VDDC Votages read problem?

[JUZTIQ]

Hi.
Question:
.......... I love using Hwinfo during gameplay - but am I doing something wrong here since GPU VDDC is reported as 1.250 (but only 1.056 in MSI Afterburner and also lower in GpuZ)????

Please help

Facts:
I got 2 x Sapphire HD7950 Vapor-X in Crossfire overclocked to 1100 mhz/1500 mhz.
I have not changed voltages of gpu or memory but only used CCC to change gpu/memory frequency and set Power Setting to +20. (Also I have disabled Ulps in system register).

Problem:
1)
HWInfo reads Gpu 1 & Gpu 2 as VDDC as Idle 0.950 (log) and Load 1.250 (See BF3 screenshot for load)
http://i1356.photobucket.com/albums/q733/JUZTIQ/HWInfo1250GPUVDDC_zps8c5bec8f.png

I have log to file and its the same in the log.

2)
MSI Afterburner reads Gpu 1 VDDC as Idle 0.806 and Gpu 2 as Idle 0.949 (log) and boths Gpus as 1.056 under load (See BF3 screenshot for load)
http://i1356.photobucket.com/albums/q733/JUZTIQ/MSIAfterburner1056GPUVDDC_zpsb8223d54.png

I have log to file and its the same in the log.

3)
Gpu-Z reads Gpu VDDC idle as stable but VDDC load is all over the place - I dont have a screenshot but the log file is like this:

Gpu 1:
Idle: 0.803
Load: lowest 0.500 and highest 1.496 (but mostly like this 1.037,1.033,1.021,1.016,1.014, 1.004, 1.006 etc. etc.)

Gpu2:
Idle: 0.941
Load: lowest 0.502 and highest 1.496 (but mostly like this 1.037, 1.037, 1.016, 1.021, 1.027, 1.006, 1.006 etc. etc.).

 

[Martin]

I need to check this deeper. For that I need the HWiNFO Report File and Debug File including sensor data of both situations (idle and load). So please generate these 4 files, name them according to the situation (idle/load) and attach here. If the files are too large, just pack them (ZIP, RAR, etc).

 

[JUZTIQ]

[attachment=536][attachment=537][attachment=538][attachment=539]

Hi again.

I did my best to attach the info you asked for..... but I am no expert at this.

Also - I had trouble making a debug under load as I had to go to "not full screen".

Hope the attached info is what you need.

Thnx in advance.

PS: If run a program to get load on Gpus and the start HWInfo debug - and then press SENSOR to open sensor window - and then press SAVE REPORT........ I get chrash with bluescreen.

I need to check this deeper. For that I need the HWiNFO Report File and Debug File including sensor data of both situations (idle and load). So please generate these 4 files, name them according to the situation (idle/load) and attach here. If the files are too large, just pack them (ZIP, RAR, etc).

 

[Martin]

Thanks for the data. I just need one more thing to be sure, since there seem to be multiple sensors on your GPUs providing various values. Could you please also attach a screenshot of the sensors window showing all sensor values for GPUs and please tell me (or mark it in the screenshot) which one is the voltage value returning inconsistent values ?

And one question - have you maybe checked the "GPU VRM Voltage Out (VOUT/VID)" values under the CHL8228 GPU sensors? Might these provide the values which are consistent with other tools ?

 

[Martin]

And one more advise - in case you see the GPU VR values fluctuating too much, try to run only one of the monitoring tools at the same time. Running multiple such tools (i.e. HWiNFO and MSI Afterburner or GPU-Z) concurrently might cause such issues.

 

[JUZTIQ]

Thanks for the data. I just need one more thing to be sure, since there seem to be multiple sensors on your GPUs providing various values. Could you please also attach a screenshot of the sensors window showing all sensor values for GPUs and please tell me (or mark it in the screenshot) which one is the voltage value returning inconsistent values ?

And one question - have you maybe checked the "GPU VRM Voltage Out (VOUT/VID)" values under the CHL8228 GPU sensors? Might these provide the values which are consistent with other tools ?

Hi Martin.

1)
You say " have you maybe checked the "GPU VRM Voltage Out (VOUT/VID)" values under the CHL8228 GPU sensors?"
.......... I think you are spot on on this! - This looks like the value of Gpu Z.

2)
See this for sensors marked with red:

http://i1356.photobucket.com/albums/q733/JUZTIQ/Sensorswitharrows_zps0398a177.png

HWInfo gives a very consistent GPU Vcore (VDDC) value - but as mentioned the stock voltage is reported by HWinfo as 1.250 under load (But according to MSI Afterburner and Sapphire Trixx stock voltage is 1.056.......... so why is HWInfo reporting 1.250 ?)

NB:

I read this:
http://www.techpowerup.com/forums/showthread.php?p=2787117

See post number 38

"I found the culprit. When the amd's gpu boost was working it also increased the gpu voltage up to 1.25v which caused heating up. As that happened card also hit the power limit and lowered the gpu clocks as well resulting worse fps and massive heat. I turned on the cards gpu boost button which essentially disables the amd gpu boots and increases the clocks to fixed 950MHz. After this the gpu speed remains constantly at 950MHz with 1.056v and temperatures below 65C with auto fan. Seems amd's driver based gpu boost needs still some work."


??? - Could this be it?

