AMD Rome S2 M1 C64
- Processor: AMD EPYC 7662 64-Core Processor
- Base frequency: 2.0 GHz
- Number of sockets: 2
- Number of memory domains per socket: 1
- Memory domain specs: 8-channel DDR4-3200
- Number of cores per socket: 64
- Number of HWThreads per core: 2
- MachineState output: NA
+----------+-------------------------------------------------------------------+
| Compiler | AMD clang |
|----------|-------------------------------------------------------------------|
| Version | AMD clang version 10.0.0 (CLANG: AOCC_2.2.0-Build#93 2020_06_25) |
+----------+-------------------------------------------------------------------+
Optimizing flags: -Ofast -fnt-store=aggressive -std=c99 -fopenmp
Remark: On the this Rome system a larger than default data set (20GB instead of 4GB) was used to rule out caching effects.
All results are in GB/s.
Summary results:
+------------------------------------------------+
| Single core | 35.58 (STriad) |
| Memory domain | 146.63 (STriad with 64 cores) |
| Socket | 146.63 (STriad with 64 cores) |
| Node | 292.55 (STriad with 64 cores) |
+------------------------------------------------+
Results for scaling within a memory domain:
#nt Init Sum Copy Update Triad Daxpy STriad SDaxpy
1 23.41 7.95 32.34 29.28 34.63 32.42 35.58 33.51
2 23.42 15.63 44.17 40.81 46.42 40.89 44.45 40.68
3 23.43 23.80 44.85 41.40 46.89 41.28 43.87 40.27
4 23.42 31.33 45.07 41.48 46.54 40.81 43.48 39.78
5 23.43 36.06 44.58 41.80 46.55 41.02 43.43 39.81
6 23.44 35.61 44.50 42.22 46.26 40.80 43.34 39.70
7 23.44 36.15 44.47 42.27 45.97 40.54 43.12 39.48
8 23.45 37.26 44.33 42.35 45.83 40.37 43.05 39.38
9 26.38 41.55 49.81 47.18 51.07 45.07 48.06 43.99
10 29.31 45.73 55.15 51.93 56.00 49.63 52.80 48.46
11 32.24 49.63 60.50 56.64 60.90 54.15 57.45 52.80
12 35.17 53.30 65.79 61.26 65.49 58.46 61.91 56.99
13 38.10 57.08 71.10 65.66 70.23 62.81 66.37 61.21
14 41.04 60.80 76.28 70.06 74.52 66.82 70.61 65.21
15 43.97 64.65 81.36 74.47 78.62 70.87 74.71 69.12
16 46.89 68.48 86.52 78.94 82.62 74.89 78.65 72.95
17 49.82 72.19 90.43 82.25 86.75 78.67 82.82 76.78
18 52.74 76.03 94.40 85.69 90.54 82.18 86.58 80.42
19 55.68 79.73 98.23 89.29 94.25 85.68 90.27 83.88
20 58.61 83.61 101.87 92.93 97.83 89.24 93.81 87.40
21 61.53 87.12 105.29 96.05 101.35 92.60 97.36 90.81
22 64.47 90.76 108.90 99.37 104.72 95.84 100.74 94.10
23 67.39 94.38 112.25 102.58 107.81 98.93 103.92 97.18
24 70.33 97.89 115.05 105.57 110.87 101.96 107.06 100.19
25 73.05 101.07 116.82 107.50 113.74 104.59 110.09 103.09
26 75.99 103.86 118.30 109.69 116.29 106.98 112.91 105.82
27 78.92 106.38 120.11 111.68 118.87 109.48 115.54 108.38
28 81.83 109.33 121.74 113.67 121.44 112.06 118.04 110.90
29 84.63 111.75 123.06 115.48 123.75 114.17 120.67 113.38
30 87.55 114.78 124.43 117.66 126.08 116.65 123.08 115.83
31 90.42 117.37 126.10 119.64 128.45 118.94 125.45 118.27
32 93.29 119.84 127.11 121.37 130.55 121.01 127.57 120.46
33 95.04 121.11 126.33 120.94 131.51 121.63 129.47 121.89
34 97.51 122.01 126.39 121.48 132.28 122.74 130.74 123.13
35 99.86 122.95 126.61 122.19 133.47 124.02 132.35 124.73
36 102.30 123.60 127.24 122.85 134.41 125.19 133.73 126.15
37 104.02 125.08 126.93 123.65 135.16 126.08 135.04 127.33
38 106.70 125.64 127.15 124.11 136.20 127.31 136.39 128.85
39 109.25 126.69 127.48 124.95 137.09 128.25 137.52 130.00
40 111.26 127.57 128.47 125.47 137.89 129.27 138.71 131.37
