My benches so far. Adding GTX480 SLI benches.

[MarcVenice]

You should also run your tests in Far Cry 2 (I've found it to be quite good at finding stable oc's). Just concluding your oc is stable on a single game isn't very wise. You should also run Furmark to find a stable oc.

Also, GF100 oc's quite well, because Nvidia isn't clocking them as high as they could, because of the powerdraw. Thats why they got quite a lot of headroom. Just a bonus for those who OC I suppose

 

[RussianSensation]

Didn't know that Dirt 2 could be run in DX10 mode. . .

Ya, you are right I made a mistake. It's DX9 vs. DX11.

 

[ArchAngel777]

If you guys are worried about heat, I suggest that you run Vsync. It will limit the GPU & CPU from being pushed hard all the time. When I tested the power draw with both Vsync on and then off, it was amazing to see the difference it made. If you are someone who doesn't mind Vsync, you should look into turning it on, along with Tripple Buffering.

Always keep in mind that a lower power draw is less heat to dissipate... So this may, in fact, make the GTX 480 a candidate for me. All of these tests are with Vsync off, basically running the GTX 480 as fast as it can go (provided the CPU can push it fully) and thus, creating a huge power draw and with it the need for a faster spinning fan (loud!!!).

Keys or Apop, are you guys up for testing these things for heat/noise with vsync toggle on/off to see the difference? You can use a Kilowatt device, or an APC unit with reporting. Or you could even do a subjective test with leaving the fan to 'auto' and running a certain game for 20-30 mins each to see how loud the fan is.

 

[apoppin]

If you guys are worried about heat, I suggest that you run Vsync. It will limit the GPU & CPU from being pushed hard all the time. When I tested the power draw with both Vsync on and then off, it was amazing to see the difference it made. If you are someone who doesn't mind Vsync, you should look into turning it on, along with Tripple Buffering.

Always keep in mind that a lower power draw is less heat to dissipate... So this may, in fact, make the GTX 480 a candidate for me. All of these tests are with Vsync off, basically running the GTX 480 as fast as it can go (provided the CPU can push it fully) and thus, creating a huge power draw and with it the need for a faster spinning fan (loud!!!).

Keys or Apop, are you guys up for testing these things for heat/noise with vsync toggle on/off to see the difference? You can use a Kilowatt device, or an APC unit with reporting. Or you could even do a subjective test with leaving the fan to 'auto' and running a certain game for 20-30 mins each to see how loud the fan is.

Personally i never use Vsynch, but i could test it for you as i have a Kill-A-Watt meter.

My GTX 480 is about as loud as my HD 4870-X2 and leaving it in my Gladator 600 (closed case w/5 fans) - for HOURS at maximum load (looping Crysis at 2560x1600 w/max details) - raised the inside case temps to an absolute max of 96.4 F while the ambients also peaked near 80F. The GTX 480 will also peak near 212 F (nearly 100C) as its hottest temp that i can find.

 

[ArchAngel777]

Personally i never use Vsynch, but i could test it for you as i have a Kill-A-Watt meter.

My GTX 480 is about as loud as my HD 4870-X2 and leaving it in my Gladator 600 (closed case w/5 fans) - for HOURS at maximum load (looping Crysis at 2560x1600 w/max details) - raised the inside case temps to an absolute max of 96.4 F while the ambients also peaked near 80F. The GTX 480 will also peak near 212 F (nearly 100C) as its hottest temp that i can find.

Any particular reason why you don't use it? I can understand why some might not if they are into competitive FPS or extremely sensative to input lag (which is hit or miss with Vsync + TB) but then you have to deal with TEARING and oh my... I cannot stand tearing at all. That ruins my entire gaming experience.

 

[apoppin]

Any particular reason why you don't use it? I can understand why some might not if they are into competitive FPS or extremely sensative to input lag (which is hit or miss with Vsync + TB) but then you have to deal with TEARING and oh my... I cannot stand tearing at all. That ruins my entire gaming experience.

It's a personal preference for me in gaming. Tearing is not so severe and it does not bother me when it happens occasionally.

