What's a good way to stress test a video card overclock?

[rivethead]

I've built a new system (see specs below) and am working out the kinks.

One of the things I'm trying to do is overclock the internal video card (Radeon 4200). I've oc'd the sideport memory frequency to 1700mhz no problem. Every review and thread I've read about this motherboard/vc says that's no problem.

I've also oc'd the vc frequency too. The base frequency is 500mhz. But some have been able to oc this to 900mhz. Some reviews say 800 is as high as they can go.

I realize that all overclocks are different and ymmv. When I set the vc frequency to 750mhz after a few minutes my screen will flicker and the ATI video driver will error and self recover. I've since lowered the frequency to 700mhz and haven't experienced any symptoms.

However, I haven't stress the vc in either scenario, other than running the Windows Experience assessment.

Other than trial and error, is there a good way to stress the video card? The only thing I could think of is to load "Spore" which is a game my kid has....I think it's the most graphically advanced game we have. But I really don't want to load this game just to stress the vc (he's not currently playing it and the game has some digital media rights restrictions associated with it...I think we can only load it 5 times or something).

Are there any "Intel Burn Test for video cards" type applications out there?

 

[Dark4ng3l]

Use furmark http://www.ozone3d.net/benchmarks/fur/

But be warned furmark is so hard on your graphics that it could potentially damage it if it has insufficient cooling(but thats why it's the best stability test bar none).

 

[rivethead]

I only see 32-bit versions of Furmark. I need to update my signature, but I'm running 64-bit W7 Home Premium. Will that be an issue?

 

[Dark4ng3l]

No everything 32 bit will run. The only thing you need is 64 bit drivers. 99% of the programs you use are 32 bit (maybe even 100%)

 

[rivethead]

Cool. Thanks for the suggestion and follow up!

 

[betasub]

Does OCing integrated graphics make a significant difference? A few points in a benchmark, or enough to get playable framerates on a particular game? Can any on-board chip really be pushed that far?

 

[Arkaign]

Does OCing integrated graphics make a significant difference? A few points in a benchmark, or enough to get playable framerates on a particular game? Can any on-board chip really be pushed that far?

IMHO, overclocking onboard video is a bad idea, gains of even 20% (that would be a rather extreme range) would still be too slow to compare to even bargain-bin new GPUs, and the chips on mainboards are usually only marginally cooled by heatsinks. Of course premium mainboards have better cooling for onboard chips, but those type of boards usually have a decent PCI-e video card stuck in there.

For reference, I just bought a 512MB GT210 for a client that needed dual-monitors, and it was $29.99 brand-new at Micro Center. It's obviously not a gamer's card, but even the 210 is better than virtually any onboard GPU.

 

[rivethead]

Does OCing integrated graphics make a significant difference? A few points in a benchmark, or enough to get playable framerates on a particular game? Can any on-board chip really be pushed that far?

The Tom's Hardware 785 Mobo Roundup article shows that ocing the onboard vc can have a noticeable impact:

http://www.tomshardware.com/reviews/amd-785g-motherboard,2441-21.html

But to be totally honest, I'm overclocking my VC....only because I can. I don't play a lot of games and I don't really need to oc the vc, but like all overclocks, there's a thrill in doing it!

 

[MagickMan]

OCCT has separate GPU tests for core and memory, give it a shot too. Like others have said, though, be careful OCing integrated video.

 

[konakona]

IMHO, overclocking onboard video is a bad idea, gains of even 20% (that would be a rather extreme range) would still be too slow to compare to even bargain-bin new GPUs, and the chips on mainboards are usually only marginally cooled by heatsinks. Of course premium mainboards have better cooling for onboard chips, but those type of boards usually have a decent PCI-e video card stuck in there.

For reference, I just bought a 512MB GT210 for a client that needed dual-monitors, and it was $29.99 brand-new at Micro Center. It's obviously not a gamer's card, but even the 210 is better than virtually any onboard GPU.

Case in point, I am using a HD3300 in 790gx for a 6year old MMORPG. It was quite sluggish and would be plagued with constant stutter at the default 400mhz, but raising it up to 466mhz made it very playable barring certain weather effects and newer map segments with higher shader requirements. 500mhz was a bit unstable so I was forced to back down to 466mhz, but it did not compromise stability of the entire system (overclocked to 3.8/2.8 too). That was with the standard dinky heatsink that came with the biostar mobo.

I might be in the minority, but this is my hands on experience for whatever that's worth. A well-informed prospective user is not likely to expect an IGP to work well enough for any modern games. For the older games where IGPs can still be relevant, OC'ing may help quite a bit.

 

[tweakboy]

The best way to test your OC is to play a couple of your games, if you don't see artifacts in any of them or the system doesnt crash and each game plays for 1 hour each.

I mean you can run 3dmark and you might get artifacts, but that doesn't mean squat. Its real world engines and games you want to test your OC on. Play Crysis or UT3 or Gears or MW2 a big engine game and if it runs for 1 hour straight without artifacts, then your core OC was good your shader is good and your memory is especially good, since you don't see artifacts.

 

[rivethead]

A well-informed prospective user is not likely to expect an IGP to work well enough for any modern games. For the older games where IGPs can still be relevant, OC'ing may help quite a bit.

That's pretty much the conclusion of the Tom's Hardware review that I linked. Although they show benchmarks for Left 4 Dead. Isn't that a new(er) game?

 

[handyrandyrc]

I vote for the "wrap PC in pillows, sleeping bags, and duct tape" method for stress-testing. If it can make it through that, it'll survive anything!

 

[konakona]

yeah well, tbh I did not click on the link and posted that reply just to throw in my 2 cents those results look decent, except the res might be a bit too low by today's standards; I am running this game at 1600x1200 with a quite playable fps. Combine that with the fact that my main rig sports a 1080p TV, the resolution which the game doesn't support on ATi cards. 1600x1200 with an IGP actually works better for me than it would have on a 5850 or whatever (not for fps, but simply cuz of supported res problem + monitor type)

 

[dark0anima]

Case in point, I am using a HD3300 in 790gx for a 6year old MMORPG. It was quite sluggish and would be plagued with constant stutter at the default 400mhz, but raising it up to 466mhz made it very playable barring certain weather effects and newer map segments with higher shader requirements. 500mhz was a bit unstable so I was forced to back down to 466mhz, but it did not compromise stability of the entire system (overclocked to 3.8/2.8 too). That was with the standard dinky heatsink that came with the biostar mobo.

I might be in the minority, but this is my hands on experience for whatever that's worth. A well-informed prospective user is not likely to expect an IGP to work well enough for any modern games. For the older games where IGPs can still be relevant, OC'ing may help quite a bit.

i didnt know you can overclock the IGPs O_O

 

[Hyperlite]

i didnt know you can overclock the IGPs O_O

yep, usually just a bios setting...