Jump to content
  • Advertisement
Sign in to follow this  
kerryhall

What exactly is the processor doing when running a video benchmark?

This topic is 2623 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I have an AMD 64 3000+ with 4 gigs of RAM, and a Geforce GTX 460. I recently installed Windows 7 on there, and the Unigen Heaven benchmark. Running it, I'm getting mediocre frame rates for my card. (7 fps) When I run the process manager, I see that my CPU usage is at 100%. Here is my question: What exactly is the benchmark doing that is using the processor? Shouldn't the video card be doing 100% of the work? There is no physics (which is done on the video card anyway!), no AI, no networking, nothing that I can think of that would use the processor. Any ideas?

Thanks!

Share this post


Link to post
Share on other sites
Advertisement
Something has to be sending the instructions to the video card; that takes some CPU load. For very good video cards and low load, you can actually be CPU bound even in a benchmark if the GPU can draw faster than you can ask for new frames.

Other than that... depends on the benchmark. I'm not personally familiar with the one in question so the best I could offer would be pure speculation.

Share this post


Link to post
Share on other sites
That does seem a *bit* low, but there are tons of things that a CPU can be doing during a 3D scene -- visibility calculations, occlusion tests, general "scene management". Honestly, you have a rather lop-sided system going on. That CPU is circa... what, 2004 or so, and single-core at that? Pairing it with a relatively top-end video card is either going to under tax the GPU or overtax the CPU in real-world applications and benchmarks alike.

Share this post


Link to post
Share on other sites

Something has to be sending the instructions to the video card; that takes some CPU load. For very good video cards and low load, you can actually be CPU bound even in a benchmark if the GPU can draw faster than you can ask for new frames.

Other than that... depends on the benchmark. I'm not personally familiar with the one in question so the best I could offer would be pure speculation.


So if I have a bunch of OpenGL calls, the processor has to send those to the video card each frame? What is the throughput on this, theoretically? Do you think that is what the bottleneck could be?

Is there a way I could run some kind of profiler to see how many calls are done per frame, and what percent of the CPU time is spent on what instruction? I realize the instructions will just be x86 asm, but I don't mind sifting through that. I am really curious now.

Share this post


Link to post
Share on other sites
Who manufactures your card? You can usually download some kind of program to tell you how much of your gpu you are using.
If you are not using all of your gpu your processor is the bottleneck for that test (and probably for 3d applications).

Share this post


Link to post
Share on other sites

I have an AMD 64 3000+ with 4 gigs of RAM, and a Geforce GTX 460....
How did you manage to assembly that? I'm surprised the power lines on the mainboard didn't blow out or something.
What's the whole point? A dual processor with a mediocre card would have likely served you much better...

Anyway, I'd check the renderer first. Maybe drivers didn't install correctly and you're going through the software path.

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!