topic Re:dapl async_event CQ (0x1fdb8e0) ERR 0 in Intel® MPI Library

dapl async_event CQ (0x1fdb8e0) ERR 0

Ambika1 — Fri, 28 Apr 2023 06:49:37 GMT

Dear Support official,

I am facing a problem while submitting jobs on an HPC . We are submitting using Slurm but the MPI assignment is being done through mpiexec.hydra.

When we are assigning large number of n-task the simulation stops midway and shows the following comments in the log, after which the simulation does not progress any further.

For e.g. while submitting a job with 240 tasks and memory of more than 17GB per task (run.script is attached for your kind reference) we get this following error:

"hm010:UCM:dec1:1f7d2700: 781258369 us(781258369 us!!!): dapl async_event CQ (0x1fdb8e0) ERR 0
hm010:UCM:dec1:1f7d2700: 781258411 us(42 us): -- dapl_evd_cq_async_error_callback (0x1f7d700, 0x18171f0, 0x2b351f7d1d30, 0x1fdb8e0)
hm010:UCM:dec1:1f7d2700: 781258429 us(18 us): dapl async_event QP (0x55c0cf0) Event 1"

Can you please help us in resolving this problem? I am a novice in this field.

Thank you.

Sincerely,

Ambika

Re:dapl async_event CQ (0x1fdb8e0) ERR 0

RabiyaSK_Intel — Tue, 02 May 2023 08:52:40 GMT

Hi,

Thanks for posting in Intel Communities.

Could you please provide the following information so we could reproduce your issue at our end:

1. Are you using Intel OneAPI toolkits? If so, please specify the version of the toolkit.

2. OS, CPU and hardware details

3. Could you please provide a sample reproducer along with the steps to reproduce?

Thanks & Regards,

Shaik Rabiya

Re: Re:dapl async_event CQ (0x1fdb8e0) ERR 0

Ambika1 — Tue, 02 May 2023 12:11:33 GMT

Dear M(r)s Rabiya,

Please find my response to your queries below:

1. I am using a software QuantumATK which utilizes Intel's mpiexec.hydra for mpi processes. I am not sure if it comes under OneAPI.

2. I am trying to run simulations on an HPC with centos 7. The system has 384 nodes with the following configuration for each node:

2* Intel Xeon SKL G-6148, Cores = 40, 2.4GHz, Memory= 192 GB, DDR4 2666 MHz.

3. Since I am using a proprietary software I wont be able to send you a sample script, however I am readily willing to share my screen/computer through a suitable app such as Anydesk and collaborate, for you to look into the issue.

I have no experience and knowledge in MPI coding and thus am solely depending on your help.

Thank you very much.

Sincerely

Ambika.

Re: dapl async_event CQ (0x1fdb8e0) ERR 0

RabiyaSK_Intel — Thu, 04 May 2023 05:45:31 GMT

Hi,

>>>I am using a software QuantumATK which utilizes Intel's mpiexec.hydra for mpi processes. I am not sure if it comes under OneAPI.

Could you please confirm which version of Intel OneAPI HPC toolkit and Intel OneAPI MPI Library you are using?

>>>I am trying to run simulations on an HPC with centos 7.

If you are using latest Intel MPI Library for mpiexec.hydra, the centOS 7 is not a supported Operating System. Please go through the below links and try on supported hardware and let us know if you face any issues:

MPI Library System Requirements:

https://www.intel.com/content/www/us/en/developer/articles/system-requirements/mpi-library-system-requirements.html

HPC Toolkit System Requirements:

https://www.intel.com/content/www/us/en/developer/articles/system-requirements/intel-oneapi-hpc-toolkit-system-requirements.html

>>>Since I am using a proprietary software I wont be able to send you a sample script, however I am readily willing to share my screen/computer through a suitable app such as Anydesk and collaborate, for you to look into the issue.

For Community support we would require you to submit a minimum reproduction sample code specific to your issue that provides us the most relevant background information for triage. In case you require privacy, and unable to share the issue / sample with us publicly and if you are a licensed oneAPI product customer and/or member of Intel’s oneAPI Academic Program please submit a ticket for Priority support so that your application can be handled with the required data protection and privacy regulations.

Thanks & Regards,

Shaik Rabiya

Re: dapl async_event CQ (0x1fdb8e0) ERR 0

Ambika1 — Fri, 05 May 2023 07:59:32 GMT

Dear M(r)s Rabiya,

Thank you very much. I will contact with the HPC administrator to find out about the Intel OneAPI subscriptions and let you know asap.

Re:dapl async_event CQ (0x1fdb8e0) ERR 0

RabiyaSK_Intel — Thu, 11 May 2023 05:47:45 GMT

Hi,

We haven't heard back from you. We are waiting for your response to our previous reply.

Could you please provide all the necessary details that we requested?

Thanks & Regards,

Shaik Rabiya

Re: Re:dapl async_event CQ (0x1fdb8e0) ERR 0

Ambika1 — Thu, 11 May 2023 11:47:20 GMT

Dear M(r)s Rabiya,

QuantumATK is compatible with CentOS 7 because it is 2019 version.

I have attached the run script , to run this script you need license of QuantumATK. It is available in our institute and not accessible from outside. You can access it by using our system.

Initially we had academic subscription of Intel but now its contract term is over. However subscription would be probabily initiated by HPC administrator.

Re: Re:dapl async_event CQ (0x1fdb8e0) ERR 0

Ambika1 — Thu, 11 May 2023 12:49:29 GMT

Dear M(r)s Rabiya,

We are not using mpiexec.hydra of Intel. QuantumATK have its own mpiexec.hydra.

If you can solve this issue kindly help us to resolve this problem

Re:dapl async_event CQ (0x1fdb8e0) ERR 0

RabiyaSK_Intel — Fri, 12 May 2023 10:11:22 GMT

Hi,

We can only offer direct support for Intel hardware platforms that the Intel® oneAPI product supports.

Could you please confirm which component of Intel is being used here?

Thanks & Regards,

Shaik Rabiya

Re:dapl async_event CQ (0x1fdb8e0) ERR 0

RabiyaSK_Intel — Thu, 18 May 2023 06:40:24 GMT

Hi,

We haven't heard back from you. Could you please respond to my previous post?

Thanks & Regards,

Shaik Rabiya

Re:dapl async_event CQ (0x1fdb8e0) ERR 0

RabiyaSK_Intel — Wed, 24 May 2023 09:39:40 GMT

Hi,

We haven't heard back from you. If you need any additional information, you can raise a new question as this thread will no longer be monitored by Intel.

Thanks & Regards,

Shaik Rabiya