GF100 evaluation thread

[rpg.314]

I think that the perf/$ wise, from AMD's pov, gf100 is god sent. When Cypress launched, it had rather high prices. In absence of immediate competition, that was to be expected. Overall, AMD was enjoying pretty good margins then. The supply shortfall served to increase prices and AMD's margins only improved. Today, after the arrival of competition, they are still quite competitive in perf/$ wrt gf100. All in all, AMD's accountants will be very happy at the perf/$ delivered by gf100.

In that respect, it would have been quite ironic if fermi launched in Nov and AMD increased prices then. Short of a B spin I don't expect the situation to change substantially in time for Hecaton/NI.

GF104, however, could lead to some price cuts in that segment of the market.

Here's hoping that extra cash at AMD's doorstep leads to better drivers for Linux.

 

[XMAN26]

On price:
470 WTF!, 480 WTF! (both are over priced)

On specs:
470 ok. 480 WTF!

On performance:
470 pretty good. 480 Pretty good

overall

Meh to both

 

[XMAN26]

For me its fail - I wouldnt pay for either more than for 5850 or 5870 because 470 and 480 are louder, more power hungry and hotter.

Fermi = R600 (neither of them is/was as bad as NV30 which was worst GPU ever IMO).

I disagree. R600, even in its fastest clothing could barly compare to the 8800GTS where as here, both perform better than ATIs single GPU counter parts, their biggest draw back is that they are power hungry, louder and more expensive. NVidia should have priced them the same as ATIs offerings. Having not done so will cause a reduction in sales(availability not with standing) and a boost to ATIs.

 

[Tchock]

Err in retrospect ATI could just as well done a 2900XTX with the zomghuge cooler and zomghuge power draw, and the 2900XT is basically the GTX 470 here.

What ATI lacked was the devout.

 

[XMAN26]

This is what I wrote in the other thread:

I just can't see any reason to buy either of these cards over their AMD counterparts. Performance is very similar, and when you take into account the big increase in noise, power, cost and heat of the Nvdia cards, the AMD cards are more attractive hands down.

The 470/480 are just not competitive cards unless you are a corner case that must have niche capabilities (ie Cuda or Physx). When you look at them as a whole, including performance, noise, heat, cost, power, the Nvidia cards are just not good when compared to the AMD products. I wouldn't have a 470/480 if you gave me one because of the noise alone - I couldn't live with it.

15-20% more performance is on par?! And while the videos are revealing I take issue with 2 things about them. In all but the 480 video, the mic is atleast 6" away, but in the 480 video, it is amost on top of the fan and if you dont think mic placement can alter a sound reading, you are sadly mistakin. Secondly, doing that sound test would have been better served with the cards in a case, side panels on as that is how most people use their computer.

 

[GZ007]

90 Celsius degree idle with two lcd-s . http://www.legitreviews.com/article/1258/15/
And the nvidia response:

"We are currently keeping memory clock high to avoid some screen flicker when changing power states, so for now we are running higher idle power in dual-screen setups. Not sure when/if this will be changed. Also note we're trading off temps for acoustic quality at idle. We could ratchet down the temp, but need to turn up the fan to do so. Our fan control is set to not start increasing fan until we're up near the 80's, so the higher temp is actually by design to keep the acoustics lower." - NVIDIA PR

So basicly u can choose betwen loud or hot .
GF100 architecture is actualy great but the chip is broken. Heat and power is catastrophic compared to 5k radeons. And thats a castrated chip clocked lower than they wanted. (and it seems the gddr5 is making them problems too)

 

[XMAN26]

Err in retrospect ATI could just as well done a 2900XTX with the zomghuge cooler and zomghuge power draw, and the 2900XT is basically the GTX 470 here.

What ATI lacked was the devout.

If memory serves, they did an XTX, then pulled it as it too sucked balls.

 

[thatdude90210]

I wonder how a GTX 480 would perform if you underclock/undervolt it to draw 188W, same as a 5870. I'm guessing it would be a total dog. Basically, if Fermi had been released by a non-"win at all cost, even if we have to feed it beyond sanity" company, it could have been a lot closer to nv30/r600 standard.

 

[Bouncing Zabaglione Bros.]

15-20% more performance is on par?! And while the videos are revealing I take issue with 2 things about them. In all but the 480 video, the mic is atleast 6" away, but in the 480 video, it is amost on top of the fan and if you dont think mic placement can alter a sound reading, you are sadly mistakin. Secondly, doing that sound test would have been better served with the cards in a case, side panels on as that is how most people use their computer.

15-20 percent is pretty much best case. That's not worth the extra cost, heat, and noise even in the few games you'll get that. Heck, I'd have to get aircon installed and pay to run that in my study if I was running a 480! 15 percent best case in some games isn't enough to balance out all the negatives for me.

I mean you could offer a card that was a thousand percent better than competitor products, but if you need a nuclear power source in your office, it's not much use to me either.

 

[neliz]

15-20% more performance is on par?! And while the videos are revealing I take issue with 2 things about them. In all but the 480 video, the mic is atleast 6" away, but in the 480 video, it is amost on top of the fan and if you dont think mic placement can alter a sound reading, you are sadly mistakin.

