum:
#nm 1 2 3 4 5 6 7 8
1 35.06 70.03 105.65 141.73 178.14 214.76 251.48 288.64
2 37.81 74.87 112.40 150.15 187.93 225.60 263.62 301.19
3 37.13 76.10 112.02 149.88 187.12 225.18 263.43 300.96
4 38.04 76.12 111.98 149.56 187.29 225.18 263.23 300.89
5 38.14 74.71 112.31 149.96 187.27 225.08 262.68 300.69
6 38.00 74.68 112.43 150.10 188.41 225.17 264.03 301.38
7 38.25 75.02 112.41 150.56 188.41 226.45 264.20 302.66
8 38.45 75.14 112.93 150.81 188.79 226.77 265.12 303.26
9 43.18 84.47 126.89 169.38 211.91 254.63 297.36 339.98
10 46.97 93.70 140.81 187.80 235.26 282.44 329.63 377.17
11 52.73 103.11 154.88 206.50 258.36 310.44 362.46 414.43
12 57.51 112.48 168.88 225.33 281.91 338.64 395.39 452.04
13 60.88 121.97 183.13 244.29 305.59 367.21 428.16 490.50
14 67.17 131.42 197.33 269.26 329.39 395.52 462.17 528.42
15 72.02 140.73 211.46 282.16 353.21 424.43 495.99 567.20
16 76.96 154.05 225.96 301.62 377.38 453.79 529.56 606.28
17 80.24 157.20 235.95 314.39 392.69 471.21 550.55 628.17
18 82.57 162.27 242.82 323.28 403.80 484.52 564.39 645.21
19 83.97 167.48 251.18 335.00 416.56 500.10 577.23 664.97
20 86.22 172.85 261.68 348.44 428.33 516.46 599.18 686.55
21 88.85 179.71 266.64 352.59 442.70 528.04 611.13 703.36
22 92.26 182.36 272.51 363.89 454.03 541.93 631.61 720.22
23 94.45 186.96 280.20 373.08 465.56 554.44 646.79 744.56
24 96.95 191.83 288.16 383.03 476.20 570.63 664.10 757.13
Copy
#nm 1 2 3 4 5 6 7 8
1 47.77 96.22 143.84 192.42 243.62 291.60 343.56 393.99
2 53.06 105.90 157.88 210.41 263.87 317.56 370.62 423.17
3 52.26 105.81 157.53 210.28 262.53 314.73 368.16 420.26
4 52.54 105.73 156.85 208.98 261.33 313.94 365.42 419.00
5 52.79 104.39 156.66 209.13 261.23 313.45 363.99 418.16
6 52.20 104.41 156.75 209.28 261.69 313.37 366.57 418.24
7 52.88 104.71 156.85 209.46 261.47 314.07 366.38 418.94
8 52.46 104.93 157.41 210.04 262.56 315.22 367.61 420.08
9 58.54 115.84 173.86 231.86 289.87 348.07 406.51 464.00
10 62.50 124.91 187.32 249.85 312.38 374.67 437.68 499.91
11 67.53 133.94 200.96 267.97 334.74 401.65 468.64 535.42
12 71.33 142.70 214.08 285.16 355.91 427.54 498.60 569.46
13 75.74 150.86 226.10 301.49 376.85 451.98 527.19 602.51
14 79.38 158.75 238.01 318.34 396.15 475.18 554.82 633.72
15 83.31 166.11 248.96 332.06 414.66 497.40 580.31 663.22
16 86.74 173.41 259.72 345.86 431.76 518.36 603.99 690.56
17 87.29 174.23 260.83 347.89 434.37 521.26 609.68 695.98
18 87.76 175.03 261.62 348.28 434.30 520.79 607.82 693.50
19 88.14 175.77 262.76 349.52 436.03 522.77 608.12 694.58
20 88.23 176.07 262.80 349.79 435.69 521.78 607.65 693.02
21 88.15 176.10 262.85 349.36 435.49 521.35 607.23 691.99
22 88.38 176.15 262.87 349.51 435.14 521.46 607.54 691.98
23 88.40 175.78 263.20 349.94 436.65 522.71 608.03 694.65
24 89.82 179.43 268.21 356.21 444.90 533.01 620.84 707.14
Update
#nm 1 2 3 4 5 6 7 8
