B3D GTX-480 Thermal Study

[HYBRID]

Pretty good read.


http://benchmarkreviews.com/index.ph...0628&Itemid=21

[LurkerLito]

A really nice read. And it looks approximately like what I am getting when I check my temp readings after I game. Some games run hotter than others but the max I have hit is about 89.

[HYBRID]

I hit 85c playing BFBC2 after a good 2 hour mp session.

[Bah!]

If only they would have done this two months ago, it could have gone a long way towards showing what many of us have been trying to tell people since these cards were released.

The funny thing is, even as low as the temps they are reporting are compared to the pre-release reviews, they are still quite a bit higher than what I see daily just by making a custom fan profile. The highest temp I've seen in anything (including benchmarks) is 85c. Most of my games run in the low 70's.

[Madpistol]

Ugh... guess I get to play devil's advocate on this one.

1. Gigantic and amazing cooling solution to cool the sucker down.

2. Extremely high power usage.

3. High heat output = higher cooling costs in the room.

4. Exposed HSF = heat back in the case = hotter components.



As well done as the article is, I'm getting sick of reading articles trying to justify the amount of heat that this thing produces. At one time not too many years ago, this was considered an absolutely ludicrous cooling solution for a video card. Now we have THIS!!!

I'm all in the name of advancing technology, but this card is just pushing the limit too far. Exposed heat pipes that get so hot that they can burn you??? More power usage than a HD 5970??? The list goes on...

I sound like I'm trying to justify my HD 5870. I know. But why does Nvidia believe that this sort of extreme heat and power usage is acceptable for retail usage? And why do so many people continually stand behind it saying it's an amazing product? Don't get me wrong, the performance is quite good from a single-chip solution, but at what cost? The 8800 GTX wasn't that bad, and lets face it, that thing was a pretty hot (temp) card. The 6800 series was quite awesome too, but it required a smaller 2 slot cooling solution in its most powerful form, and it got up into the 70c range temperature wise. It didn't get into the 90s.

These articles are hilarious. How can you justify those temperatures? According to them, the 5870 gets up to within 5C of the GTX 480... Look at the difference in cooling solutions! The GTX 480 has 4 exposed heatpipes. The HD 5870 has 1/2 the upper slot for exhaust. Why are the temperatures of each of these cards being compared? If you put the GTX 480's cooling solution on the HD 5870, the 5870 would probably hit full load in the 60's or low 70's max. Likewise, if you put the HD 5870's cooling solution on the GTX 480, you'd have a card with a fan that never shuts up because it's constantly spinning up to cool the card down.

Once again, I fail to see reasoning behind this article. Heat is still heat, and the GTX 480 creates a lot of it. No amount of testing or writing is going to change that. Only a revision of the architecture will.

[WeReWoLf]

Yes, we know it runs hot.This article proves that. This article also proves that the radeons only run slightly cooler however, so from what you're saying both sets of cards are unacceptable.

[Bah!]

Originally Posted by Madpistol

As well done as the article is, I'm getting sick of reading articles trying to justify the amount of heat that this thing produces.

But why does Nvidia believe that this sort of extreme heat and power usage is acceptable for retail usage? And why do so many people continually stand behind it saying it's an amazing product?

How can you get sick of reading these articles when there aren't any? This is literally one of the very few out there.

As far as why Nvidia believes this is acceptable, it's because it is acceptable, and the product is amazing. Most people who spend $500-2000 on video cards don't care about the stuff you are complaining about.

My year to year my power bill went up a whopping $3 in May. That's insignificant to me. I've seen reports saying it would be $12...also insignificant. As far as heat, 80% of the time they produce no more heat than a 5870. The other 20% of the time my gaming is improved and I couldn't care less.


I used to own this bad ass Firebird. Thing was loud as hell and sucked gas like it was going out of style. It was also fast as lightning and is to this day my favorite of all the cars I've ever owned. The 480 is my Firebird.

