View Full Version : Nvidia to demo Fermi-based Tesla GPUs November 17
josiahsuarez
11-13-09, 03:54 AM
http://www.nvidia.com/object/sc09.html
See “Fermi” in Action
Come see a live demonstration of the next generation CUDA™ GPU computing architecture, codenamed “Fermi”. The Fermi-based Tesla™ GPUs are mass market parallel processors and fuel the HPC industry’s transition to parallel processing. Compared to the latest quad-core CPUs, the Fermi-based Tesla GPUs deliver equivalent performance at 1/20th the power consumption and 1/10th the cost.
LordJuanlo
11-13-09, 04:39 AM
I don't want to see f***ing Tesla, I want to see GAMES
CaptNKILL
11-13-09, 04:49 AM
I don't want to see f***ing Tesla, I want to see GAMES
Agreed.
AMC_Duke
11-13-09, 05:08 AM
I don't want to see f***ing Tesla, I want to see GAMES
+1
I don't want to see f***ing Tesla, I want to see GAMES
+1337
Noriega
11-13-09, 08:18 AM
I don't want to see f***ing Tesla, I want to see GAMES
+PONY!
candle_86
11-13-09, 03:15 PM
+candle
FlakMagnet
11-13-09, 04:48 PM
+plus
Johnny C
11-13-09, 05:51 PM
I don't want to see f***ing Tesla, I want to see GAMES
It might be a while before you see any real games....nV is still trying to figure out how to make Fermi run them....
The good news is...it's really good as an add-in card for dedicated Physx....
HAHA...
Ninja Prime
11-14-09, 01:35 AM
http://www.nvidia.com/object/sc09.html
That's a great brag, I can't wait to load up windows on my new NV GPU thats 1/20th the whatever marketing units more betterer than my quadcore CPU... oh wait.
Fail.
That's a great brag, I can't wait to load up windows on my new NV GPU thats 1/20th the whatever marketing units more betterer than my quadcore CPU... oh wait.
Fail.
On what reasoning is it a fail. Nobody but Nvidia knows that atm. It could be a real winner and blow the competition away. Only when we get some real info will we know for sure. Either way it's to early to call it a failed GPU.
LordJuanlo
11-14-09, 07:51 AM
Why would nVidia show Tesla before gaming cards?. They know their fans need something to believe in, something that will stop them from getting ATI cards on Christmas. The only thing that I can think of is... it does not work yet. Hardware problems, driver problems... maybe performance problems?. I don't know, but I smell something fishy here
josiahsuarez
11-14-09, 08:06 AM
Why would nVidia show Tesla before gaming cards?
Tesla is much more profitable than gaming cards
Johnny C
11-14-09, 12:36 PM
Why would nVidia show Tesla before gaming cards?. They know their fans need something to believe in, something that will stop them from getting ATI cards on Christmas. The only thing that I can think of is... it does not work yet. Hardware problems, driver problems... maybe performance problems?. I don't know, but I smell something fishy here
Ding Ding.....we have a WINNAR!
If you had the next big thing and wanted your target audience of consumers to put off a major purchase in order to capture those sales...wouldn't you want to demonstrate the cards full potential as soon as possible?
Only 2 reasons there haven't been leaked gaming performance numbers....either:
5800 series currently beats or matches Fermi gaming performance or...
Fermi simply can't run games as of yet.....
I'm going with the second reason....since they can't even demo the damn thing running a game to show it works....
Maybe all of those wood screws are causing EMI problems or grounding issues with the card...
Who knows???
lightman
11-14-09, 01:08 PM
Maybe it's simply because 1) as someone already said, Teslas are much more profitable than graphics cards, and 2) we scientists (or at least quite some of us) are almost salivating and can't wait to get one (or in some cases many more). Fermi is expected to be incredibly fast in double precision floating point, a thing that is seriously missing in current Teslas...
Moreover, it looks like TMSC is having quite some yield problems, so it is quite an understandable thing to go with the higher margin parts first.
Fermi is starting to look like "Lame-i".
Johnny C
11-14-09, 02:06 PM
Maybe it's simply because 1) as someone already said, Teslas are much more profitable than graphics cards, and 2) we scientists (or at least quite some of us) are almost salivating and can't wait to get one (or in some cases many more). Fermi is expected to be incredibly fast in double precision floating point, a thing that is seriously missing in current Teslas...
Moreover, it looks like TMSC is having quite some yield problems, so it is quite an understandable thing to go with the higher margin parts first.
Excellent point if we were talking about why nVidia was supplying Tesla demand before AIB partners, but that's not what what I was talking about. Since there is no supply or demand at this junction either way.
I'm talking about why...nV has not shown a single game demo running DX11, or even DX10 for that matter, with the new Fermi architecture....
There are no leaked performance numbers, at least that have any merit what-so-ever.
I'll agree that there are probably some scientists out there fighting a bad case of tent-pants over this demo...but like every other GAMER out there...I want to see some GAMING performance numbers...
I've got 500 bucks burning a hole in my pocket....and I'm asking the same silly question as everyone else....should I wait for Fermi?
This totally reminds me of the 7950GX2 Quad SLI marketing...oh it sounded sooooo gooood all that raw processing horsepower and I took the bait, spent the cash, and was rewarded with a very disappointing gaming experience....one which left me longing for both my 7800GT's SLi setup and my hard earned cash....
LydianKnight
11-14-09, 04:34 PM
Why would nVidia show Tesla before gaming cards?. They know their fans need something to believe in, something that will stop them from getting ATI cards on Christmas. The only thing that I can think of is... it does not work yet. Hardware problems, driver problems... maybe performance problems?. I don't know, but I smell something fishy here
Why do you think NVIDIA will be able to stop ATI dead on their tracks on Christmas time? I never trust Mr. Demerjian's articles as they're full of hatred and lack of objectiveness but... if even Fuad says an A3 revision will be probably needed... I would not count on NVIDIA this year.