 

[HurleyBird]

Well, most big sites are guily of this as well, but I think posting min and max FPS values by themselves are not very informative. There's not much difference between 15 and 9 minimum FPS when those values are reached for only a couple frames at the start of a level. There's a HUGE difference between 15 and 9 minimum FPS when that level is reached for an extended period of time during heavy action where you need good frame rates the most. Contrary to popular beleif, high maximum frame rates are a bad thing, assuming the average is equal, stable frame rate values are better than flucuating ones. The more time a game spends near the average FPS the better. The best thing to do, if you want to report more than just the average, is to provide a graph of frame rate over time, and then describe what is happening over the course of the run. This method is a lot more informative and I wish more sites used it.

 

[3DVagabond]

The best thing to do, if you want to report more than just the average, is to provide a graph of frame rate over time, and then describe what is happening over the course of the run. This method is a lot more informative and I wish more sites used it.

Yes, graphs give you the whole picture. I agree that it's important to know what happened to drop frame rates as well. No two runs of a game are going to put the same demands on the card. Also, how often the minimums (or maximums) were approached matters. This will be reflected in the avg. frame rates, but still doesn't tell the whole story. Maximum and minimum frame rates are nearly worthless without qualifying them somehow.

 

[Rubycon]

Turning off vsync may look better (no tearing) but game play suffers from pumping and jerkiness too much.

 

[Keysplayr]

Yes, graphs give you the whole picture. I agree that it's important to know what happened to drop frame rates as well. No two runs of a game are going to put the same demands on the card. Also, how often the minimums (or maximums) were approached matters. This will be reflected in the avg. frame rates, but still doesn't tell the whole story. Maximum and minimum frame rates are nearly worthless without qualifying them somehow.

So, any review out there that does not use line graphs is no good then. That's what you're telling me. All of them useless. When I first announced via thread that I was going to run benches, why didn't you suggest it? You posted 4 times in that thread, but no suggestion to do so.
http://forums.anandtech.com/showthread.php?t=2055912
Posts: 37, 39, 51, 65.
I really wish you would have mentioned this earlier, but we have what we have.

 

[ArchAngel777]

Turning off vsync may look better (no tearing) but game play suffers from pumping and jerkiness too much.

You must have meant to say on, and not off? Or am I missing something?

Going to presume that you meant 'on' and that being the case, what is this jerkiness you are talking about? I would imagine you would be referring to someone not turning on TB? I wouldn't dream of turning on Vsync without TB, if that is where you are going with it. I have never experienced any pumping or jerkiness with Vsync + TB. So I am gonna have to ask you to clarify your post, because at face value here, your post seems a bit off.

 

[Rubycon]

You must have meant to say on, and not off? Or am I missing something?

Going to presume that you meant 'on' and that being the case, what is this jerkiness you are talking about? I would imagine you would be referring to someone not turning on TB? I wouldn't dream of turning on Vsync without TB, if that is where you are going with it. I have never experienced any pumping or jerkiness with Vsync + TB. So I am gonna have to ask you to clarify your post, because at face value here, your post seems a bit off.

Yes turning it ON makes it jerky. In games where render ahead is uncapped with no vertical sync and frame rates can soar to 10+ sometimes as much as 30X refresh frequency it will get all screwy as if the system appears struggling to play it! For this reason I force vsync OFF always.

Yes my original post was meant to say OFF is better. (41 hours no sleep / no coffee does that. )

 

[ArchAngel777]

Yes turning it ON makes it jerky. In games where render ahead is uncapped with no vertical sync and frame rates can soar to 10+ sometimes as much as 30X refresh frequency it will get all screwy as if the system appears struggling to play it! For this reason I force vsync OFF always.

Yes my original post was meant to say OFF is better. (41 hours no sleep / no coffee does that. )

Can you give me some games where you have experienced this? Every game I have ever tested to date has never given me a problem with Vsync + TB. Never experienced what you are describing and I always force Vsync+TB and generally run a fast video card.

Why are you at 41 hours of being awake with no sleep?

 

[MagickMan]

Can you give me some games where you have experienced this? Every game I have ever tested to date has never given me a problem with Vsync + TB. Never experienced what you are describing and I always force Vsync+TB and generally run a fast video card.

Why are you at 41 hours of being awake with no sleep?

WoW gets jerky with vsync on, not run into it with other games.

 

[evolucion8]

Triple Buffering fixes the stuttering of V-Sync, is a pity that it can't be forced in 64-Bit games (At least didn't work with Crysis Warhead 64-Bit)

 

[ArchAngel777]

WoW gets jerky with vsync on, not run into it with other games.

