200万网格并行算力测试(OpenFOAM版本)
-
CPU型号:AMD Ryzen 5950X 系统:Ubuntu 20.04 (WSL2) OpenFOAM版本:OpenFOAM-8 16 407.96 8 369.83 4 382.25 2 402.88 1 615.98
-
@stickyrice 高主频单核还是很能打的。另外,可否用真系统跑一下测试下WSL的效率?
-
我也测试一下算力玩玩。
CPU型号:E5-2630 V3 系统:Ubuntu 16.04 LTS OpenFOAM版本:OpenFOAM-7 16 165.57 8 209.59 4 327.4 2 668.9 1 1453.02
-
@stickyrice 后来用真系统跑了没 老铁
-
@stickyrice 多谢。我之前用虚拟机跑的时候特别慢。看来我需要再测一下。我那个服务器上装了双系统。等4月份我测一下。
-
@李东岳
根据下方链接的java编译速度作为依据
https://raspberrypi.stackexchange.com/questions/114144/pi-4-performance-against-x86-ci-cd-server-with-java-mavenComputer Arch CPU RAM Drive Java Build Time ------------- ------ -------- ----- -------- ------ ---------- Lenovo T420 64-bit i5-2520M 16 GB SATA SSD 11+28 0:25 Lenovo T480 64-bit i5-8350U 16 GB NVMe SSD 11+28 0:12 Raspberry Pi3 32-bit 1 GB USB2 SSD 11.0.7 4:32 Raspberry Pi4 64-bit 8 GB USB3 SSD 11.0.8 1:10
假如不考虑java与c++的区别,单纯按照下面链接的benchmark对比的话,8750H比8350U快21%
https://cpu.userbenchmark.com/Compare/Intel-Core-i7-8750H-vs-Intel-Core-i5-8350U/m470418vsm388461咱们的benchmark 8750H 4核是910秒,
只按照线性的预测的话,则8350U是1101秒
单独一个raspberry pi 4的话就差不多 6422s。。So I can buy a 48-port switch ($750), 48 Pi 4s (48 x $55) and 48 PoE [Power over Ethernet] HATs (48 x $30) and for around $5000 you've got 192 Arm cores.
但是真实的计算速度还要考虑到数据传输带来的损耗,内存的带宽什么的
考虑到设备维护等因素
所以说在3W这个价位的话,是不是一个正常的x86更合算一点?上面的文章也提到了
In traditional server land you'll be lucky to get 32 Intel or AMD cores for the same money. So if you're running micro-services you might get significantly better value out of the Raspberry Pi.