About Tesla products for the HPC market? Yes, probably the A2 silicon is good enough for raw computing, but for games the better refined you get your silicon, the better, and it could be not... whether if this is true or not, I don't really know... I would like to have a Fermi card for christmas but I'm starting to make my mind for a late Q1/Q2 2010 launch...
i do not care if the cards come out this year i am leaving Sweden fore Hout Bay near Cape Town South Africa over X-mas and new year and a bit in January so i wont buy any fermi cards earliest in feb or march
http://www.southafrica-travel.net/westcape/capepics/hbpanorama.jpg
http://www.southafrica-travel.net/westcape/capepics/hb_02.jpg
lightman
11-15-09, 07:38 AM
About Tesla products for the HPC market? Yes, probably the A2 silicon is good enough for raw computing, but for games the better refined you get your silicon, the better, and it could be not...
Not to sound rude, but you don't know what you're talking about.
HPC is all about being sure of the results you're getting. If a single pixel of a displayed texture in a game is not RGB(255,250,123) but it is RGB(254,251,126) no one would notice or care. Now, if I get, e.g., a value of 2.007129 instead of 2.007131 it could easily screw up some computations (especially if it is not a reproducible error, and thus beyond software correction).
So in the end, it matters way more to the HPC crowd (myself included) than to the gamers crowd to have a perfect (whatever "perfect" might mean in this particular case) silicon.
I would be really disappointed (or, better, incredibly angry) if my $2000+ Teslas would give me perfectly unusable results, and as myself everyone else in the scientific community. And that would be such a bad press that NVIDIA would easily lose almost all its (potential or real) market within the HPC community. I can assure you of that.
LydianKnight
11-15-09, 08:02 AM
Not to sound rude, but you don't know what you're talking about.
No, I don't take it as a rude comment, no worries :P
The comment I made is just based on my own thoughts, not real knowledge. What bugs me is the following: as other users state, there's not a single gaming performance bit of data, but they're going to demonstrate Fermi for the HPC market, ok...
Now some sites (which I trust (relatively) some of them wheras not others... (but based on my own opinion)) says NVIDIA will probably need an A3 revision of their Fermi silicon, ok...
Now the difficult part, extrapolation.
A. NVIDIA is currently on A2 silicon
B. NVIDIA is going to show Fermi for the HPC market
C. NVIDIA doesn't provide any gaming numbers
D. NVIDIA would probably need an A3 revision of the Fermi silicon
What would you think at first thought? It's not like NVIDIA is considering the A2 silicon 'good enough' (like I have posted) for HPC, but why Tesla is ready for showtime before any kind of gaming demonstration? Maybe the current A2 silicon works good enough, maybe with a higher-than-expected TDP, less clock speed, whatever, or maybe not... we don't have solid information into this, and NVIDIA is not precisely very helpful at the time to mitigate false or incomplete or unreliable rumors, so all we have is... extrapolation.
That's where my comments come from, nothing more, but it's not that ilogical to think, or maybe it's just they're saving the gaming numbers for later, but doesn't make sense...
josiahsuarez
11-15-09, 09:42 AM
Nvidia really has no need to show any gaming numbers until 5870 is produced in larger numbers. they're still pretty scarce as of right now.
CES in January should be when they talk about a "GTX 380" or whatever it will be called. probably the sky won't fall on Nvidia between now and then.
shadow001
11-15-09, 01:03 PM
Maybe it's simply because 1) as someone already said, Teslas are much more profitable than graphics cards, and 2) we scientists (or at least quite some of us) are almost salivating and can't wait to get one (or in some cases many more). Fermi is expected to be incredibly fast in double precision floating point, a thing that is seriously missing in current Teslas...
Moreover, it looks like TMSC is having quite some yield problems, so it is quite an understandable thing to go with the higher margin parts first.
Well to put it into perspective,the HD5870 does 2.7 terraflops in single precision math,and dual precision math is 1/5th that value,so it comes in at about 550 gigaflops,which isn't exactly a slouch,and still way faster than current teslas as well.
Rumors suggest Fermi cranks that up to about 700 Gigaflops,better than the HD5870,but not a night and day difference either,and that's ignoring the delay it's suffering,and that a refresh version of the HD5870,probably clocked much closer to the 1 Ghz mark,would likely close into that 700 Gigaflop double precision floating point math of fermi quite easily.
shadow001
11-15-09, 01:11 PM
Nvidia really has no need to show any gaming numbers until 5870 is produced in larger numbers. they're still pretty scarce as of right now.
CES in January should be when they talk about a "GTX 380" or whatever it will be called. probably the sky won't fall on Nvidia between now and then.
Seems that according to ATI,they've sold close to 100 000 HD5800 cards so far,so at 300 to 400$ a pop(depending which version purchased),so that's already between 30 and 40 million in sales and it's only been 7 weeks since it was released,and the christmas shopping season is coming,and the Bad yeilds from TSMC won't continue right up until the month of march/April of next year,when the gaming version of fermi is released either,so the pressure on Nvidia is very much on,no matter how much damage control they try to do.
Ninja Prime
11-15-09, 03:59 PM
On what reasoning is it a fail. Nobody but Nvidia knows that atm. It could be a real winner and blow the competition away. Only when we get some real info will we know for sure. Either way it's to early to call it a failed GPU.
*whoosh* Thats the sound of the joke going over your head.
I can't wait to load up windows
They want to compare themselves to a CPU, they can't do half the things I can do with my CPU, get it?
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.