博文
Fluent 并行核数对比
|
测试平台：2颗 epyc 7742 Fluent2020
网格12w，density based，开启energy，计算161 迭代步
| 核数 | 时间/迭代步 [s/iter] |
| 10 | 0.124 |
| 20 | 0.068 |
| 40 | 0.023 |
| 60 | 0.018 |
| 80 | 0.016 |
可见，核数越多，越快。
但是，如果同时开启2个以上fluent，
速度会相互影响。
| Performance Timer for 161 iterations on 10 compute nodes Average wall-clock time per iteration: 0.124 sec Global reductions per iteration: 101 ops Global reductions time per iteration: 0.000 sec (0.0%) Message count per iteration: 2284 messages Data transfer per iteration: 2.981 MB LE solves per iteration: 3 solvesLE wall-clock time per iteration: 0.035 sec (28.5%) LE global solves per iteration: 1 solves LE global wall-clock time per iteration: 0.000 sec (0.2%) LE global matrix maximum size: 62 AMG cycles per iteration: 3.000 cycles Relaxation sweeps per iteration: 226 sweeps Relaxation exchanges per iteration: 0 exchanges LE early protections (stall) per iteration: 0.000 times LE early protections (divergence) per iteration: 0.000 times Total SVARS touched: 369 Time-step updates per iteration: 0.31 updates Time-step wall-clock time per iteration: 0.002 sec (1.9%) Total wall-clock time: 19.916 sec |
| Performance Timer for 330 iterations on 20 compute nodes Average wall-clock time per iteration: 0.068 sec Global reductions per iteration: 101 ops Global reductions time per iteration: 0.000 sec (0.0%) Message count per iteration: 7687 messages Data transfer per iteration: 5.265 MB LE solves per iteration: 3 solves LE wall-clock time per iteration: 0.020 sec (28.7%) LE global solves per iteration: 1 solves LE global wall-clock time per iteration: 0.000 sec (0.6%) LE global matrix maximum size: 192 AMG cycles per iteration: 3.000 cycles Relaxation sweeps per iteration: 212 sweeps Relaxation exchanges per iteration: 0 exchanges LE early protections (stall) per iteration: 0.000 times LE early protections (divergence) per iteration: 0.000 times Total SVARS touched: 369 Time-step updates per iteration: 0.30 updates Time-step wall-clock time per iteration: 0.001 sec (2.1%) Total wall-clock time: 22.477 sec |
| Performance Timer for 161 iterations on 40 compute nodes Average wall-clock time per iteration: 0.023 sec Global reductions per iteration: 101 ops Global reductions time per iteration: 0.000 sec (0.0%) Message count per iteration: 13744 messages Data transfer per iteration: 7.996 MB LE solves per iteration: 3 solves LE wall-clock time per iteration: 0.008 sec (33.1%) LE global solves per iteration: 1 solves LE global wall-clock time per iteration: 0.001 sec (4.0%) LE global matrix maximum size: 602 AMG cycles per iteration: 3.000 cycles Relaxation sweeps per iteration: 200 sweeps Relaxation exchanges per iteration: 0 exchanges LE early protections (stall) per iteration: 0.000 times LE early protections (divergence) per iteration: 0.000 times Total SVARS touched: 369 Time-step updates per iteration: 0.31 updates Time-step wall-clock time per iteration: 0.001 sec (2.6%) Total wall-clock time: 3.732 sec |
| Performance Timer for 161 iterations on 60 compute nodes Average wall-clock time per iteration: 0.018 sec Global reductions per iteration: 101 ops Global reductions time per iteration: 0.000 sec (0.0%) Message count per iteration: 22381 messages Data transfer per iteration: 10.313 MB LE solves per iteration: 3 solves LE wall-clock time per iteration: 0.006 sec (34.2%) LE global solves per iteration: 1 solves LE global wall-clock time per iteration: 0.001 sec (5.6%) LE global matrix maximum size: 612 AMG cycles per iteration: 3.000 cycles Relaxation sweeps per iteration: 197 sweeps Relaxation exchanges per iteration: 0 exchanges LE early protections (stall) per iteration: 0.000 times LE early protections (divergence) per iteration: 0.000 times Total SVARS touched: 369 Time-step updates per iteration: 0.31 updates Time-step wall-clock time per iteration: 0.001 sec (3.1%) Total wall-clock time: 2.843 sec |
| Performance Timer for161 iterations on 80 compute nodes Average wall-clock time per iteration: 0.016 sec Global reductions per iteration: 101 ops Global reductions time per iteration: 0.000 sec (0.0%) Message count per iteration: 31017 messages Data transfer per iteration: 12.356 MB LE solves per iteration: 3 solves LE wall-clock time per iteration: 0.007 sec (41.4%) LE global solves per iteration: 1 solves LE global wall-clock time per iteration: 0.001 sec (7.4%) LE global matrix maximum size: 623 AMG cycles per iteration: 3.000 cycles Relaxation sweeps per iteration: 199 sweeps Relaxation exchanges per iteration: 0 exchanges LE early protections (stall) per iteration: 0.000 times LE early protections (divergence) per iteration: 0.000 times Total SVARS touched: 369 Time-step updates per iteration: 0.31 updates Time-step wall-clock time per iteration: 0.001 sec (3.3%) Total wall-clock time: 2.631 sec |
https://blog.sciencenet.cn/blog-531760-1361900.html
上一篇：Sajben扩压器型线python计算
下一篇：[转载]百度网盘上传网络异常
扫一扫，分享此博文