Radeon 7850 VDDC Current Jumps

[ladydub]

Hi everyone, I have a question about GPU-Z and my Radeon 7850.

Thing is that VDDC showings are not constant and there are huge current and voltage jumps.
Please look at this screens and notice how VDDC graphs are not constant
http://imageshack.us/photo/my-images/52/currentjumps.gif/
http://imageshack.us/photo/my-images/687/currentjumps2.gif/
http://imageshack.us/photo/my-images/689/currentjumpmax.gif/

As you can see the highest VDDC Current jump was 255A when playing Witcher 2 with ubersampling (for testing purposes)

I also have a log file with 112, 120, 130, 190 and 200A jumps which seemed to increase as I was running Unigine Benchmark.
There were also lots 1.564V Jumps during Unigine (1.7 max during Witcher...) while its 1.064-1.066 most of the time.

Is that normal? Maybe this is the cause of my blacksreens during gaming when my monitor looses power requiring a hard reset?

 

[Zanovar]

Do you have sandra or something similar running in the background?turn it off *winks*

 

[chimaxi83]

GPU-Z isn't properly reading the sensors, that's all. I get some pretty stupid readings on my 680 too. I normally use HWiNFO64 for hardware sensor reading.

 

[Zanovar]

GPU-Z isn't properly reading the sensors, that's all. I get some pretty stupid readings on my 680 too. I normally use HWiNFO64 for hardware sensor reading.

Although not always the case,i would like to believe this to*coughs*

 

[chimaxi83]

Although not always the case,i would like to believe this to*coughs*

Don't get what you mean, but okay.

 

[Zanovar]

Im guessing varios progs are playing with v control,wasnt having a go mate.

 

[ladydub]

Nope, absolutely everything turned off, even browser. Just played some NFS:HP where GPU load was around 60-70% and temps were 57-60. This session caused no jumpes, vddc current was around 44-45A constantly. Seems like if GPU isnt loaded 100% it runs stable? Thats wrong.

BTW gonna try this HWiNFO64, and compare readings.

 

[ladydub]

While playing smth heavier, for example Saints Row 3 on max settings, GPU is loaded 100% and I get Huge jumps again.

 

[ladydub]

BTW Zanovar cant reply to your PM since I don't have 10 posts yet hehe.

 

[TakeNoPrisoners]

I would say it is just GPU-Z, it is known to show strange readings at times.

 

[Zanovar]

I would say it is just GPU-Z, it is known to show strange readings at times.

possibly,but not always.ive had some crazy jumps in voltge.various progs reading v has caused this.

 

[Cookie Monster]

Software sensors are often rubbish and report the wrong values (or values that aren't very accurate).

Also Im surprised the suppose current reading is so high because normally all GPUs have around vCore of ~1V. Now given the power consumption figures for a HD7850, these shouldn't be drawing 200As (most likely a false reading), since that results in a very very basic approximation of 200W.

These spikes shouldn't be normal since the vCore should be regulated to a fixed voltage (with a tolerance limit).

Id think before looking into this "issue", you'd want to go through a basic trouble shooting procedure to find why its suffering from a blackscreen? d

If there are no issues and you are just paranoid about these "jumps" then im happy to tell you that software readings are quite frankly rubbish.

 

[Zanovar]

Software sensors are often rubbish and report the wrong values (or values that aren't very accurate).

Also Im surprised the suppose current reading is so high because normally all GPUs have around vCore of ~1V. Now given the power consumption figures for a HD7850, these shouldn't be drawing 200As (most likely a false reading), since that results in a very very basic approximation of 200W.

These spikes shouldn't be normal since the vCore should be regulated to a fixed voltage (with a tolerance limit).

Id think before looking into this "issue", you'd want to go through a basic trouble shooting procedure to find why its suffering from a blackscreen? d

If there are no issues and you are just paranoid about these "jumps" then im happy to tell you that software readings are quite frankly rubbish.

*laughs*gcard voltage spikes is very real,though your policy of hide your eyes and not believe it is a sound one in the short term,i would kind of look into it*coughs*.

