7870 Myst edition (Tahiti LE) question

[canadiantrex]

Just got my new Powercolor [not] 7870 Myst edition (the Tahiti LE chip) card in today and installed it.

Got drivers installed and tested the card, seems to be working find and be a decent overclocker.

I have it hooked up to 2 monitors, so I understand in desktop mode it won't downclock the RAM. My question is, however, is it normal for the card to idle on the desktop at 64*C at 30% fan (1450ish RPM)?

I had a side fan in my case that I just unplugged. Thought it might have been causing turbulence, however the idle temps have stayed the same.

Sitting on the desktop with just Firefox, GPU-Z, and CPUID Hardware Monitor on, my temps are as follows:

FX-8320 - 35*C
Mainboard - 30*C
HDD's 1, 2, & 3 - 32*C
GPU - 64*C



As you can see, I'm not feeding hot air into the case for the other temps to be like this. Should I be concerned? Should I change out the thermal paste on the GPU? Should I just shoot it? :biggrin:
​

 

[chimaxi83]

My 7870 LE idles around 40C at the desktop, on auto fan with no custom profile set. All I've done to it was change out the thermal paste. Make sure that the heatsink screws are all tight, some have been known to be loose, causing poor contact. Case has decent airflow?

 

[canadiantrex]

Yes, airflow is very flowish. :biggrin:

Perhaps Sapphire Trixx screwed up my voltage settings? At 501/1600 clocks for dual desktop, the voltage was well up over 1.1 volts. Bumped it down to 0.973 a few minutes ago and temps have dropped to 53*C and 25% fan on stock profile.

What voltage should this thing be at for desktop clocks?

 

[GaiaHunter]

How hot is it with only 1 monitor or playing?

I have a sapphire 7870 XT and I have to say these cards run on the hot side (although mine hooked to a single monitor idles at around 33C.

 

[canadiantrex]

How hot is it with only 1 monitor or playing?

I have a sapphire 7870 XT and I have to say these cards run on the hot side (although mine hooked to a single monitor idles at around 33C.

I'll go hit some World of Tanks and let ya know. (Though that's not very graphically demanding, heh)

Also, just reset card back to stock settings.

In furmark, stock clocks, it's at 1.104 V VDDC.
On desktop, 501/1500 clocks, it's at 1.152 V VDDC.

Why is it higher on the desktop? My previous Pitcairn 7870 would lower the voltage when the clocks were lower.

*edit*

Just mirrored the displays, clock is now 300/150 mhz, temps are 55*C. Voltage is 1.158

 

[BallaTheFeared]

If you're confident enough, you can flash your card with new clocks and voltage for low power and base/boost.

http://www.techpowerup.com/forums/showthread.php?t=189089


To answer your question, it could be a case of vdroop if you're reading from GPUz.

 

[canadiantrex]

73*C in World of Tanks, about 50% GPU load.

 

[BallaTheFeared]

What was the fan speed at?

BTW can you link what card you have, picture of it?http://redirect.anandtech.com/r?url...duct.aspx?Item=N82E16814131484&user=u00000687

 

[GaiaHunter]

73*C in World of Tanks, about 50% GPU load.

73C with a demanding game is likely normal.
With the GPU@50% maybe it is the fan staying really low.

If you manually set the fan around 40% (from what I've seen seems to be the target for 100% GPU load for that particular model), what temperature do you get?

 

[canadiantrex]

73C with a demanding game is likely normal.
With the GPU@50% maybe it is the fan staying really low.

If you manually set the fan around 40% (from what I've seen seems to be the target for 100% GPU load for that particular model), what temperature do you get?

Doing this now for a match, will report back.

*edit*

Got up to 79*C this time around with fan at 40% GPU load was around 50ish for the whole session.

 

[GaiaHunter]

Doing this now for a match, will report back.

*edit*

Got up to 79*C this time around with fan at 40% GPU load was around 50ish for the whole session.

That does seem high for a 50% load.
There might be something wrong with how the cooler/thermal paste is set.

Try a demanding game if you can or some benchmark even with auto fan.

To be clear, is that with only 1 monitor or both? Is there any difference in temperatures while playing a game with 1 monitor and with 2 monitors?

 

[canadiantrex]

To be clear, is that with only 1 monitor or both? Is there any difference in temperatures while playing a game with 1 monitor and with 2 monitors?

Negatory, same temps.

I ran furmark with it slightly overclocked. Last 15 seconds of the test were with the fan at basically 100% and the GPU mid 80s.

*edit* Hmm, discovered I had voltage locked in Trixx. Changed that and reset fan profile to stock. Ran furmark again, power limit +20. (just changing the power limit boosted the 720p benchmark by 604 points)

GPU hit 88, VRM's hit 104, fan hit 49%.

And yeah, turning down that power limit. My god those VRM's were hot.

 

[Erenhardt]

Negatory, same temps.

I ran furmark with it slightly overclocked. Last 15 seconds of the test were with the fan at basically 100% and the GPU mid 80s.

*edit* Hmm, discovered I had voltage locked in Trixx. Changed that and reset fan profile to stock. Ran furmark again, power limit +20. (just changing the power limit boosted the 720p benchmark by 604 points)

GPU hit 88, VRM's hit 104, fan hit 49%.

And yeah, turning down that power limit. My god those VRM's were hot.

104 for VRM is not even close to hot. Uninstall trixx. Sometimes it causes increased voltage and card not downclocking to idle. Check GPU usage when idling in desktop - someone could mine with your cards

 

[GaiaHunter]

Well if furmark didn't kill your card it should be fine.

 

[canadiantrex]

104 for VRM is not even close to hot. Uninstall trixx. Sometimes it causes increased voltage and card not downclocking to idle. Check GPU usage when idling in desktop - someone could mine with your cards

No, on desktop it's idling at 0% load.

And 104 isn't too hot? I used to mine on my old 7850 and 7870, and I would worry when the VRM's got into the high 70s. :whiste:

 

[BallaTheFeared]

104 is still safe, but it is unreasonable hot.

I don't think you want them any hotter than 120 or 130C forget which, but 104C is hotter than I'd want to see (thou I'd never run furmark).

I would do as chimaxi83 suggested and remove the cooler, replace the TIM, and remount. If your card has a poor application or mounting of the cooler it would increase all temps and power usage because a hotter chip uses more power.

 

[canadiantrex]

Pulled the card and heat sink. Paste looked like crap, dry and crumbly.

Unfortunately, now the pc won't power up. Everything is connected properly, but its like the power supply broke. It's less than a year old, so I doubt it could just die suddenly. Guess I'll be taking it all apart.

*edit*

Poked at it again. The power switch on the back of the power supply got turned off accidentally. Oops.

I hooked back up the side fan on the case, GPU is currently idling at 47*C with the new paste.

 

[canadiantrex]

Been messing with this card some more. Discovered that the difference between mid-high 40's idle and idling in the low 60s is the RAM speed. Stock speed of 1500 MHz it'll idle in the 40s on the desktop. Bump the RAM to 1600 MHz, and over the next few minutes the temps will climb to low 60s, still doing nothing but sitting on the desktop.

Is this normal for these cards?