I took screenshot with no overclocking and CCC Power Setting to 0:

And got this - where you can see that the voltages "fluctuates"

http://i1356.photobucket.com/albums/q733/JUZTIQ/Gpuvoltagechanges_zps2b03c3b0.png

CONCLUSION:

HWInfo is a great tool and the one I always use - I just want to find out what stock voltage is (everyone says 1.056 and they raise it manually to 1.175 or so when overclocking - thus I just wondered why HWInfo says mine is 1.250 when I have not change voltages.............. BUT then maybe it is AMD Bios (Boost) or when CCC Power Setting is on +20 .... or a combination of the 2?) - I am going insane

Thnx for you help and a great tool for montering

 

[Martin]

Don't worry I think I know what's happening there and going to explain this:

The GPU VDDC reported by HWiNFO is the VID value coming from AMD drivers. This is the voltage value that the AMD GPU/driver requests and thinks it gets.
But... in reality the real VID which is supplied to the GPU comes from the Voltage Regulator - the CHL8228 chip which is also monitored by HWiNFO. So the "GPU VRM Voltage Out (VOUT/VID)" value is the real actual value of GPU Vcore supplied to the card. And they are different because as described at TPU.

This situation is similar to CPU VID and Vcore. VID is the requested value and Vcore is the monitored real value.

Moreover to understand the CHL8228 values: there are 2 instances of this chip per GPU. The 1st one manages the GPU Vcore and the 2nd one most probably the GPU memory voltage. They also allow you to monitor current and power consumption of both phases (GPU Core and GPU Mem).

I hope this clears the situation a bit Just let me know if you have any questions or think there's something wrong ...

 

[JUZTIQ]

That seems to be correct - so I have added "GPU VRM Voltage Out (VOUT/VID)" to my monitoring as well.

Thank you again for you help.

NEW QUESTION::blush:

When I log onto my PC - start MSI OSD server - then start HWIinfo sensor reading - sensor reading starts - I then minimalize - start BF3 (game) and I get perfect OSD reading in game.

..... BUT

If I then exit BF3 game - exit HWInfo - exit MSI OSD Server (so I am back to where I started when I logged on originally - but I do not re-boot)..... and then start MSI OSD Server again - start HWInfo again - start sensor reading again...... then HWI freezes at ATI sensor numer 65 everytime and totally freezes my PC and I need to re-boot.

Anything I can do?

(This means that everytime I make changes in the sensors after I have played a game with HWinfo open and then make changes in the sensors I want in the OSD and press quit and save - then I have to reboot ....otherwise I get freeze).

PS:
In my log file sensor number 65 is "chassis fan number 2"....... or is this not valid informaition.

 

[Martin]

Please try the following procedure:
Go into Configure screen of HWiNFO and select the SMBus/I2C tab.
In the "GPU I2C Bus Exclusion" check all boxes except number 6.
Now check how it works...

If that doesn't help, then please attach the debug file which captures the freeze, so I can see more precisely what happens there.

 

[JUZTIQ]

I checked the boxes but it didnt help...

Here is the debug file of the crash

[attachment=540]

 

[Martin]

Thanks. Please go to the Safety tab and disable the "Audio Codec Check" option.
Let me know if that helped, if not then please attach a new Debug File.

 

[JUZTIQ]

Hi again.

I disabled the "Audio Codec Check" option - but still gets a freeze.

Now it stops at GPU ATI number 89

Debug file of chrash is here:

[attachment=547]

Thanx in advance.

 

[Martin]

If you disable GPU I2C bus 6 too, does that help?
Disabling that will cause you too loose the GPU VR sensors, it's just to prove if the problem is there...
Also, if you're running any other system monitoring/control tools like MSI Afterburner, GPU-Z, try to close them if that maybe has influence on this problem.

 

[JUZTIQ]

I disabled GPU I2C bus 6 too - but still I got freeze

I also only started HWinfo (normally I start MSI OSD Server and then HWInfo)

I do not use any other monitoring tools:

Screenshot of processes are here - dont know if that helps you?

http://i1356.photobucket.com/albums/q733/JUZTIQ/Processes_zps5976f562.png

Freeze with MSI OSD + HWInfo Debug file is here:

[attachment=548]

Freeze with only HWInfo Debug file is here:

[attachment=549]

 

[Martin]

Thanks for the additional tests. This is quite strange, since it seems everytime the crash occurs at a different place. So you probably can enable back the options I suggested to disable, because they don't seem to be the cause.
Sounds more like a stability issue... Does this always happen only after starting and quitting BF3?
Maybe try to upgrade the GPU drivers to the latest version (Catalyst 13.1) ?

 

[JUZTIQ]

Hi Martin.

Thank you for your effort - Maybe I try CCC 13.1 or else I just reboot after changes - when I have set up OSD I rarely make changes anywas.

Have a nice week-end

 

[Martin]

OK, let me know in case you find something more..
Anyway, at least to be able to monitor the real GPU VDDC value please do not forget to enable the GPU I2C bus 6 back.
Have a nice weekend too.

 

[JUZTIQ]

OK, let me know in case you find something more..
Anyway, at least to be able to monitor the real GPU VDDC value please do not forget to enable the GPU I2C bus 6 back.
Have a nice weekend too.

UPDATE - Problem Solved

Hi Martin.

My problem described in post number 8 above is totally gone now

What I did was:

install AMD CCC 13.2 Beta
install HWINFO 4.12 (latest version)
change CAS latency on RAM from 9 to 10 (eventhough its rated CAS 9)

..... and then HWINFO works - and I can now open/configure/save before/at thes ame time/and after other programs.
(But which of the 3 above changes that made it work - i dunno?)

 

[Martin]

That's interesting Thanks for the feedback.
I guess that most probably the CCC update fixed it.