 

[chimaxi83]

*laughs*gcard voltage spikes is very real,though your policy of hide your eyes and not believe it is a sound one in the short term,i would kind of look into it*coughs*.

Perhaps if you knew a little bit about the subject, you wouldn't *cough cough* your way through the post. Do you really think that a card which normally operates at 1.06-1.1 volts will spike to 1.564-1.7 volts, regardless of load? Do you also think that a card which normally operates at 1.06-1.1 volts will spike to a current of 255 amps, regardless of load? If you do, then there is nothing anyone can say to help you.

I find the software sensor reading of GPU-Z to be blah, to say the least.

 

[ladydub]

When I get home from work I will test with some other voltage monitoring tool, which one would you recommend? I agree that GPU-Z might show things wrong, but still, those showings are not crazy impossible, and seem to be happening only on 100% GPU load.

As for Blackscreens and monitor loosing power, it is a very known problem with 7000 Series Radeons, plenty of topics about this on the internet, and no fix for it. Seems like a driver issue "probably". Majority of ppl have blackscreens when upgrading to drivers newer than 12.4. But I was getting those crashes with any version.
Trying to solve this issue made me look into GPU-Z readings in the first place, and I found those strange spikes. Every other screenshot of GPU readings I saw had VDDC showings constant, only my card has those jumps.
I "hope" that those current and voltage jumps and card powering off with monitor loosing signal is just a coincidense.

 

[Cookie Monster]

*laughs*gcard voltage spikes is very real,though your policy of hide your eyes and not believe it is a sound one in the short term,i would kind of look into it*coughs*.

Theres a difference between being "real" and being "normal". These so called spikes can exist, and whether or not they are intentional is the question at hand. In my experience, any power regulator that I have dealth with simply do not exhibit this behaviour. And often measuring such spikes requires very expensive equipment to detect which leads to believe that the software sensor is just rubbish which they normally are (on video cards anyway).

But regulators that have hardware problems could have these "spike issues" and if it did exist with the OPs card, then the card itself would have died by now due to the hardware damages caused. Putting it into perspective, 1.7V is over 50% of the required output voltage.

Because of how software sensors tend to be inaccurate on video cards, my suggestion to the OP is to forget about it for now and try to trouble the videocard normally e.g. driver re-installation, stability tests etc. Once the OP figures out what the actual problem is, then we can start thinking about what the potential causes are. And by the looks of it, it could very well be a hardware problem and im guessing the only solution is for RMA which im sure the manufacturer will accept.

If you think otherwise, then instead of resorting to some sort of pretty personal remarks, please stay constructive and explain why you think its wrong or false.

@OP: One thing you can also do is swap it out into a different PC. What is your PC spec by the way including the power supply?

 

[ladydub]

Because of how software sensors tend to be inaccurate on video cards, my suggestion to the OP is to forget about it for now and try to trouble the videocard normally e.g. driver re-installation, stability tests etc. Once the OP figures out what the actual problem is, then we can start thinking about what the potential causes are. And by the looks of it, it could very well be a hardware problem and im guessing the only solution is for RMA which im sure the manufacturer will accept.

@OP: One thing you can also do is swap it out into a different PC. What is your PC spec by the way including the power supply?

Specs are: Phenom2x4 965BE, Asus m5a97 PRO, 4GB RAM, PSU is chieftec 550w (kind of average PSU I know). Unfortunately I cant test the card in other PC, although I can put back my old 5770 and see of GPU-Z will give funny readings...

As for blackscreens, yes I tried clean installing different driver versions, I tried so much stuff its not even funny. The card is stable in stress testing with up to 1250 MHZ Core, 3dmark and running Unigine for an hour(kind of lucky with that card, actually bought it for overclocking performance). Blackscreens dont occur in every game and dont depend on clocks, happens both on stock and any frequency up to 1250core. Games that gave me blackscreens are Crysis, Max Payne 3 and Saints Row 3, while never happening in Witcher 2, NWN2, NFS:HP, Rage, Game of Thrones, Warhammer:SM.
Maybe I'm being paranoid and those are driver problems, otherwise it would happen in every game?
Or I should try and return this card and buy a new PSU and GTX 670 and forget all this headache.