HWiNFO64 and MSI GX60 3BE 8970M

VAMET

Active Member
Dear Martin

Martin said:
If there's 99% load and HWiNFO has only the main window open (no sensors, no summary active), then this is definitively a bug in the AMD drivers.

Yes, there are 99% load on no matter which kind of window is open. So everyone with AMD GPU has got the same problem? Maybe you have other solution, an idea, how in other way enable GPU, to be detected by HWiNFO64 monitoring purpose? I like this application, but I don't want to be 99% load all the time.

Martin, I believe in you and your experience to solve this problem without fixing drivers by AMD :)

Best regards.

Sincerely
 

Martin

HWiNFO Author
Staff member
No, I don't think all AMD GPUs are affected, since I already tested lots and there were no such issues. The short OpenCL sequence performed is absolutely legitimate and causes no issues on several other AMD and NVIDIA GPUs.
Maybe it's not really 99% load, just some sort of occupying some of the GPU resources.
What temperatures do you see on that GPU when reporting 99% load in HWiNFO and in other GPU-intensive applications ?
 

VAMET

Active Member
Dear Martin

Martin said:
Maybe it's not really 99% load, just some sort of occupying some of the GPU resources.
What temperatures do you see on that GPU when reporting 99% load in HWiNFO and in other GPU-intensive applications ?

Your HWiNFO64 Sensors shows for 8970M:
- 0 minutes = 55'C
- 3 minutes = 65'C
- 4 minutes = 70'C
- 5 minutes = 72'C
- 8 minutes = 74'C

Max Payne 3 (max settings 1920x1080 AA x4):
- >6 hours = 79'C

Far Cry 3 (max settings 1920x1080 AA x0):
- >4 hours 84'C

84'C is the maximum temperature I have ever noticed. So I think that this 99% load is not fake.

Best regards.

Sincerely
 

Martin

HWiNFO Author
Staff member
OK, let's make one more test.
Please get this build: www.hwinfo.com/beta/hw64_437_2175.zip
Make sure the GPU is off. Then launch that build and check if it wakes up and recognizes the GPU. Then continue and check if the GPU goes back into off state.
Please let me know the results.
 

VAMET

Active Member
Dear Martin

Martin said:
OK, let's make one more test.
Please get this build: www.hwinfo.com/beta/hw64_437_2175.zip
Make sure the GPU is off. Then launch that build and check if it wakes up and recognizes the GPU. Then continue and check if the GPU goes back into off state.
Please let me know the results.

So far the best build I think :) GPU is recognized, but in sensors there are only information:

GPU [#1] ATI/AMD Radeon HD 7950M/8970M: ATI Internal
GPU Utilization 0.0%

In GPU [#0] section for 8650G there are many more, but this time 8970M is waking up and then goes off state mode. Why there are no other descriptions in GPU [#1] like:
- GPU Thermal Diode
- GPU VDDC
- GPU Clock
- GPU Memory Clock
- GPU Utilization <= only this exist
- GPU D3D Usage
- GPU D3D Memory Dedicated
- GPU D3D Memory Dynamic

Best regards.

Sincerely
 

Martin

HWiNFO Author
Staff member
When the GPU goes back into off state, HWiNFO doesn't read its parameters to avoid waking it up again. So that's the reason for not showing the other values. Try to wake up the GPU and you should see the rest of values appearing.
But please attach a new Debug File, so I can check that more precisely.
 

VAMET

Active Member
Dear Martin

Martin said:
When the GPU goes back into off state, HWiNFO doesn't read its parameters to avoid waking it up again. So that's the reason for not showing the other values. Try to wake up the GPU and you should see the rest of values appearing.
But please attach a new Debug File, so I can check that more precisely.

You are right, parameters are during playing session, I have already checked it. So everything is alright? :) ...so what have you done? :D Magic, I know... :)

Here you go with attachments:
- 1_HWiNFO64.DBG - Wake disabled GPUs (disabled)
- 2_HWiNFO64.DBG - Wake disabled GPUs (enabled)

Best regards.

Sincerely
 

Attachments

  • 1_HWiNFO64.DBG
    344.9 KB · Views: 1
  • 2_HWiNFO64.DBG
    350 KB · Views: 1

Martin

HWiNFO Author
Staff member
Then it's OK and the DBG confirms it :) I have implemented a workaround and I hope this will not affect other AMD discrete GPUs that are powered off and they will still work as expected. I still believe the true problem was (and actually is) the AMD drivers.
Thanks for all the testing you have done.
 

VAMET

Active Member
Dear Martin

Martin said:
Thanks for all the testing you have done.

No problem at all, I am glad, if I may help to someone :)
Thank you for your great support! :)

Best regards.

Sincerely
 
Top