portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-11-03 21:44:33 +00:00
Code Releases Activity

Benchmark results

Browse Source
This commit is contained in:
Peter Boyle
2024-03-06 19:03:35 +01:00
parent b812a7b4c6
commit 228bbb9d81
1 changed files with 70 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

70
systems/Booster/benchmarks/Benchmark_usqcd.csv Normal file
View File

@@ -0,0 +1,70 @@
Memory Bandwidth
Bytes, GB/s per node
3145728, 225.900365
50331648, 2858.859504
254803968, 4145.556367
805306368, 4905.772480
1966080000, 4978.312557
GEMM
M, N, K, BATCH, GF/s per rank
16, 8, 16, 256, 1.713639
16, 16, 16, 256, 288.268316
16, 32, 16, 256, 597.053950
32, 8, 32, 256, 557.382591
32, 16, 32, 256, 1100.145311
32, 32, 32, 256, 1885.080449
64, 8, 64, 256, 1725.163599
64, 16, 64, 256, 3389.336566
64, 32, 64, 256, 4168.252422
16, 8, 256, 256, 1326.262134
16, 16, 256, 256, 2318.095475
16, 32, 256, 256, 3555.436503
32, 8, 256, 256, 1920.139170
32, 16, 256, 256, 3486.174753
32, 32, 256, 256, 5320.821724
64, 8, 256, 256, 2539.597502
64, 16, 256, 256, 5003.456775
64, 32, 256, 256, 7837.531562
8, 256, 16, 256, 1427.848170
16, 256, 16, 256, 2222.147815
32, 256, 16, 256, 2877.121715
8, 256, 32, 256, 1922.890086
16, 256, 32, 256, 3199.469082
32, 256, 32, 256, 4845.405343
8, 256, 64, 256, 2639.483343
16, 256, 64, 256, 5012.800299
32, 256, 64, 256, 7216.006882
Communications
Packet bytes, direction, GB/s per node
4718592, 2, 206.570734
4718592, 3, 207.501847
4718592, 6, 189.730277
4718592, 7, 204.301218
15925248, 2, 307.882997
15925248, 3, 287.901076
15925248, 6, 295.603109
15925248, 7, 300.682033
37748736, 2, 331.740364
37748736, 3, 338.610627
37748736, 6, 332.580657
37748736, 7, 336.336579
Per node summary table
L , Wilson, DWF4, Staggered, GF/s per node
8 , 16, 1165, 10
12 , 473, 4901, 163
16 , 1436, 8464, 442
24 , 4133, 10139, 1530
32 , 5726, 11487, 2518
1 Memory Bandwidth
2 Bytes, GB/s per node
3 3145728, 225.900365
4 50331648, 2858.859504
5 254803968, 4145.556367
6 805306368, 4905.772480
7 1966080000, 4978.312557
8 GEMM
9 M, N, K, BATCH, GF/s per rank
10 16, 8, 16, 256, 1.713639
11 16, 16, 16, 256, 288.268316
12 16, 32, 16, 256, 597.053950
13 32, 8, 32, 256, 557.382591
14 32, 16, 32, 256, 1100.145311
15 32, 32, 32, 256, 1885.080449
16 64, 8, 64, 256, 1725.163599
17 64, 16, 64, 256, 3389.336566
18 64, 32, 64, 256, 4168.252422
19 16, 8, 256, 256, 1326.262134
20 16, 16, 256, 256, 2318.095475
21 16, 32, 256, 256, 3555.436503
22 32, 8, 256, 256, 1920.139170
23 32, 16, 256, 256, 3486.174753
24 32, 32, 256, 256, 5320.821724
25 64, 8, 256, 256, 2539.597502
26 64, 16, 256, 256, 5003.456775
27 64, 32, 256, 256, 7837.531562
28 8, 256, 16, 256, 1427.848170
29 16, 256, 16, 256, 2222.147815
30 32, 256, 16, 256, 2877.121715
31 8, 256, 32, 256, 1922.890086
32 16, 256, 32, 256, 3199.469082
33 32, 256, 32, 256, 4845.405343
34 8, 256, 64, 256, 2639.483343
35 16, 256, 64, 256, 5012.800299
36 32, 256, 64, 256, 7216.006882
37 Communications
38 Packet bytes, direction, GB/s per node
39 4718592, 2, 206.570734
40 4718592, 3, 207.501847
41 4718592, 6, 189.730277
42 4718592, 7, 204.301218
43 15925248, 2, 307.882997
44 15925248, 3, 287.901076
45 15925248, 6, 295.603109
46 15925248, 7, 300.682033
47 37748736, 2, 331.740364
48 37748736, 3, 338.610627
49 37748736, 6, 332.580657
50 37748736, 7, 336.336579
51 Per node summary table
52 L , Wilson, DWF4, Staggered, GF/s per node
53 8 , 16, 1165, 10
54 12 , 473, 4901, 163
55 16 , 1436, 8464, 442
56 24 , 4133, 10139, 1530
57 32 , 5726, 11487, 2518