GT300 "Fermi" Game Performance Figures Leaked?

[josiahsuarez]

http://www.xtremesystems.org/forums/showthread.php?t=236661

unfortunately I can't view this thread. hopefully one of you can, is there anything juicy there?

[josiahsuarez]

woohoo, google cache to the rescue :)

http://74.125.47.132/search?q=cache:http://www.xtremesystems.org/forums/showthread.php?t=236661

basically it leaks to this supposedly leaked youtube video

http://www.youtube.com/watch?v=10xKBSEEvwQ

Crysis 1680x1050, very high DX10
GT300: 75
HD5870: 43

Far Cry 2 1680x1050, very high DX10
GT300: 177
HD5870: 106

Vantage performance score
GT300: 23098
HD5870: 16098

[TheANIMAL]

10 - 1 odds on, that is fake.

[v3rninater]

Yea even the guys on there seem to think that's fake, would be awesome if that were true!

[Sazar]

woohoo, google cache to the rescue :)

http://74.125.47.132/search?q=cache:http://www.xtremesystems.org/forums/showthread.php?t=236661

basically it leaks to this supposedly leaked youtube video

http://www.youtube.com/watch?v=10xKBSEEvwQ

Crysis 1680x1050, very high DX10
GT300: 75
HD5870: 43

Far Cry 2 1680x1050, very high DX10
GT300: 177
HD5870: 106

Vantage performance score
GT300: 23098
HD5870: 16098

Here's to hoping that we get this card at a reasonable price and with performance on the final version as good as the apparently fake numbers posted online :D

Would be nice.

-edit-

the response video had me cracking up :D

g8EKtCK8zbY

[Ninja Prime]

If that is true, not looking too good there on the NV front. 4+ months later and only 40% faster? In all likelyhood the 5870x2 will launch before NV does and it will be faster than that by far, leaving NV to launch with a slower than top end card that is supposed to be the top end card... Unless they launch with a x2 type part too, but I doubt that as well, they had problems with the GTX295 and the GT300 is supposedly larger and certainly more power hungry.

[mojoman0]

If that is true, not looking too good there on the NV front. 4+ months later and only 40% faster? In all likelyhood the 5870x2 will launch before NV does and it will be faster than that by far, leaving NV to launch with a slower than top end card that is supposed to be the top end card... Unless they launch with a x2 type part too, but I doubt that as well, they had problems with the GTX295 and the GT300 is supposedly larger and certainly more power hungry.

LOL only 40? for the SAME GENERATION? I don't think anybody is expecting that big of a performance difference hence the hilarious youtube reply. ATI would also have no way in hell of getting 40% more out of their cards in 4 months. :headexplode:

[Johnny C]

LOL only 40? for the SAME GENERATION? I don't think anybody is expecting that big of a performance difference hence the hilarious youtube reply. ATI would also have no way in hell of getting 40% more out of their cards in 4 months. :headexplode:

Where ohhh Where could AMD find 40% more outta their card?

Oh wait here
http://hothardware.com/News/Asus-Voltage-Tweaks-Radeon-HD-5870-To-A-38-Performance-Boost/

and we don't have to wait 4 months.

In fact I'd only have to wait 2-3 business days....as it's IN STOCK right now.

http://www.ncix.com/products/index.php?sku=44304&vpn=EAH5870%2FG%2F2DIS%2F1GD5&manufacture=ASUS

[CaptNKILL]

LOL only 40? for the SAME GENERATION? I don't think anybody is expecting that big of a performance difference hence the hilarious youtube reply. ATI would also have no way in hell of getting 40% more out of their cards in 4 months. :headexplode:

Agreed. :lol:

40% over the 5870 would be crazy. I'd spend $600 on a graphics card again if this was the case. That said, I don't know where 40% came from. 43 to 75 is over 70% increase.

Just in case anyone is forgetting, the 5870 is already 30-40% faster than a GTX 285 in most cases. Anandtech's results in Crysis Warhead at 1680 with 4xAA show the 5870 pulling 45fps and the 285 getting 34. If a GT300 could manage 75 in a similar situation that'd be more than 100% increase from one generation to the next.

