Furmark stresstest; overdo or clever?

[discy]

I've heared from people (and AMD themselves) that Furmark doesn't represent a real gaming situation beacause it stresses the GPU to a level way beyond what you would expect in games.

Therefore a lot of people use 3dMark loops and such, as they are expected to be a way better representation of playing games.

This made me wonder. Is Furmark really THAT overdone?. I liked to play it safe and just overclock until Furmark crashes;

My fear is that 3dMark won't stress my GPU enough and that it will crash just when I was about to get first @ MW2..but is that fear legitimate?

What would you guys say? Furmark overdone or is it indeed a valid way to play it safe?

 

[SnW]

If you want to stress it (GPU/CPU) properly, use F@H....

sorry in regards to Furmark i can't help ..all i hear is that it gets GPU's cooked

EDIT keep in mind i am not asking you to join the F@H team (would be awesome tho

) i am serious in what this software can do !

 

[dugdiamond]

+1 use F@H

 

[Aurus_OC]

Yea use F@H. I think the newer NVidia cards, also have some kind of limitation on them so it doesn't stress them all the way. However I'm not 100% sure on that.

 

[discy]

If you want to stress it (GPU/CPU) properly, use F@H

Should I just use memtestcl? because that doesn't crash even when my core clock is up to a level it does crash in Crysis 2

.

What am I doing wrong? (AMD 6950)

in regards to Furmark i can't help ..all i hear is that it gets GPU's cooked

In my case, the GPU only gets up to 80 degrees and it lets me find when my GPU shuts down or not. So I would guess it also does some stress testing instead of only cooking?

 

[SieB]

Use Furmark or F@H to test your card, if it doesn't crash in those it's not going to crash in games.

Here is a link to GPU tracker for F@H LINK

 

[Aurus_OC]

Should I just use memtestcl? because that doesn't crash even when my core clock is up to a level it does crash in Crysis 2

.

What am I doing wrong? (AMD 6950)

Just download GPU tracker and test your card by folding on it for some time. That way you can combine the stress testing with helping out science.

 

[SeanBond]

Furmark doesnt test for 100% stable, i had my 460 @ 910mhz core stable in furmark, but when i open Heaven beanchmark it just crapped out and froze

 

[discy]

Use Furmark or F@H to test your card, if it doesn't crash in those it's not going to crash in games.

That is what I'm doing at the moment. But isn't Furmark overdone in comparison to play games?

I won't have to make it stable up to a point I will never reach in games.

 

[discy]

I had my 460 @ 910mhz core stable in furmark, but when i open Heaven beanchmark it just crapped out and froze

I've never heard of such a situation; thanks. Did you run Furmark for 30+ minutes?

 

[Aurus_OC]

That is what I'm doing at the moment. But isn't Furmark overdone in comparison to play games?

I won't have to make it stable up to a point I will never reach in games.

Furmark doesn't test the GPU to a full extend, even though it is showing 100% load. You will want to have a 100% stable overclock, because having a 99% stable overclock is an overclock that is not stable. At some point a game will stress you GPU to that extend. You can also try using the Heaven Benchmark and have it just cycling through over and over. That will work better than testing in Furmark aswell.

 

[SieB]

That is what I'm doing at the moment. But isn't Furmark overdone in comparison to play games?

I won't have to make it stable up to a point I will never reach in games.

That's why it's good to test your card on, if it does not crash under extreme tests like Furmark it won't crash in games. Just get the clock you want test it in Furmark or F@H and then if it does not crash you know no matter what else you do it's not going to crash either. Thats the way I see it anyway.

 

[SeanBond]

I've never heard of such a situation; thanks. Did you run Furmark for 30+ minutes?

Overnight, so 8+ hours. I assumed the card was stable as a rock untill i wanted to do some benches

 

[discy]

At some point a game will stress you GPU to that extend. You can also try using the Heaven Benchmark and have it just cycling through over and over. That will work better than testing in Furmark aswell.

So a heaven benchmark is more demanding than even the newest Furmark?

That's a surprice to me.

If I do such a benchmark; should I use the highest settings or would it be better to let it run on a more playable FPS like 40?

 

[Aurus_OC]

So a heaven benchmark is more demanding than even the newest Furmark?

That's a surprice to me.

If I do such a benchmark; should I use the highest settings or would it be better to let it run on a more playable FPS like 40?

I most often use the settings that max out my configuration. I don't have a 1080p display so the only thing I turn down is the resolution to 1680x1050. But if you have a 1080p display use that res. I use normal tesselation and everything else up to it's maximum.

Edit: Oh and I also prefer looking at the great heaven environment to the turning hairy donut that Furmark displays you

 

[marsey99]

+1 for unigine

it wont take you long for it to find out if your gpu is stable at those settings either, like sean i have tried settings that are stable at everything else crash in its 1st loop of heaven.

 

[adamjamesroe]

I would say funning Prime and MSI Kombuster at the same time should stress the PC enough.