| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 14
|
|
| Author |
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
This image shows what I am talking about. I have an i7-980X and 2 GTX 590s with 12GB memory. You can see that even though two of the GPUs are in full usage there is no drop in CPU usage. I have noticed that those talking about CPU usage dropping have both AMD CPUs and GPUs, which leads me to believe that there is something very different between the two type of setups. ![]() The problem is Nvidia driver support for OpenCl. Until that improves you'll see near 100% CPU core usage per GPU task running on any Nvidia graphics card. It's the same whether you're using an Intel or AMD CPU.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
----------------------------------------![]() ![]() [Edit 1 times, last edit by nanoprobe at Oct 22, 2012 1:26:47 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
That's a valid point. Funny is, was reading last night a test of the Intel 3770 [Think it was a long test at Phoronix]. The On-Chip GPU is not supporting OpenCL either.
|
||
|
|
mikey
Veteran Cruncher Joined: May 10, 2009 Post Count: 824 Status: Offline Project Badges:
|
The problem is Nvidia driver support for OpenCl. Until that improves you'll see near 100% CPU core usage per GPU task running on any Nvidia graphics card. It's the same whether you're using an Intel or AMD CPU. AND is exactly what I am seeing at other projects when I crunch with my Nvidia gpu. In fact at DistRTgen it is using 0.939% of the cpu and they have been using gpu crunching for awhile now. The thing to remember is if you just leave one cpu unused in Boinc, the gpu will FAR out crunch anything you think you are losing if you used that cpu for crunching instead. Right now the software isn't efficient enough to allow us to use both at once. PART of that is that we are NOT gamers, gamers do not care about crunching and they are BY FAR the largest consumers of high end gpu's. That means the software is written for them and not for us crunchers. And in gaming the MORE resources you can throw at it the faster the game plays for you and the more chance you have of winning!! We crunchers like to divide our resources but still use them all. Right now it favors the gamers way. ![]() ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for the info nanoprobe. I may just have to pick up an AMD card to dedicate to this project soon.
It's funny how that completely reverses with Folding@Home. With F@H, the nVidia cards have always tended to use very little CPU while folding, but AMD/ATI always used a full CPu core. One thing to note is newer units in folding are showing more CPu use even for nVida cards, and I believe there was a recent shift to move GPU units away from CUDA coding towards OpenCL. This shift may explain the increasing CPU usage these days. |
||
|
|
|