Seti@Home optimized science apps and information
Optimized Seti@Home apps => Discussion Forum => Topic started by: chronek on 18 Apr 2009, 08:21:09 am
-
Hello
I wondering if is any tool to monitor GPU load ... I didnt find yet , and riva tuner dont have for nvidia that , but maybe someone know where to look for it
-
You can use GPU-Z (http://www.techpowerup.com/gpuz/)
It does for Nvidia and ATI.
-
i don't saw sensor of gpu load , i saw speed core and memory and temp. , but it not show any % of all shaders used and overall load ...
-
I use GPU-Z 0.3.3
see picture--> gif-file attached.
My GPU is not active, so it shows zero.
Perhaps you must click on the "GPU Load" field and enable it.
[attachment deleted by admin]
-
GPUZ only montiors GPU load on ATI cards.
-
I use also GPU-Z 0.3.3.
But I'm little confused about the view.
I have two GTX260 Core216 installed and in both GPU overviews I see:
GPU mhz
MEM mhz
GPU temp
PCB temp
GPU temp
PCB temp
fan %
fan RPM
[..not like you _heinz.. ;)]
And different temps. :o
The first temp is the real GPU temp?
And what's the voltage transformer temp?
And is there really a temp sensor on the back side of the GPU?
-
try EVGA precision... much better than gpu-z
-
also... can someone answer why the GPU load is usually just on 80%? benchmarking apps and games tend to increase it to 98%... wouldn't some optimizations benefit the crunching speed?
-
also... can someone answer why the GPU load is usually just on 80%? benchmarking apps and games tend to increase it to 98%... wouldn't some optimizations benefit the crunching speed?
Certainly would ;)
-
Which would be a normal GPU load in a GTX280 processing Seti enhanced 6.09 beta? Mine is 70-80%
-
Hi Edboard,
mine running 2 astropulse is:
91%
if I screw it higher than 95% the system will be sluggish...
With 90% load I can use my system as a normal PC and working with it.
Everybody must self decide what he will do.
-
I've always used EVGA Precision, but nice to know there are options.
-
How can I increase the GPU load? My GTX 275 card's GPU load is barely ever over 50%. Am I wasting the other half of my GPU?
-
Try to run 2 CUDA MB tasks at once (set coproc value to 0.5 instead 1).
But it could not help at all. Unfortunately, some kernels in current CUDA MB implementation really can't load all available SIMD engines with work, especially if number of those engines pretty high one.
-
If I may speak :)
I noticed also this phenomen on Windows. But what is difference: on Windows 7 max load is about 75% , but on Wind XP load can go high as 95%, and yes Win XP compute much fasten then Win 7. Does compute speed is associated with GPU load I dont know, but it looks like that different kernels ( Win 7 and Win XP) with different drivers can produce on same machine huge difference in speed. After one whole day of benchmarking I down crunch time over 15% and in same cases over 30% translating to Win XP and using older drivers.
Try to run 2 CUDA MB tasks at once (set coproc value to 0.5 instead 1)
This will make your GPU load fully operated, but I doesnot noticed at all speed difference. But on the other side it does make your video sluggish :(
-
Interesting observation.
For CPU apps Vista's kernel is faster than XP's one, I checked it on AP tasks on my netbook. Vista very slow in user interface part on that PC, but crunch time is better...
-
Interesting observation.
For CPU apps Vista's kernel is faster than XP's one, I checked it on AP tasks on my netbook. Vista very slow in user interface part on that PC, but crunch time is better...
That is for CPU, but what is for GPU? I doubt that Vista can beat XP in crunch time. The only thing I didn't try is Win 98: if logic is correct then crunch time will be much better then WP crunch time :)