*** 25/02/2011 EXTENSION! *** DEAL OF CENTURY: Asus GeForce GTX 480 1536MB Graphics Card with 2 FREE

[Warpy]

very tempted...... must resist!!

 

[^Andy^]

A word about temps.

It seems all these ridiculously high load temps on a 480 are reported using Furmark.

Now, I have a GTX 280. In the 2.5 years I've owned it I have never found the fan noise at load to be anything like as bad as people say it is. I've also never found it to be particularly hot (ie almost 100c hot) under load. You can imagine my suprise then when I see all the crying about the GTX 480 only to find the stats being bandied about are about the same as those for my 280.

So, I downloaded Furmark and ran it.

Wow. Load temps of almost 90c and the fan SO loud I could hear it from outside the room. Never before have I ever seen temps as high as this or the fan spinning as loud as this.

It seems Furmark places HUGE stress on the video card - more so than any normal game.

I'm going to record temperature data when I play BC2 tonight. I suspect that the ridiculous load temps are the product of simulated benchmarks and simulated benchmarks only, and are in the real world therefore a non-issue.

Whilst you are at it record the gpu load % too.

 

[[TW]Fox]

Any tips on useful software for recording this?

 

[Zefan]

Riva tuner should do the job.

 

[setter]

Msi afterburner, you can have it display gpu temps, gpu usage, gpu memory usage and fan speed on screen while playing

 

[[TW]Fox]

Riva Tuner is being a nightmare, it wont run because of stupid Windows 7 Device Driver Signing. I've turned it off - heck Windows now even reports its in 'Test Mode' but Windows STILL wont let Riva Tuner install the driver it needs.

 

[Aedus]

EVGA precision?

 

[Michaeljcox24]

Msi afterburner, you can have it display gpu temps, gpu usage, gpu memory usage and fan speed on screen while playing

have mine running during games showing card temp and usage but was unaware it showed memory info. top tip

 

[elpedro]

Yeah at 1680x1050 crysis 4xAA goes over 1GB, you must really enjoy stuttering gameplay with large FPS drops.

Yeah, absolutely love all the above you mentioned...

 

[Flare]

I really can't decide between this and the 560! Ahhh driving me mental!

 

[mattjd]

snip

That's Warhead though, slightly less demanding than the original Crysis and you're not using Enthusiast settings

I would like a screenshot showing VRAM usage please

 

[[TW]Fox]

30 minutes of BC2 completed:

Maximum GPU load: 96%. Mostly sitting at 93%.
Maximum GPU temperature: 81c. Mostly sitting at 80c.
Maximum GPU fan speed: 57%.

10 minutes of Furmark completed.

Maximum GPU load: 97%. Mostly sitting at 96%.
Maximum GPU temperature: 89c. Mostly sitting at 85c.
Maximum GPU fan speed: 100%.

As we can see then Furmark results in noticeably MORE heat and noticeably MORE fan noise than your average high end game.

I do hope people are not making graphics card purchasing decisions based over a cards performance in an overly stressful synthetic benchmark instead of real world games?

 

[Vertigo]

Firstly, let me just say that I detest Furmark and I believe it has come in for a lot of criticism in that it's utterly unrepresentative of the load a real game would place on a card. It stresses them in ways no game ever would and I believe it has actually damaged some cards as a result.

That said, there's something that doesn't quite sit right with those results. The load figures suggest that BC2 is stressing the GPU to almost the same degree as Furmark, yet with BC2 the temp is kept at 81C with only 57% fan speed whilst Furmark needs 100% fan speed to maintain 89C? That just doesn't add up.

I fished out Anandtech's review of the 480 from when it was released and their figures suggest it's just as hot running Crysis as it is running Furmark, which is in itself surprising.

As I said earlier, everyone's perception of "loud" differs. When I got my GTX280 (standard non-overclocked reference model) I found the noise intolerable when playing demanding games, not just when using Furmark. It may be that you had a better (quieter) card, or were playing different games, or are just less sensitive to the noise than I.

FWIW: http://www.youtube.com/watch?v=xfQOU3T0O3Q

 

[DIABLO]

It's widely known that Furmark gets the GPU a lot hotter then playing a game. Thats why the 580 throttles itself when it detects furmark running.

 

[elpedro]

That's Warhead though, slightly less demanding than the original Crysis and you're not using Enthusiast settings

I would like a screenshot showing VRAM usage please

4xAA, enthusiast, 1080p on a £110~ graphics card with 768MB VRAM

No evidence of an erratic framerate, major dips or anything.

I know this chart is a bit old and the resolution isn't quite 1080p but notice there's 0xAA. See how my 28.92FPS compares.

Regardless how much memory is being used, it clearly isn't having an effect on my performance.

 

[bhavv]

Furmark is to graphics cards what Prime95 is to CPUs. No game will ever stress either your CPU or GPU as much as those programs do, however they are used to test for stability when overclocking.

If it doesnt crash in furmark, then its never going to crash in a game. That is what Furmark is meant to be for.

 

[[TW]Fox]

That said, there's something that doesn't quite sit right with those results. The load figures suggest that BC2 is stressing the GPU to almost the same degree as Furmark, yet with BC2 the temp is kept at 81C with only 57% fan speed whilst Furmark needs 100% fan speed to maintain 89C? That just doesn't add up.

What do you want, screenshots? They can be arranged?

 

[95thrifles]

No ones debating (I think) that what is loud is subjective, from one to the next, but equally no one should be debating the NV 5 series are quieter than the 4, as are AMD/ATi's equivalents, the 4 may well be quiet enough for you but the others are quieter, thats all

 

[MikeHiow]

I do hope people are not making graphics card purchasing decisions based over a cards performance in an overly stressful synthetic benchmark instead of real world games?

Ironic, seeing as earlier you said;

There is no other way [than benchmarks] - anything else is pure conjecture.

 

[bhavv]

Regardless how much memory is being used, it clearly isn't having an effect on my performance.

Nooooeeeesss! Your screenshot is photoshopped and your game is using up WAAAY more than 768 Mb of Vram, and you must be getting hitching and game crashes and playing at slideshow speeds cos you only have 768 mb Vram

Jokes.

Ironic, seeing as earlier you said;

Before he meant gaming benchmarks, now he means synthetic ones like furmark.