topic Re:Bug in intelMPI on AMD processors in Intel® MPI Library

Bug in intelMPI on AMD processors

MichalKrupicka — Thu, 22 Jun 2023 13:26:24 GMT

The problem occurs in pamcrash and starccm with intelmpi (2018, 2019, 2021) on AMD processor (AMD EPYC 7763 64-Core Processor, 2 sockets, 1 node has 128 cores) on SLES 15 SP4.

On SLES 15 SP3 there was one workaroud with LD_PRELOADed strtok library. But when SP4 came, it doesn’t work.

The computation hangs everytime when the count of mpi ranks is higher than 256.

Re:Bug in intelMPI on AMD processors

ShivaniK_Intel — Fri, 23 Jun 2023 06:23:51 GMT

Hi,

Thanks for posting in the Intel forums.

Could you please let us know whether you are facing a similar issue with the Intel processor?

We can only offer direct support for Intel hardware platforms that the Intel® oneAPI product supports. Intel provides instructions on how to compile oneAPI code for both CPU and a wide range of GPU accelerators.

https://intel.github.io/llvm-docs/GetStartedGuide.html

Thanks & Regards

Shivani

Re: Re:Bug in intelMPI on AMD processors

MichalKrupicka — Fri, 23 Jun 2023 07:05:45 GMT

Hi Shivani,

thanks for the replay. We do not face this issue on intel processors.

Does it mean that intel mpi is not prepared for other processors than intel`s? Is intelmpi supposed to be run only on intel processors?

Thanks & regards,

Michal

Re:Bug in intelMPI on AMD processors

ShivaniK_Intel — Thu, 29 Jun 2023 06:25:11 GMT

Hi,

Could you please provide us with the sample reproducer code and steps to reproduce the issue at our end?

Thanks & Regards

Shivani

Re:Bug in intelMPI on AMD processors

ShivaniK_Intel — Wed, 05 Jul 2023 11:01:35 GMT

Hi,

Could you please provide us with the details of the interconnect used? Also please let us know which Intel MPI variables you set to change the fabric.

Thanks & Regards

Shivani

Re:Bug in intelMPI on AMD processors

ShivaniK_Intel — Wed, 12 Jul 2023 10:32:50 GMT

Hi,

As we did not hear back from you could please respond to my previous post?

Could you please let us know if this issue also happens with IMB-MPI1 benchmarks included in IMPI distribution?

$ mpirun -n 256 IMB-MPI1 sendrecv

Could you also please let us know whether you are using 256+ ranks on a single node with 128 cores or if this happens when using more than a single node?

Thanks & Regards

Shivani

Re:Bug in intelMPI on AMD processors

ShivaniK_Intel — Thu, 20 Jul 2023 16:57:14 GMT

Hi,

As we did not hear back from you could you please respond to my previous post?

Thanks & Regards

Shivani

Re:Bug in intelMPI on AMD processors

ShivaniK_Intel — Thu, 27 Jul 2023 09:44:59 GMT

Hi,

We have not heard back from you. This thread will no longer be monitored by Intel. If you need further assistance, please post a new question.

Thanks & Regards

Shivani