Linux下使用AVX_512指令集编译的gromacs会比用AVX_256编译的计算速度要快吗？

12313

想问一下大家，目前我们服务器使用的是AVX_256指令集进行Gromacs进行的计算，如果换成是AVX_512指令集的话计算速度会进一步提升吗？

mizu-bai

AVX_512 Skylake-X desktop and Skylake-SP Xeon processors (2017) and AMD Zen4 (2022); on Intel it will generally be fastest on the higher-end desktop and server processors with two 512-bit fused multiply-add units (e.g. Core i9 and Xeon Gold). However, certain desktop and server models (e.g. Xeon Bronze and Silver) come with only one AVX512 FMA unit and therefore on these processors AVX2_256 is faster (compile- and runtime checks try to inform about such cases). On AMD it is beneficial to use starting with Zen4. Additionally, with GPU accelerated runs AVX2_256 can also be faster on high-end Skylake CPUs with both 512-bit FMA units enabled.

https://manual.gromacs.org/docum ... x.html#simd-support

建议你两个版本都编译了跑跑试试，你也没给 CPU 型号、是否带 GPU 以及体系的一些信息，所以还是得自己算算看。

12313

mizu-bai 发表于 2024-5-10 09:38
https://manual.gromacs.org/docum ... x.html#simd-support

建议你两个版本都编译了跑跑试试，你也没给 ...

好的，谢谢老师