Re: [OMPI users] Horovod Performance with OpenMPI

2025-06-04 Thread 'George Bosilca' via Open MPI users
Please ignore my prior answer, I just noticed you are running single-node. In addition to Howard's suggestions, check if you have nvlink between GPUs. George. On Wed, Jun 4, 2025 at 10:11 AM George Bosilca wrote: > What's the network on your cluster ? Without a very good network you > canno

Re: [EXTERNAL] Re: [OMPI users] Horovod Performance with OpenMPI

2025-06-04 Thread 'Pritchard Jr., Howard' via Open MPI users
" Subject: [EXTERNAL] Re: [OMPI users] Horovod Performance with OpenMPI What's the network on your cluster ? Without a very good network you cannot obtain anything closer to the single GPU, because the data exchanged between the two GPUs will become the bottleneck. George. On Wed, J

Re: [OMPI users] Horovod Performance with OpenMPI

2025-06-04 Thread 'George Bosilca' via Open MPI users
What's the network on your cluster ? Without a very good network you cannot obtain anything closer to the single GPU, because the data exchanged between the two GPUs will become the bottleneck. George. On Wed, Jun 4, 2025 at 5:56 AM Shruti Sharma wrote: > Hi > I am currently running Horovod

[OMPI users] Horovod Performance with OpenMPI

2025-06-04 Thread Shruti Sharma
Hi I am currently running Horovod benchmarks in an intra-node setup. However, I have observed that increasing the number of GPUs does not result in a proportional increase in total throughput. Specifically, the throughput per GPU with a single GPU is approximately 842.6 ± 2.4, whereas with two

[OMPI users] Horovod Performance with OpenMPI-5.0.6

2025-06-04 Thread Shruti Sharma
Hi I am currently running Horovod benchmarks in an intra-node setup. However, I have observed that increasing the number of GPUs does not result in a proportional increase in total throughput. Specifically, the throughput per GPU with a single GPU is approximately 842.6 ± 2.4, whereas with two