41 112.17 127.82 127.33 125.56 137.76 129.77 139.04 131.41
42 113.46 128.67 128.55 126.34 138.27 130.49 139.43 132.09
43 114.24 128.92 128.77 126.47 138.48 130.89 139.80 132.44
44 115.84 129.45 129.51 126.95 138.86 131.74 139.94 133.00
45 116.62 130.36 129.70 128.43 138.91 132.45 140.30 133.33
46 117.95 130.62 130.37 128.61 139.25 132.82 140.56 133.83
47 118.86 131.39 131.15 129.11 139.52 133.15 140.89 133.92
48 120.58 133.28 131.42 129.93 140.82 134.86 142.09 135.39
49 120.82 133.81 131.22 130.33 140.97 134.86 142.34 135.70
50 121.75 134.49 131.27 130.58 141.56 135.34 142.82 136.28
51 122.60 135.06 131.56 131.34 141.80 135.97 143.26 136.64
52 123.05 135.83 132.16 131.62 141.89 136.37 143.62 136.83
53 123.69 136.29 132.03 132.01 142.44 136.87 143.97 137.43
54 124.13 136.44 133.12 132.76 142.58 137.00 144.17 137.57
55 125.38 137.62 132.99 132.84 143.05 137.69 144.33 138.37
56 125.37 138.06 133.79 132.98 143.51 138.38 144.51 138.85
57 125.93 138.35 133.99 133.59 143.57 137.98 144.98 138.70
58 126.67 138.54 133.52 133.45 143.51 138.45 145.40 138.72
59 127.13 139.46 134.90 134.49 144.15 138.98 145.66 139.34
60 128.15 139.76 134.63 134.88 144.28 139.26 145.79 139.36
61 128.49 140.40 134.72 134.46 144.31 139.34 146.18 139.69
62 128.41 140.72 135.17 135.69 144.61 139.50 146.30 140.00
63 129.29 141.65 136.11 135.90 144.89 140.05 146.38 140.48
64 129.58 141.37 135.92 135.92 144.84 140.04 146.63 140.30
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
1 23.41 46.82
2 23.42 46.81
3 23.43 46.85
4 23.42 46.83
5 23.43 46.80
6 23.44 46.83
7 23.44 46.87
8 23.45 46.89
9 26.38 52.74
10 29.31 58.60
11 32.24 64.46
12 35.17 70.32
13 38.10 76.18
14 41.04 82.04
15 43.97 87.90
16 46.89 93.75
17 49.82 99.57
18 52.74 105.43
19 55.68 111.27
20 58.61 117.15
21 61.53 122.97
22 64.47 128.89
23 67.39 134.72
24 70.33 140.53
25 73.05 146.06
26 75.99 151.93
27 78.92 157.71
28 81.83 163.51
29 84.63 169.15
30 87.55 174.98
31 90.42 180.75
32 93.29 186.36
33 95.04 190.06
34 97.51 195.14
35 99.86 199.77
36 102.30 204.42
37 104.02 208.44
38 106.70 212.89
39 109.25 218.36
40 111.26 222.66
41 112.17 223.74
42 113.46 227.22
43 114.24 229.16
44 115.84 231.24
45 116.62 232.13
46 117.95 235.75
47 118.86 236.44
48 120.58 241.21
49 120.82 241.56
50 121.75 243.13
51 122.60 245.09
52 123.05 245.14
53 123.69 246.55
54 124.13 249.58
55 125.38 250.68
56 125.37 251.71
57 125.93 251.41
58 126.67 252.44
59 127.13 253.79
60 128.15 255.78
61 128.49 255.31
62 128.41 257.10
63 129.29 257.85
64 129.58 258.81
Sum:
#nm 1 2
1 7.95 15.85
2 15.63 31.07
3 23.80 47.50
4 31.33 62.95
5 36.06 72.18
6 35.61 71.20
7 36.15 72.28
8 37.26 74.52
9 41.55 83.10
10 45.73 91.49
11 49.63 99.21
12 53.30 106.51
13 57.08 114.16
14 60.80 121.77
15 64.65 129.20
16 68.48 136.42
17 72.19 144.13
18 76.03 151.84
19 79.73 159.19
20 83.61 166.33
21 87.12 173.75
22 90.76 181.40
23 94.38 188.36
24 97.89 195.82
25 101.07 201.65
26 103.86 207.23
27 106.38 212.44
28 109.33 218.16
29 111.75 223.64
30 114.78 228.97
31 117.37 234.30
32 119.84 239.79
33 121.11 241.61
34 122.01 243.98
35 122.95 245.24
36 123.60 247.17
37 125.08 249.79
38 125.64 250.59
39 126.69 252.69
40 127.57 253.66
41 127.82 254.35
42 128.67 255.51
43 128.92 258.11
44 129.45 259.01
45 130.36 258.58
46 130.62 260.09
47 131.39 263.21
48 133.28 264.82
49 133.81 266.57
50 134.49 267.33
51 135.06 269.11
52 135.83 268.84
53 136.29 271.36
54 136.44 272.80
55 137.62 273.30