You can see the microphone?
Kyle's set-up being fixed on a tripod makes me believe the distance is equal in all cases.

Secondly, doing that sound test would have been better served with the cards in a case, side panels on as that is how most people use their computer.

How do you measure the cards noise when you'd have other components interfering? you'll need some hefty fans to keep the airflow going in a case with a 480.

 

[Bouncing Zabaglione Bros.]

How do you measure the cards noise when you'd have other components interfering? you'll need some hefty fans to keep the airflow going in a case with a 480.

I think that a large number of cases that won't supply the 470/480 with enough air, and cards will get even louder and noiser. Even good cases will have problems, especially if there are other components in there that generate heat or block off airflow. I think at those temps and fanspeeds, 480 is at the edge of it's thermal capabilities, and those of nearly all cases to cool.

I think it was Guru3d that said you can't touch the top of the card, it's so hot. That heat is going to go into your case too.

 

[flippin_waffles]

Stick a fork in it, it's done. Option 5.

 

[XMAN26]

You can see the microphone?
Kyle's set-up being fixed on a tripod makes me believe the distance is equal in all cases.



How do you measure the cards noise when you'd have other components interfering? you'll need some hefty fans to keep the airflow going in a case with a 480.

If that little red thing in the 470/480 video isn't a mic, what the hell is it then because it looks like a mic to me.

You don't need to use load case fans, they do make silent ones. not to mention I believe he runs water for his CPU.

 

[XMAN26]

15-20 percent is pretty much best case. That's not worth the extra cost, heat, and noise even in the few games you'll get that. Heck, I'd have to get aircon installed and pay to run that in my study if I was running a 480! 15 percent best case in some games isn't enough to balance out all the negatives for me.

I mean you could offer a card that was a thousand percent better than competitor products, but if you need a nuclear power source in your office, it's not much use to me either.

15-20% is the average range I saw in the reviews I looked at with a few exceptions here and there. I'll give you the cost, power and heat side as I agree, but I wouldn't call an average advantage of 15-20% with launch drivers against mature drivers being on PAR.

As to cases, I have a Cosmos 1000 case with a pair of GTX260s with fan set to 100% and the sound range isn't loud at all in any way shape or form. Now if I take the cover off, it gets damn loud.

 

[neliz]

If that little red thing in the 470/480 video isn't a mic, what the hell is it then because it looks like a mic to me.

That's probably something on his desk, because the mic. used actually is black. http://ep.yimg.com/ca/I/yhst-16162446720287_2099_2234201631

 

[swaaye]

What has me worried is the whopping 95c at load.

It's not that uncommon in notebooks. Also, a friend of mine has a 4850 that cooks up around there.

 

[swaaye]

All that's wrong with it is that it runs too hot, is too loud, pulls in too much power and costs more. Trivial stuff. Everything else is great. Maybe if Intel would build it on 32nm for them. lol.

I just wish we'd have a price war but that's not gonna happen with that gigantron megachip.

 

[BRiT]

All that's wrong with it is that it runs too hot, is too loud, pulls in too much power and costs more. Trivial stuff.

Completely trivial. They'll fix that with newer drivers.

 

[mboeller]

15-20% more performance is on par?! And while the videos are revealing I take issue with 2 things about them. In all but the 480 video, the mic is atleast 6" away, but in the 480 video, it is amost on top of the fan and if you dont think mic placement can alter a sound reading, you are sadly mistakin. Secondly, doing that sound test would have been better served with the cards in a case, side panels on as that is how most people use their computer.

Kyle was not the only one reporting that the GTX480 is really really loud:

Even in Grid the GTX 480 exceeds its TDP by 16 watt and when running Furmark it even requires more than 300 watt. The loudness in 3D mode is, with 7 up to almost 12 Sone, extremely high.

source: http://www.pcgameshardware.de/aid,7...ias-GF100-Generation/Grafikkarte/Test/?page=2

you can see the test setup in the link. They have a video too on youtube showing the noise at ~2.50min in the video:
http://www.pcgameshardware.de/aid,7...hnell-ist-die-Geforce-GTX-480/Multimeda/News/

edit:
PS.: I somehow mixed up the german review (link) with the english review (text)

 

[no-X]

If memory serves, they did an XTX, then pulled it as it too sucked balls.

They have never released XTX because of its power draw. I think they sold the XTX parts underclocked to XT level as HD2900XT 1GB GDDR4 (I remeber some of the boards had "850MHz" sticker or something like that)

The key difference is that R600 and G80 were developed as direct competitors. R600 failed (too high power draw + broken MSAA resolve), so ATi decided to compete only with nVidia's top-1 model.

RV870 wasn't designed as direct competitor of Fermi. Dual-RV870 was meant to compete to Fermi. Despite that, Fermi is only able to outperform single-RV870 by 15% at the cost of 30%+ higher street-price, 50% larger die, 50%+ higher power draw, extreme noise, 6 months+ delay etc.

GTX400 simply missed its target significantly more than HD2900... But I agree, that both of them had quite similar story. Delay, high power consumption, noisy cooling, overkill performance at wrong place (FP16 tex + BW / geometry performance + tessellation)