Amazing performance, but I really doubt this is real. Actually, I'm almost positive that the video isn't real (why would nvidia have a video slide show with a water mark for internal use? LOL) which means the performance numbers are most likely BS too.

Besides, why are they calling it the Tesla? Tesla isn't even a graphics card. It has no video outputs.

[mojoman0]

Where ohhh Where could AMD find 40% more outta their card?

Oh wait here
http://hothardware.com/News/Asus-Voltage-Tweaks-Radeon-HD-5870-To-A-38-Performance-Boost/

and we don't have to wait 4 months.

In fact I'd only have to wait 2-3 business days....as it's IN STOCK right now.

http://www.ncix.com/products/index.php?sku=44304&vpn=EAH5870%2FG%2F2DIS%2F1GD5&manufacture=ASUS

if that card were made to last even a year ATI would have made it standard out of the box. Why in the world would they nerf their card 38% purely from a voltage limit? Reliable ? :thumbdwn:

[Ninja Prime]

Agreed. :lol:

40% over the 5870 would be crazy. I'd spend $600 on a graphics card again if this was the case. That said, I don't know where 40% came from. 43 to 75 is over 70% increase.

Just in case anyone is forgetting, the 5870 is already 30-40% faster than a GTX 285 in most cases. Anandtech's results in Crysis Warhead at 1680 with 4xAA show the 5870 pulling 45fps and the 285 getting 34. If a GT300 could manage 75 in a similar situation that'd be more than 100% increase from one generation to the next.

Amazing performance, but I really doubt this is real. Actually, I'm almost positive that the video isn't real (why would nvidia have a video slide show with a water mark for internal use? LOL) which means the performance numbers are most likely BS too.

Besides, why are they calling it the Tesla? Tesla isn't even a graphics card. It has no video outputs.

Need I remind you that the GTX280 was only about 40% faster than the 4870 at launch, and the 4870 won last round, and it launched only a few days before. Here we have a launch months before. The 4870x2 came along a month or so later and destroyed the GTX280, while NV promised for months they would have an x2 card out "soon" but it took forever and when it came out it was pretty much too little too late. I see some parallels, don't you?

[mojoman0]

Need I remind you that the GTX280 was only about 40% faster than the 4870 at launch, and the 4870 won last round, and it launched only a few days before. Here we have a launch months before. The 4870x2 came along a month or so later and destroyed the GTX280, while NV promised for months they would have an x2 card out "soon" but it took forever and when it came out it was pretty much too little too late. I see some parallels, don't you?

this isn't about pricing, it's about performance. ATI pushed their cards out the door and will get sales for being first, and for being cheaper after their first refresh. You get what you pay for tho.

ATI = Cheaper, slower, little more bang for your buck
Nvidia = Pricier, faster, feature heavy

This is how AMD has been staying in the PC market for the last 3 years

[JH24]

It would be cool if this was true, but it must be fake. Why hasn't Nvidia released any game benchmarks themselves? Although I wouldn't fully trust either company's (Nvidia/AMD) own tests, it would show Nvidia's confidence in their product.

[CaptNKILL]

Need I remind you that the GTX280 was only about 40% faster than the 4870 at launch, and the 4870 won last round, and it launched only a few days before. Here we have a launch months before. The 4870x2 came along a month or so later and destroyed the GTX280, while NV promised for months they would have an x2 card out "soon" but it took forever and when it came out it was pretty much too little too late. I see some parallels, don't you?

Ok I'm confused. :p

The GTX 280 was released first and was at best 20-30% faster than the 4870 and was normally only 15-20% faster. It also required an 8pin PCI-E power connector which hardly anyone had, and it was $200 more expensive.

How are there parallels between that and nvidia releasing a card late which is (possibly) 70% faster than the 5870?

If ATI manages to put out a 5870x2 around the same time nvidia releases the GT300 its going to be a major power hog and its going to be really expensive (remember, the 5870 is $80 more than the 4870 was at launch). Not to mention it will use crossfire which many people aren't interested in.

