Fermi hardware might still be
two months away, but NVIDIA has done the sage thing and released some tantalizing numbers and architectural details to keep the fanboys chirping in the meantime. The GF100 will signal the end of tiresome rebadging and clock speed massaging, and early adopters will find 512 CUDA cores, 48 ROPs, and a 384-bit GDDR5 memory interface sprawled across
three billion transistors. Big changes are also afoot in how the card will do its work, with a reorganization toward a more parallel workflow leading to promises of up to eight times better geometry performance than on the GT200.
HardOCP reports that anti-aliasing results have improved "notably," while the video we've got stashed after the break for you shows the GF100 beating the
GTX 285 handily in a
Far Cry 2 benchmark. Still, the
PC Perspective crew expressed some apprehension about the massive die size and how it could impact yields given the still young 40nm production process -- a sentiment echoed by other publications who questioned whether NVIDIA would not have been better off trying for a less ambitious, more gaming-oriented board. We should all know that answer soon enough.
Read - AnandTech
Read - Hot Hardware
Read - PC Perspective
Read - HardOCP
Read - Tom's Hardware
posted January 18th 2010 1:15AM
Can't wait for the Radeon 5000 series to drop in price now!
@M C yea but only nvidia has 3D! wish ati could bring it to their cards , can anyone comment about the possibility that ati will add 3D?
@Bryce
I'll throw it in the air and say 5890.
Nvidia IS like half a year late for DX11, so...
@MegaJapan late? why late? how many games do take advantage of dx11 today? how many games were you crying ati crocodile tears because you couldnt play them on your perfectly fine for ANY game released today gtx260core216 you can get at 150 euros, TODAY?
Dirt2. One game. Dirt2. You think i would buy a 5870 just for Dirt2?
Nvidia works with game devs, The batch of games taking advantage of tesselation will come WHEN nvidia has cards out.
If anything you could say ATI suffered of premature ejaculation. 1 Game for a few months, was all they could base their harsh marketing on. Well... that, and the fact they own Charlie Demerjan and his buddy Ryan Shrout.
Fact you can do 3 monitor gaming using older cards is great. I wonder if we can do hybrid generation or both cards need to be same. We can already do triple gaming using software solutions like http://www.kegetys.net/SoftTH/ which is great (thanks kegetys!)
Hell, my current card can even be used to power nextgen gaming by dedicating it to PhysX. Now thats value. My investment from the past gets used to amplify my future. Funny no one talks about it in these opinion making reviews, because they all LOADED, and get hardware for FREE (actually they get paid without getting paid).
I'll just have fun with some popcorn, reading comments for fanboys on either side.
ps: I applaud nvidia for the socially responsible work and donations you have been doing even before haiti.
@doutorpiranha They don't work with game devs. They pay them cash to get the "MEANT TO BE PLAYED WITH HURRVIDIA" logo on the game bootup, and to have exclusive graphics perks that the radeons could render just as well or better. I can't support a company that does that, I will be going with ATi [my current card is an Nvidia 7900gtx, I'm no fanboy]. The sooner the nvidia cards come out, the sooner I can get my radeons for cheaper :3
Man I am excited to check this out.
@M C
Can't wait for the high-end Radeon 5000 series to stop being so ridiculously huge.
@Luffy
That's a fine argument for your fanboy friends, but around here people are aware of the existence of games like:
Batman Arkham Asylum: that can only use PhysX with nvidia cards
or
Saboteur: that wouldn't even RUN on ATI when it was first released.
You're going to have to do better than 'LOLNO' when its clearly true.
@M C nobody answered meeee!!!!
@E71 For starters, yes its closest rival company got a headstart with them on manufacturing DX11 cards.. Albeit, Nvidia is now, not only focusing on PC-GPU cards but on mobile procs as well. Understanding their Tesla/Tegra: http://bit.ly/tesla-and-tegra-what-are-they
That's why there's some stalling but hey, I think its good to see what this company has to offer again for gamers alike.
@Bryce
thats cause ur a disgrace to the community here
wtf does 3d have to do with anything here
We need the real card.