56 138.06 273.27
57 138.35 274.84
58 138.54 274.17
59 139.46 277.61
60 139.76 279.88
61 140.40 279.17
62 140.72 280.11
63 141.65 281.39
64 141.37 283.20
Copy
#nm 1 2
1 32.34 64.78
2 44.17 87.99
3 44.85 89.53
4 45.07 90.15
5 44.58 89.11
6 44.50 89.04
7 44.47 88.92
8 44.33 88.63
9 49.81 99.62
10 55.15 110.34
11 60.50 121.05
12 65.79 131.62
13 71.10 142.29
14 76.28 152.58
15 81.36 162.88
16 86.52 172.79
17 90.43 181.08
18 94.40 189.04
19 98.23 196.59
20 101.87 203.45
21 105.29 210.59
22 108.90 217.31
23 112.25 224.51
24 115.05 230.37
25 116.82 233.38
26 118.30 236.45
27 120.11 239.82
28 121.74 243.03
29 123.06 246.15
30 124.43 249.10
31 126.10 251.85
32 127.11 253.64
33 126.33 252.30
34 126.39 251.85
35 126.61 253.03
36 127.24 253.91
37 126.93 253.53
38 127.15 252.71
39 127.48 255.05
40 128.47 256.37
41 127.33 255.21
42 128.55 256.86
43 128.77 257.60
44 129.51 259.00
45 129.70 258.55
46 130.37 260.45
47 131.15 260.60
48 131.42 262.46
49 131.22 260.62
50 131.27 261.90
51 131.56 263.47
52 132.16 263.37
53 132.03 263.88
54 133.12 265.20
55 132.99 265.73
56 133.79 266.90
57 133.99 265.57
58 133.52 266.70
59 134.90 269.04
60 134.63 269.52
61 134.72 268.63
62 135.17 270.71
63 136.11 270.76
64 135.92 272.10
Update
#nm 1 2
1 29.28 58.75
2 40.81 81.64
3 41.40 82.81
4 41.48 83.31
5 41.80 83.63
6 42.22 84.45
7 42.27 84.71
8 42.35 84.99
9 47.18 94.57
10 51.93 104.27
11 56.64 113.69
12 61.26 122.91
13 65.66 131.83
14 70.06 140.66
15 74.47 149.55
16 78.94 158.14
17 82.25 165.24
18 85.69 172.40
19 89.29 179.51
20 92.93 186.06
21 96.05 192.64
22 99.37 199.60
23 102.58 206.05
24 105.57 212.42
25 107.50 216.56
26 109.69 221.21
27 111.68 224.72
28 113.67 228.93
29 115.48 232.81
30 117.66 237.03
31 119.64 240.79
32 121.37 245.02
33 120.94 243.36
34 121.48 245.53
35 122.19 246.38
36 122.85 247.67
37 123.65 248.43
38 124.11 250.69
39 124.95 251.76
40 125.47 252.19
41 125.56 253.16
42 126.34 254.71
43 126.47 256.57
44 126.95 256.93
45 128.43 258.40
46 128.61 258.74
47 129.11 260.25
48 129.93 262.90
49 130.33 262.76
50 130.58 265.21
51 131.34 265.26
52 131.62 265.91
53 132.01 267.26
54 132.76 268.43
55 132.84 268.34
56 132.98 269.71
57 133.59 269.45
58 133.45 270.15
59 134.49 272.94
60 134.88 274.14
61 134.46 275.19
62 135.69 275.00
63 135.90 275.13
64 135.92 277.10
Triad
#nm 1 2
1 34.63 69.67
2 46.42 92.75
3 46.89 93.84
4 46.54 93.02
5 46.55 93.02
6 46.26 92.55
7 45.97 91.97
8 45.83 91.76
9 51.07 102.21
10 56.00 112.29
11 60.90 121.85
12 65.49 131.13
13 70.23 140.37
14 74.52 149.25
15 78.62 157.49
16 82.62 165.35
17 86.75 173.67
18 90.54 181.59
19 94.25 188.91
20 97.83 195.39
21 101.35 202.80
22 104.72 209.45
23 107.81 215.84
24 110.87 221.58
25 113.74 227.18
26 116.29 232.33
27 118.87 237.72
28 121.44 242.63
29 123.75 247.46
30 126.08 252.35
31 128.45 256.86
32 130.55 260.75
33 131.51 262.79
34 132.28 264.14
35 133.47 266.23
36 134.41 268.49
37 135.16 269.84
38 136.20 271.39
39 137.09 273.63
40 137.89 275.23
41 137.76 275.19
42 138.27 275.88
43 138.48 276.90
44 138.86 277.24
45 138.91 277.08
46 139.25 277.43
47 139.52 278.63
48 140.82 280.94
49 140.97 281.13
50 141.56 281.46
51 141.80 282.74
52 141.89 283.95
53 142.44 283.92
54 142.58 284.64
55 143.05 285.40
56 143.51 285.48
57 143.57 285.94
58 143.51 286.37
59 144.15 287.28
60 144.28 288.09
61 144.31 288.37
62 144.61 288.47
63 144.89 288.95
64 144.84 289.52
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