If nvidia releases the GT300 with a significant performance lead over the 5870 and a competitive price they will have no problems this time around, even with a 5870x2. Especially with Physx, CUDA and 3D Vision support.

[slaWter]

I'd also say this is fake but imagine 2-3 of those beasts in SLI :drooling:

[CaptNKILL]

I'd also say this is fake but imagine 2-3 of those beasts in SLI :drooling:

My god. :drooling:

That could probably achieve the impossible... Crysis at 1920x1200 with 8xAA transparency AA (multisampling ;)) at 60fps constant in DX10 mode, maxed.

[FlakMagnet]

Would that actually be possible? Or would the CPU be a bottleneck before that could be achieved? I guess I could try it at 640x480 on my i7 and see if 60FPS is possible.

Just imagine it though... 60FPS at max details.... :drooling:

[CaptNKILL]

Would that actually be possible? Or would the CPU be a bottleneck before that could be achieved? I guess I could try it at 640x480 on my i7 and see if 60FPS is possible.

Just imagine it though... 60FPS at max details.... :drooling:

Crysis isn't that CPU heavy unless you're doing some crazy physics experimentation. It ran extremely smooth when I dropped the resolution way down on my 8800GTX and Core 2 Duo.

[Heinz68]

Fake it's. Anybody can make such video. The problem is the video wont play Crysis.
XtremeSystem deleted the thread, that's why the link doesn't work.

[MikeC]

If that is true, not looking too good there on the NV front. 4+ months later and only 40% faster?

http://www.percent-change.com/index.php

[FlakMagnet]

Crysis isn't that CPU heavy unless you're doing some crazy physics experimentation. It ran extremely smooth when I dropped the resolution way down on my 8800GTX and Core 2 Duo.

Oh stop it! You're getting me *really* excited now :D

<me in squeeky school boy voice>Crysis... 60 FPS.... on max... he he he...</me in squeeky school boy voice>
(nana2)

[Johnny C]

if that card were made to last even a year ATI would have made it standard out of the box. Why in the world would they nerf their card 38% purely from a voltage limit? Reliable ? :thumbdwn:

http://support.asus.com/repair/repair.aspx?no=201&SLanguage=en-us


3 year warranty....who cares if the card dies every year for 3 years(Kinds like my Evga 8800GTX). It's 20% faster across the board and I can buy it today.

Your challenge was how will ATi compete with the 40% faster "paper masche" Fermi cards.

I found a product that meets those performance requirements that anyone could buy right now.

Lately NV sucks. They are doing petty things like disabling physx for ATi users and showing "mock up" boards as the real thing(think how desperate they are if they don't even have a board for their CEO to hold up). I hope that Fermi is fast and cheap. Fast it may be, cheap it won't be.

On the other hand, AMD has brought new dx11 products to the table, in an uncontested market, and are we paying $649 for a 5870? No. This is the opposite to the GTX280 last year, where Nv gouged consumers then issued credits to the early adopters when pricing dropped ~$200 a month after release to compete with AMD's offerings.

wipe the green off your nose and open your eyes.

Even if Fermi is the next messiah, you still won't be able to buy it till 2010.

Cheers Mate,

JC

[InqWoN1776]

Johnny C., you can post vendor OC’d graphics cards all you want but it remains that the market will not depend on them to make up for any potential performance/feature/architectural delta. Regardless, AMD has great cards. At the moment, they have the performance crown. I doubt that these benchmarks are real… However, if a performance delta of this magnitude is present between the single gpu graphics cards upon fermi’s release…. ATI will have a difficult time to put it mildly.

[Ninja Prime]

http://www.percent-change.com/index.php

I was looking at the 3dmark scores, which is odd, because usually the 3dmark scores show more spread than in game benchmarks.

16098 vs 23098, 43%, aka, about 40%.

[josiahsuarez]

http://support.asus.com/repair/repair.aspx?no=201&SLanguage=en-us
I found a product that meets those performance requirements that anyone could buy right now.


you're assuming their won't be factory overclocked Fermi boards as well. Asus makes nvidia cards too...