I played WoW for over 3 years (not proud of it BTW... lol...) and never had any stutter. And lets define stutter right now, because it isn't a true 'stutter' in the sense that it is used in games where HDD is accessing the textures from the HDD. The sutter that people are describing is merely the act of going from a steady frame rate to a fraction of that frame rate. To give an example: 60fps to 30fps back to 60, etc... when performance was not able to keep the frame-rate pegged at 60. This is in fact very noticable, but that is why we have TB, to completely eliminate it. I have never ran accross a game that didn't have TB, or didn't let the TB work via nVidia CP. However, I am sure there are games out there, which is why I'd like to know what those games are. I think it is fair to say that I have played all the top AA or AAA titles, or at least their respective demos and have not ran into this issue.

Also, from what recall, SLI had an issue with Vsync. I am wondering if some of these people think it is worse than it is because they are using multi-gpu setups. I can't comment on those, because I don't run them.

 

[MagickMan]

What I see in WoW is a very slight, yet noticeable and annoying, pause while moving. It's similar to the micro-stutter I've seen in SLI and some xfire setups in certain games. It happens to me when I have vsync On, even with TB enabled. Not a big deal to me, really, I just turn it off and run at 200+ FPS. There's some screen tearing but it isn't that bad, it's better than the stutter.

 

[evolucion8]

A nice example of stuttering I experimented recently was in Vantage, the reported FPS in the first level reported 30 to 60fps, but never looked like that, at 40fps looked like 20, and 30fps looked like 10fps, is all jerky and horrible.

 

[Magusigne]

Ah,

nKey's..my friend...

I actually believe your post, mainly because Apoppin backed it up. That's the price you pay I suppose...

Anyways, It does really indeed like the GTX 480 is the better performer out of the GTX 480 vs 5870, and while I believe there are people (A great many) that only care about performance obviously that doesn't effect 100% of a persons FX card decision...you make the power/heat issue sound rather minor..but when comparied to the red team..you can see how inefficient it really is.
See Infared pics--->
http://www.geeks3d.com/20100405/infrared-pictures-of-the-geforce-gtx-480/

I think before I consider the GTX 480 they are going to need to figure out there heat/power issue.. maybe the next revision or two..

 

[Madcatatlas]

Those pictures reminded me of Predator..

 

[apoppin]

Those pictures reminded me of Predator..

i turned off the gas in my house so that the hissing from the pilot could not be heard, so my house is really cold without the gas heater.
- So, i have turned my case around on the table so the rear faces me and the warm GTX 480 exhaust air is blowing on me right now . . .it's kinda nice now for Winter/early Spring; but maybe not so cool for Summer


But i am looking at the temperature difference in the ambient temps and inside my case (near the HD) and it is only ten degrees (F) difference.

i can conclude from this that my Gladiator 600 case with five 120-140 mm fans has an *Excellent airflow*. Perhaps now you can understand why i am not so bothered and why my PoV is perhaps different; especially coming from another noisy and hot video card - the HD 4870-X2 and living in a relatively cold place.

 

[MrK6]

It's a personal preference for me in gaming. Tearing is not so severe and it does not bother me when it happens occasionally.

Turning off vsync may look better (no tearing) but game play suffers from pumping and jerkiness too much.

I think you meant turning on vsync, but with that addendum I agree with both of you. I can't stand the input lag and jerkiness of vsync either, and am glad to see apoppin and Keys benchmarking without it. However, if there's time, a few runs with it on for comparison's sake would be an appreciated addition.

 

[ArchAngel777]

I think you meant turning on vsync, but with that addendum I agree with both of you. I can't stand the input lag and jerkiness of vsync either, and am glad to see apoppin and Keys benchmarking without it. However, if there's time, a few runs with it on for comparison's sake would be an appreciated addition.

You would have to benchmark without it, or else you would rarely see any differences between any of the cards. Benchmarking with it off is totally acceptable and supported by me. I don't think it should be done any other way. Just wanted to clarify... But for actual game-play, I won't ever run without it.

 

[Keysplayr]

I never run with V-Sync in any gaming I do. If there is any tearing, I don't bother to notice, or to be more precise, it doesn't diminish my gameplay experiences.

 

[Rubycon]

I never run with V-Sync in any gaming I do. If there is any tearing, I don't bother to notice, or to be more precise, it doesn't diminish my gameplay experiences.

Yes with it on how would you tear through a room? You would get shot!