I'll gladly sacrifice a little power usage and heat for more power. And don't kid yourself, the performance isn't "quite good", it's ****ing excellent, and the gap between it and the 5870 is even larger than the benchmarks would lead you to believe.

[Viral]

I don't exactly see the point in discrediting furmark results. Yes, the temps in actual games are less, but so are the temps of the competiting GPU's. It's a worst case scenario and any scaling down of temps in actual games should apply in a relative manner to competing solutions.

[Madpistol]

Originally Posted by Bah!

And don't kid yourself, the performance isn't "quite good", it's ****ing excellent, and the gap between it and the 5870 is even larger than the benchmarks would lead you to believe.

No. It's quite good. People, especially fanboys of all brands, tend to over inflate numbers. That's the sad fact of this world. People trying to justify a product that they own by creating numbers that are artificially high. There are people that claim that Crysis runs fantastically, while others claim it only runs as good as a 5870. There are also people that claim to get 170 FPS in BC2, while others complain about load times and 70 FPS max in BC2. I will admit that I'm glad the FPS minimum is higher. That's one area where the GTX 480 got it right.

The performance is "quite good" If you want "****ing excellent" performance, free free to pic one of these insanely expensive behemoths up. Just make sure you've got a small nuclear reactor to power it if you plan to add in 2 of them. Make sure you've got a swiss bank account as well.

[Madpistol]

Originally Posted by Viral

I don't exactly see the point in discrediting furmark results. Yes, the temps in actual games are less, but so are the temps of the competiting GPU's. It's a worst case scenario and any scaling down of temps in actual games should apply in a relative manner to competing solutions.

It's funny that you mention that. I've actually found that OCCT's GPU stress test creates more heat than Furmark. I wonder how many people actually know of this...

Furmark is able to push my video card up to 86C, but at that point the fan spins up and cools it back down. OCCT's GPU test is able to push it into the low 90's with fan speeds reaching 50%. At that level, the fan is pretty audible over the rest of the computer.

So if you want to make your GTX 480 scream, OCCT's GPU test is the way to go.

[Rollo]

Madpistol:
Can you tell me why you think the heat that video cards generate running synthetic benchmarks matters if it is not comparable to the heat generated while running games?

What does it matter if the exposed heat pipes can burn you? Usually I'm not gaming with my hands in my case- I have one on the keyboard and one on the mouse.

[Xion X2]

The logic of this entire thing is pretty simple.

Can the GTX480 be cooled to the temps of the 5870 and under? Sure. Slap a massive heatsink or a waterblock on it and watch the temps nose dive.

So what? Trying to accurately gauge temps of a GPU based on whatever heatsink it has is useless. The heatsink, case, fan speed and ambient temps will differ a hundred times over between users. This is why you should base how warm a GPU runs on something concrete like its power usage because that is an independent variable that is constant.

These last cards from Nvidia are good performers, but I, like Madpistol, am sick of seeing people pull any argument that they can out of their &#^ trying to justify the temps. It is the highest power-drawing single-GPU card ever; therefore it is the hottest running single GPU ever. It's as simple as that. Any arguments against this are simply rhetoric and aren't based in fact.

In addition to that, take a look at Nvidia's stock. Their stock has taken a nose dive since Fermi released because investors know that AMD's latest line of GPUs are more power efficient and priced better and have sold better. Are they better? No, not for everyone (and not for me, because I think their drivers suck) but for the mainstream, yes, because of the above reasons (power usage and price.)

Nvidia's stock has lost 6$ a share (~33%) since April, Fermi's release:



http://finance.yahoo.com/q/ta?s=NVDA...+Analysis&t=6m

Anyone who cares to look at 5870 vs 480 logically (I'm not talking about you Nvidia fanboys who fail to ever see anything that Nvidia does wrong) just needs to accept that AMD offered the better product this time around. This is how it's viewed on Wall Street, and this is how it is in reality. Nvidia's saving grace at this point are their drivers and their feature set, but unfortunately for them, only the true enthusiast crowd seems to care about these. Not the mainstream.