I predict we're a year, maybe 2, away from external GPUs. These things are getting so big and needing so much power & better cooling that it seems inevitable that they'll escape the PC case and live in their own air- or water-conditioned little boxes.
@CRA1G That is closer than you realize.. http://www.pcper.com/comments.php?nid=8222
@CRA1G
i could see that being awesome or sucking
the more vacuum tubes the more awesome
@CRA1G
What I'd like to see happen is dropping your video chip of choice onto the mo.board like the processor.
@CRA1G Vidock 2 a external GPU for laptops already
Cool! In two months I'll be able to afford a Radeon 5970!
@spaz1 lets hope! then again i'm still running DDR2 RAM...
@kojo87
DDR 133 here.
just wait till 6000 series comes out...
i've been an nvidia card owner since back in the TNT era but i'm wondering wether they'll manage to pull through ...
i don't remember if it was that similar but this delay reminds me of the bad moves and delays by 3DFx back in the day ...
@neonn "but i'm wondering wether they'll manage to pull through ..."
They will: nVidia has diversified now its portfolio enough. Though I suspect that desktop graphics is still the cash cow.
My only question is - how much would it cost? nVidia has the problem with their product strategy, treating mass market as an afterthought.
As much as I prefer their quality to ATI, it is very hard to justify prices of the nVidia.
The good new between the lines is:
- The "old-gen" amazing-and-totally-adequate cards you couldn't afford are gonna get a price drop.
- ATI is gonna release something similar only notably cheaper.
No wonder PC Perspective is AMD BIASED.
Ryan Shrout is AMD "sponsored". He prefers to talk about an external laptop card addon for a niche market by amd than to talk about FERMI BEAST that will power 2010 nvidia pc machines!
Here's a nice realtime raytracing output from 3 gpus that will change 3D modeling as we know it (basically wiremesh will be worked overlayed on top of constant updated rendering (lightwave core etc).
http://www.youtube.com/watch?v=Cbnv_z6VDj8
ATI fanboys (sponsored or not) or any kind of fanboys are annoying noise.
@doutorpiranha
Watched the vid... I see what you mean. That is very nice.
Of course, my comparatively pedestrian 9600 GSO still holds up rather well for me. Then again, I'm also not a gamer so there's no way I will ever truly tax any of those cards, so...
Just my 2¢...
@doutorpiranha Wow man, you're way off base.
"talk about FERMI BEAST that will power 2010 nvidia pc machines!"
"ATI fanboys (sponsored or not) or any kind of fanboys are annoying noise."
I'm die hard NVIDIA personally, but you sound more like a fanboy than I or any review site I've ever seen for that matter.
http://www.pcper.com/content_home.php?s=5
For AMD fanboys there sure is a bunch of GeForce in that list. Oh look, Fermi was covered 4 months ago. You claim fanboys are annoying noise? Stop being so noisy then.
@malventano malventano, get back in your submarine man :)
no need to come defend your buddies from pcperspective. You work with them, have a show with them and do nice ssd reviews for them. Isnt that enough? you also need to come here and spread the nvidia vitriol?
@doutorpiranha
How come you know the performance of a not even released card?
How come your given proof except the common "Oh they will kick ass, (at least i want them to)!" is a crappy youtube video?#
How can you be sure that the vid is a rendering on 3 GF100 ?
- There might be some seriouse cluster right left of the LCD rendering
- There might be a 3 Nvidia GTX 295 Mars working overclocked as hell in that case
- There might even be ATI Cards in the case.
The thing is this vid and your "GF 100 Beast statement is nothing that backups your GF 100 ftw statement.
@yakill go back to the place where they pay you to be spin doctoring semiaccurate.com crap. Thats the only place where nonsense is debated as news.
i said nothing about performance that isnt stated on the BENCHMARKS the articles show. Also the demo of 3x gf100 is welldocumented on youtube, and you also have single card demos. what exactly is your problem? Unlike many i dont earn to be spreading filth. Im just a user, glad tech is advancing. Unlike some here i dont need disclaimer.
How come all gf100 demos area always with 3 cards? Does it even run as a single? Who besides gates can afford 3 gf100? And will gates be content being poor after purchase or will he miss all his money?
