Using a PCI-E Connector

krooton · 1 May 2012 at 21:36

What drivers are you using?
Have you set a custom profile in Afterburner for the fan?

I'm busy looking in to if Catalyst can cause conflicts with Afterburner if certain settings are used.

Drake5 · 1 May 2012 at 21:42

krooton said:
I assumed it was just a bit of dust, and didn't want to try and wipe it for fear of staticing the card in to an early grave. I did want to give it a compressed air blast, but don't have any

Really doesn't look like dust tbh, it's spread over the back of the card a fair bit.

I've messed about with the fan profiles in afterburner to no luck, as said though even running the GPU at 100% fan speed seems to have zero effect on the temps which is really strange.

Here's a pic of the inside of my case, excuse crappy camera quality. I don't have the best cable management but I don't think it'd cause any issues either.

Cheers

Using latest 12.4 Drivers.

krooton · 1 May 2012 at 21:46

Is that using Furmark only?

Can you tested with a few 3D games please? Furmark seems to be a bad way to test things from what I have been reading, cards are throttling it at the drop of a hat because of the load it puts on cards.

Are you getting throttling/performance loss in games?

And what is your ambient case temperature? Use Speedfan or HWMonitor

Drake5 · 1 May 2012 at 21:48

I haven't had throttling in games yet, but the temps running at 87c after under 20 minutes is a bit concerning I think.

Drake5 · 1 May 2012 at 22:09

Nothing comes up in HW Monitor or speedfan about ambient temps.

Cheers

krooton · 1 May 2012 at 22:12

In HW Monitor there is a section with SYSTIN, that is the ambient temp in the case.

What temps were you getting before you overclocked?

I don't want to be funny, but I never overclocked the card and never got temps as high, and as it is within the 90ºC specification, I don't think XFX are going to be considering it to be a warranty fault.

Overclocking and then pushing a card to it's limit with Furmark may have caused an issue?

Drake5 · 1 May 2012 at 22:17

Nope, still don't see SYSTIN.

I see airflow, it says 26c.

Temps were the same idle before I overclocked, I know that much. I didn't let Furmark run longer than 7 minutes.

What else can I do now? Card is running stupidly hot, don't even want to risk gaming on it to be honest.

krooton · 1 May 2012 at 22:24

Idle temps aren't affected my overclocks, only load.

I'm just concerned that by pushing it to the max overclock for the card straight away and running furmark could have damaged something, maybe pushed the VRMs too much.

Furmark stresses a card more than any game does, hence my concern. If it's running under 90ºC, whilst not ideal, it is still performing with specs.

Drake5 · 1 May 2012 at 22:29

krooton said:
Idle temps aren't affected my overclocks, only load.

I'm just concerned that by pushing it to the max overclock for the card straight away and running furmark could have damaged something, maybe pushed the VRMs too much.

Furmark stresses a card more than any game does, hence my concern. If it's running under 90ºC, whilst not ideal, it is still performing with specs.

If that was the case there would be more issues than heat though?

Just a bit gutted really, don't see anything on Google about other 5850's running so hot. Wondering if that stuff on the back of the card has anything to do with it?

krooton · 1 May 2012 at 22:40

As I've said, the card has been working fine in my machine for ages at stock with lower temperatures.

I'm certain it's dust, I don't see how it could be anything else as the card has been sitting in my machine since I bought it.

Again, I'm not trying to be funny, but you pushed for a maximum overclock right off the bat, so I can't vouch for the card any more. It's running within specified temperatures, so I don't really see what more there is to do.

It still has a few months warranty left should anything happen, but I am sure it wont.

Sorry if this seems harsh, but it is working within the specs, I don't really see why a return would be warranted.

Drake5 · 1 May 2012 at 22:46

Well it's not like I expect you to say the card has been running hot in your PC aswell, let's be honest you wouldn't admit that anyway.

I just don't see even if it was because of the overclock the heat would be suddenly effected.

Whatever it is, I'm certain it's not dust, I know what dust looks like.

It's actually stuck to the surface, looks grounded in. If I had seen it in pictures before, I wouldn't have purchased the card.

When you sold the card you said the egg cooler cooled even better.

Drake5 · 1 May 2012 at 23:09

double post

Drake5 · 2 May 2012 at 02:03

It's been a late night, I think I came to the bottom of it though.

As I suspected Krooton probabaly wouldn't honour a return, so I decided to take the heatsink off the GPU.

To my suprise, that was virtually next to no thermal paste on the GPU.

Anyway, I applied new thermal paste and got as much of the little specs of dust from the heatsink as I could.

Loaded back up, temps were LOADS better. I created a proper fan profile in Afterburner which has helped a great deal to.

Before, I couldn't run Furmark for longer than 2 mins without the temps going to 90c. Now I just completed the full burn in test (15 mins), temps were still high but they didn't peak anymore than 86c. ) This was with 775/1125 OC To.

Running Tribes for a fair bit, 99% GPU Usage, 67/68c max.

Fan profile (You guys might think it looks dodgy, but it works well. Any tips for improvement on it though?)

Finally glad to have it all sorted. I'm hoping it stays fine as I would have been royally screwed without a graphics card, last pennies went on the cost of it!

I am still concerned as to what the white stuff is on the back of the card though... So if anyone has any ideas, please let me know.

So would you guys agree these temps are much safer to run at now?

Cheers

Rusty0611 · 2 May 2012 at 06:28

Temps look fine now. As I said before: I wouldn't pay too much attention to Furmark temps at all.

Blue160 · 2 May 2012 at 08:22

Unigine Heaven is a much more realistic test for graphics cards. Use this while monitoring temperature with afterburner and see if you genuinely gave an overheating issue. My SLI setup overheats if I run Furmark, but I can run Heaven or Metro 2033 for hours without going over 80C.

Edit: didn't read whole thread :rolleyes:

glad it's sorted

Wazz-1 · 2 May 2012 at 08:50

Blue160 said:
Unigine Heaven is a much more realistic test for graphics cards. Use this while monitoring temperature with afterburner and see if you genuinely gave an overheating issue. My SLI setup overheats if I run Furmark, but I can run Heaven or Metro 2033 for hours without going over 80C.

Edit: didn't read whole thread glad it's sorted

+1
Temps look fine now, forget about Furmark its a nasty test for any card, as above run Heaven or play your games for couple of hours to test your card/Oc.
The white stuff i wouldnt worry about too much (looks like someone got a bit card happy :rolleyes:

) wipe it off and enjoy your card

krooton · 2 May 2012 at 09:00

So it looks like it's XFX to blame for applying TIM like noobs, glad you got it sorted.

cmndr_andi · 2 May 2012 at 09:39

Great to hear you figured out the issue

Sometimes these graphics card manufacturers do cheap out on the thermal paste - which seems a bit mad considering how cheap it is.

Those temps now look perfectly safe - so now time to enjoy some games

Drake5 · 2 May 2012 at 11:37

krooton said:
So it looks like it's XFX to blame for applying TIM like noobs, glad you got it sorted.

Would have been nice if the blame wasn't pointed straight to me.

Really surprised you didn't notice the high temps.

Thanks for the help / advice guys, appreciate it.

Will run some unigine later..

Drake5 · 2 May 2012 at 12:27

Just ran Unigine on max settings for about 30 mins, 99% GPU Usage, pretty solid 60C/61C.

Really pleased. Could someone tell me why tribes gets my GPU hotter than Unigine though?

Thanks