Overclock.net › Forums › Benchmarks › Benchmarking Software and Discussion › [OFFICIAL] HWiNFO/32/64 Thread
New Posts  All Forums:Forum Nav:

[OFFICIAL] HWiNFO/32/64 Thread - Page 149

post #1481 of 1525
Thread Starter 
Quote:
Originally Posted by mynm View Post

Is like if it reports only the DPM value, the red highlight value is AfterBurner, the green highlight one is Hwinfo: Warning: Spoiler! (Click to show)
Warning: Spoiler! (Click to show)

Thanks.

Is the value AfterBurner reports the same as Radeon Wattman does ?
Please attach a HWiNFO Debug File of those situations and note what was the correct clock (reported either by MSI AB or Wattman). I will then check all details.
post #1482 of 1525
Quote:
Originally Posted by Mumak View Post

Is the value AfterBurner reports the same as Radeon Wattman does ?
Please attach a HWiNFO Debug File of those situations and note what was the correct clock (reported either by MSI AB or Wattman). I will then check all details.

Wattman report isn't the same as AfterBurner. AfterBurner is reporting the correct clock.

HWiNFO64.zip 328k .zip file

Thanks
post #1483 of 1525
Thread Starter 
Quote:
Originally Posted by mynm View Post

Wattman report isn't the same as AfterBurner. AfterBurner is reporting the correct clock.

HWiNFO64.zip 328k .zip file

Thanks

I see heavy clock fluctuations in the Debug File. Would it be possible to create one at stable clock and tell me also what you believe the correct clock was in that case ?
And shouldn't Wattman be correct rather than MSI AB ?
Edited by Mumak - 5/1/17 at 7:24am
post #1484 of 1525
Quote:
Originally Posted by Mumak View Post

I see heavy clock fluctuations in the Debug File. Would it be possible to create one at stable clock and tell me also what you believe the correct clock was in that case ?
And shouldn't Wattman be correct rather than MSI AB ?

Because Hwinfo is reporting like by steps, the same steps as the DPM cloks and DPM vddc VID. Although Wattman isn't very clear and maybe is right but with a lower polling frequency, I don't know. AB and Hwinfo are using 1000ms polling frequency.

You can se here highlighted that clock is power throttling, but hwinfo only reports three drops to the lower DPM 980mhz although Afterburner reported clock was higher than that:
Warning: Spoiler! (Click to show)

HWiNFO642.zip 180k .zip file

Thanks thumb.gif
post #1485 of 1525
Thread Starter 
I believe both Wattman and MSI AB report clocks that are averaged, so that's why you see intermediate values. Such averaged values are not truly correct, as the clock can change only to preset levels.
HWiNFO on the other hand reports actual values when they were sampled. If you'd use a much higher polling rate in HWiNFO, you'd see more frequent changes.
Not sure which approach is better as both have their advantages/disadvantages. Averaged values better catch frequent changes in clock, while actual values report true clocks at the time of sampling.
post #1486 of 1525
Quote:
Originally Posted by Mumak View Post

I believe both Wattman and MSI AB report clocks that are averaged, so that's why you see intermediate values. Such averaged values are not truly correct, as the clock can change only to preset levels.
HWiNFO on the other hand reports actual values when they were sampled. If you'd use a much higher polling rate in HWiNFO, you'd see more frequent changes.
Not sure which approach is better as both have their advantages/disadvantages. Averaged values better catch frequent changes in clock, while actual values report true clocks at the time of sampling.

Ok thanks for the explanation thumb.gif, I will test with a 100ms polling frequency. Maybe including both clock measurement cuold be better.
post #1487 of 1525
Thread Starter 
Quote:
Originally Posted by mynm View Post

Ok thanks for the explanation thumb.gif, I will test with a 100ms polling frequency. Maybe including both clock measurement cuold be better.

Yes, I'm considering either a switch to choose between methods, or to display both values.
Note that if you want to use a too high polling frequency, you might need to disable other unnecessary sensors, which might take longer time to read out (i.e. Disk SMART).
post #1488 of 1525
Quote:
Originally Posted by Mumak View Post

Yes, I'm considering either a switch to choose between methods, or to display both values.
Note that if you want to use a too high polling frequency, you might need to disable other unnecessary sensors, which might take longer time to read out (i.e. Disk SMART).

OK thanks you very much for all thumb.gif
post #1489 of 1525
Thread Starter 
Quote:
Originally Posted by mynm View Post

Ok thanks for the explanation thumb.gif, I will test with a 100ms polling frequency. Maybe including both clock measurement cuold be better.

The next build of HWiNFO will add an option to choose between clock/utilization measuring methods: "Prefer AMD ADL"
post #1490 of 1525
Quote:
Originally Posted by Mumak View Post

The next build of HWiNFO will add an option to choose between clock/utilization measuring methods: "Prefer AMD ADL"

Perfect, thanks thumb.gif
New Posts  All Forums:Forum Nav:
  Return Home
Overclock.net › Forums › Benchmarks › Benchmarking Software and Discussion › [OFFICIAL] HWiNFO/32/64 Thread