Dammit Nvidia, I don't want to wait 2 more months for 5850s to hit $200.
2 months out and still just vague details? Me thinks there is going to be a nice shiny paper launch of this product in a few months with actual hardware to ship at a later day. Been done before by Nvidia and ATI. According to a few articles production is already under way, but you would think they would list such details as core and mem clock as under NDA, not as 'not available at this time'.
So will this be available in mobile form for a laptop?
@Paul
Yes but not in march, they said in CES Q1 (March) for desktops and Q2(June) for laptops
Intel will release 32nm Corei7 Quad in june too so it's great time to buy a laptop in 2010 summer
@GoogleCEO
Care to source that?
Everything(except one line on Wikipedia) I've seen points to Intel sticking with 45nm for quads until Sandybridge in late Q4.
And this is where my 5870 starts showing its age...
Oh well, as long as it can do Crysis :D.
@EA575
Isn't it awesome? :D
That's a lot of big numbers I don't understand, but they are bigger than the last numbers so they are better!
@ch3burashka
Yea "New numbers no one understands that are bigger then the old numbers no one understands!!" Funny!
@EA575
It is the hour that the NDA is lifted.
" In short, here’s what we still don’t know and will not be able to cover today:
Die size
What cards will be made from the GF100
Clock speeds
Power usage (we only know that it’s more than GT200)
Pricing
Performance"
Yeah but who needs to sweat those details right?
This seems really surreal. But then it IS surreal, since we don't have a REAL CARD.
But I will welcome the day Fermi stumbles into notebooks.
They already missed the holiday season, are 6 months late to the DX11 party, and even their mobile discreet cards are being displaced majorly by AMD. Fermi better be good, Nvidia.
For the price this thing will undoubtly have, I find the video very underwhelming. I've been a Nvidia diehard for a long time (my current system is my 6th SLi rig to date between my teenage son and myself) but looking at the farcry demo apart from the frame rates on the screen showing different numbers you wouldn't be able to tell the two screen from one another. Both rendered the scene and the explosion equally well. When you spend 400+ dollars on a video card it should give you that "hell yeah!" feeling the first time you start up your game compared to the last generation piece.
I've been looking at the ATI line seriously for the first time in 10+ years for my next build.
The whole PhysX thing is not much of a feature as some would make it out to be. Like someone said previously you can't talk DX11 as a feature with only one game that really uses it and the same goes for PhysX, which only has a handfull of games that really use it well to begin with. Also people will say "but Nvidia has CUDA!" , but having spent the money for the CUDA enabled versions of several video transcoding application, BR players and other applications I am very less than impressed with them so far. Yes they transcode faster but so far they seem to trade some picture quality for the speed.
133% increase at 8x on far cry 2.
for all we know they could lower the resolution from 2560x1600 to 1280x1024 and/or optimized the crap out of the engine for that build like they do with console games.
Can not wait for this baby. I might wait to buy behind the curve, but this is still looking pretty good so far.
I was comparing the settings on the 2 result screens, and they were both the same BUT... the left screen has as date above it 1/10/2010 and the right side has as date 1/13/2010, yet it's suppose to be a live comparison.
Draw your own conclusions.
Nvidia GF100 pulls 280W and is unmanufacturable
Details it won't talk about publicly
http://www.semiaccurate.com/2010/01/17/nvidia-gf100-takes-280w-and-unmanufacturable/
@alvin3486
That's written by Charlie Demerjian who used to work for theinq and is generally considered a bit overwhelmed with his dislike for nvidia, although he does know a lot of insider stuff and isn't exactly stupid, however he does seem overwhelmed by his emotion of that subject.
And since nvidia is reportingly already baking the chips we'll see in 2 months how it pans out.
Incidentally being ambitious in design isn't necessarily a bad thing, making stuff in tiny baby steps doesn't advance technology and just makes people pointlessly pay over and over for small improvements.
Last but not least ATI has a history of putting stuff in that didn't really work too, sometimes it's hard to predict how the final chip comes out it seems, and the more complex the trickier I imagine, it becoems harder and harder to emulate all interactions of components when you are dealing with millions upon millions of transistors.