Best way to stability test that isn't furmark?

spleenharvester · 11 Apr 2011 at 20:22

What's the best way to test my card that isn't furmark?

I have an 8800GT with a Zotac cooler on (Like reference but with more vents and a full-metal design), which gets 69*c gaming and 85*c under very heavy load, which is pretty good for an 8800.

I don't want furmark since as some of you might remember, the 8800GT had pretty bad VRMs, I've had one smoke on me personally a few years back.

Would 3 hours of OCCT do it? I wanna make sure it's stable since I'm going to BIOS flash it to those clocks and better fan regulation.

Also, in Nibitor, am I alright to remove the "extra" performance level?

Cheers

chipps101 · 11 Apr 2011 at 20:32

i loop unigine tropics, sanctuary or heaven benchmarks and find that is an excellent test for gfx cards, i wouldn't use furmark for testing cards as all it really tells you is their maximum heat output.

i can run my gtx 460 happily at 850/4200 in furmark and it is fine, anything higher than 820/4100 in heaven and it crashes to desktop not far in.

spleenharvester · 11 Apr 2011 at 20:42

I'll give heaven a shot cheers, how long for though? And do I need to keep watching for artifacts?

Problem is my card is really jekyll and hyde, some days it'll work brilliantly at 785/1900, other days it'll artifact at 750/1700, think it's down to heat.

chipps101 · 11 Apr 2011 at 20:47

when i'm testing i generally loop one of them for at least an hour, keeping an eye for anything while doing other stuff.

i've found that most 8800 gt's default fan settings let the card get far too hot before ramping up the fan, most of the ones i have had get into the 90's before the fan ramps up to get the temps back down.

spleenharvester · 11 Apr 2011 at 22:04

Done and all stable

Trying to flash new BIOS under the Zotac brand, since it's now got a zotac heatsink on and I'm OCD as hell, but it won't let me because of the subsystem ID mismatch. Is there any harm in forcing it? It's the original BIOS, just with modified fan speeds, clocks, and IDs.

Cheers

Yellowbeard · 11 Apr 2011 at 22:09

Are you OCing for games? If yes, play the games for your benchmarking. If it's stable doing what you want it to, then who cares if it's going to run some demo or benchmark tool?

Exsurgo · 11 Apr 2011 at 22:27

It depends on the card. On the 5870s I find Heaven pretty bad for stability testing. Crysis looping shows up instability a good 30Mhz earlier in core clock speed.

Furmark/comp buster is only really effective for pusing VRMs and will usually only fail because of temps/voltage. OCCT is rather effective however it also has a habit of failing due to temps rather than actually core instabiliy.

spleenharvester · 11 Apr 2011 at 22:30

Yellowbeard said:
Are you OCing for games? If yes, play the games for your benchmarking. If it's stable doing what you want it to, then who cares if it's going to run some demo or benchmark tool?

'Cause under certain games, such as newer ones, it could crash if not completely stable. I agree to an extent though, stuff like people Prime95'ing for 24 hours or things like that, what's the point

Anyway force flashed the card and all working brilliantly, fan control is so smooth now. Might reflash with higher clocks later on.

EDIT: That's my problem with OCCT, once my card hits 85*c it loses stability. I've got the clocks at 730/1782/900 right now 'cause they're definitely stable.

Exsurgo · 11 Apr 2011 at 22:37

Thats because VRMs loose current as the temps increase, and at 85c it would seem that they cannot satisfy the core demand.

What are the VRM temps?

Operation · 11 Apr 2011 at 22:51

Yellowbeard said:
Are you OCing for games? If yes, play the games for your benchmarking. If it's stable doing what you want it to, then who cares if it's going to run some demo or benchmark tool?

I agree. I find the best way to get things running nice and stable is to do what I inteded to do when buying the stuff. If it crashes I tone it down a bit and keep doing normal stuff. When it stops crashing when you are doing normal stuff, you're sorted. I really don't get the idea that it's not stable unless it can run some crazy stress software for days on end. Uness you run benchmarks for fun, it seems like a lot of bother to me.

spleenharvester · 11 Apr 2011 at 22:56

If I don't stress test I keep thinking I'm seeing artifacts in game, even though when I stress it afterwards it's perfectly stable. Maybe just OCD again lol.

Exsurgo said:
Thats because VRMs loose current as the temps increase, and at 85c it would seem that they cannot satisfy the core demand.

What are the VRM temps?

Anyone's guess, they show as -128*c.

Mine just failed with one error and I need to figure out if it's shaders or core that's failing the test, rookie mistake

At least the 8800GT Reference VRMs aren't as bad as on a non-reference HIS 4870 I had, christ that was awful. Both cards died once they hit 100% load in Far Cry 2 'cause there was no memory heatsink and the VRMs weren't up to the job :rolleyes:

Yellowbeard · 11 Apr 2011 at 23:46

spleenharvester said:
'Cause under certain games, such as newer ones, it could crash if not completely stable. .

That's the point I was making. That's why I said if. "IF" it is in fact 100% playing the game, then who cares what it does with a benchmark unless you are a benchmark junkie?