Discussion Thread for the Continuous GW Search known as O2MD1 (now O2MDF - GPUs only)

..it was several months ago when I first signed up for Einstein and I was doing some testing On different hardware and configs. I noticed that I had good performance (~80% GPU utilization) with the GW GPU app on the system running 10x 2070s with the E5-2680v2 CPUs, but pretty poor performance (40-50% GPU utilization) on the system with 7x 2080s and just the E5-2630Lv2. swapping to the E5-2667v2 pretty much solved that and brought the GPU utilization back up. But I’ve just stuck to running the Gamma Ray tasks since then since they run better.

Ian&Steve C. wrote:

..it was several months ago when I first signed up for Einstein and I was doing some testing On different hardware and configs. I noticed that I had good performance (~80% GPU utilization) with the GW GPU app on the system running 10x 2070s with the E5-2680v2 CPUs, but pretty poor performance (40-50% GPU utilization) on the system with 7x 2080s and just the E5-2630Lv2. swapping to the E5-2667v2 pretty much solved that and brought the GPU utilization back up. But I’ve just stuck to running the Gamma Ray tasks since then since they run better.

Thanks. Interesting results. It seems that for 10 GPUs running with a 10 core E5-2680v2 CPU and 7 GPUs running with an 8 core E5-2667v2, things worked well; it was only when you ran 7 GPUs with a 6 core E5-2630Lv2 that things didn't work so well.

So perhaps the solution is to ensure a full core is available for each GW GPU task. I assume you had hyperthreading enabled, so that tentative conclusion would apply just as well to 2 threads/task. This would fit with my observations on my 2 core/4 thread system.

also not sure if you saw this at the end of the last page. you can track this system if you want. i'll leave it running GW on 2 GPUs until i get the parts i  need to get it on GPUGRID full-time. looks like the fist two tasks ran for 11+ min, but all the ones after ran for about ~9:50, different freq. ...

I actually have two CPUs in that system. 2x 10c/20t with HT enabled so 40 total threads for the system. CPU use is only about 25-27%. lots of space CPU cycles to handle anything the GPUs need. it's overkill for what it's doing now. I used to run CPU work, but I decided to stop all CPU processing as GPUs are way more efficient and CPU work just seems like a waste of electricity anymore.

Maybe you can find another project that will work with a more modest cpu/ram combo while still running E@H Pulsar#1?