1 48.89 98.38 147.71 197.00 246.72 296.25 346.27 396.20
2 53.27 106.83 159.88 213.84 267.45 321.11 374.80 428.81
3 52.58 106.51 158.46 211.56 264.90 318.18 373.10 425.52
4 53.00 106.32 158.08 210.99 264.27 317.86 370.66 424.62
5 53.64 106.28 159.47 212.77 266.24 320.09 373.25 427.26
6 53.46 105.65 158.57 211.75 265.10 318.53 371.73 425.12
7 53.27 105.79 158.67 211.86 265.02 318.46 372.16 425.60
8 53.41 106.06 158.66 211.79 265.33 318.59 372.52 425.63
9 58.25 115.39 173.19 231.55 289.95 348.77 407.76 467.12
10 62.35 123.65 185.64 248.16 310.85 373.68 437.04 500.42
11 66.63 132.40 198.79 265.66 332.52 399.71 467.42 535.02
12 70.55 140.36 210.87 281.78 353.04 424.29 496.04 567.91
13 74.21 147.78 222.21 297.16 372.24 447.84 523.06 599.14
14 77.36 154.84 232.89 312.43 389.85 468.86 548.51 628.29
15 80.66 161.55 242.88 324.86 407.28 489.70 572.64 656.77
16 84.03 168.34 252.95 339.08 425.31 512.69 600.96 689.87
17 84.98 170.12 256.19 343.64 430.84 520.18 610.56 701.63
18 85.99 171.94 258.47 346.10 434.84 525.15 615.40 706.29
19 87.08 174.01 261.62 350.40 439.55 529.28 619.64 712.28
20 87.81 175.52 263.26 352.34 441.65 532.47 623.44 716.55
21 88.29 176.79 265.10 353.98 443.52 534.70 627.10 718.23
22 89.02 177.86 266.86 356.50 446.40 537.52 630.55 722.20
23 89.83 178.95 268.45 358.42 448.90 540.02 630.22 724.65
24 90.88 181.35 272.95 364.15 458.64 552.49 647.41 742.60
Triad
#nm 1 2 3 4 5 6 7 8
1 45.94 93.06 138.37 183.10 233.96 279.77 322.84 374.52
2 47.21 94.91 141.67 190.68 238.45 286.49 334.60 383.00
3 47.49 96.77 142.96 190.79 238.89 287.02 335.27 383.45
4 48.62 97.34 143.70 191.81 240.00 288.15 336.57 385.43
5 48.30 96.40 144.79 193.23 241.46 290.06 338.17 387.32
6 49.32 96.49 144.88 193.35 242.08 290.29 339.38 387.30
7 48.81 96.76 145.16 193.62 242.11 290.94 339.68 388.37
8 49.16 96.91 145.39 193.85 242.68 291.29 340.07 389.04
9 54.38 108.11 160.75 214.49 268.09 321.94 375.74 429.06
10 59.56 116.95 175.40 233.76 292.28 350.52 409.34 467.33
11 63.30 126.59 189.97 253.08 316.16 379.13 442.88 505.55
12 68.02 136.02 203.94 271.41 339.29 406.92 475.11 542.29
13 73.36 144.90 217.09 289.90 361.77 434.05 506.14 578.56
14 76.84 153.70 230.38 309.62 383.07 459.64 536.47 612.79
15 81.10 161.95 242.69 323.97 403.99 484.49 565.12 646.35
16 85.28 170.83 254.79 339.54 423.67 508.49 593.30 678.28
17 86.48 171.99 257.76 343.62 428.77 514.06 600.38 685.92
18 87.44 174.14 260.41 346.87 432.87 519.45 605.41 691.80
19 87.86 175.34 262.99 350.40 436.52 523.14 607.88 695.12
20 88.70 177.40 265.84 353.82 440.37 528.56 615.96 703.69
21 89.55 179.25 267.99 355.66 444.65 532.85 619.35 707.73
22 90.22 179.86 269.01 358.28 446.85 535.84 624.45 711.63
23 90.75 180.55 270.40 359.60 448.76 537.76 626.61 714.75
24 91.79 182.74 273.23 362.46 453.35 543.15 632.72 721.91
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
