The GT300/Fermi Thread

[LordEC911]

Actually, all he has posted has turned out true so far. (Which is 3DFinity or whatever, the deep dive, the NDA date.) And seeing as how he said the power consumtion for GTX380 was also huge, it's probably in the ~250W or else it wouldn't have been all that much. A 250W card should beat a 188W card.

Nvidia's answer to eyefinity has been known about for quite awhile, first heard about it over a month ago. The deep dive and NDA expiration date had been known for a couple of weeks.

Not saying he is wrong or pulling our leg but all that info has been floating around for quite awhile.

[marten_larsson]

Nvidia's answer to eyefinity has been known about for quite awhile, first heard about it over a month ago. The deep dive and NDA expiration date had been known for a couple of weeks.

Not saying he is wrong or pulling our leg but all that info has been floating around for quite awhile.

Okay, haven't read anything about it but it figures.

[trinibwoy]

Not saying he is wrong or pulling our leg but all that info has been floating around for quite awhile.

Not in public.

[annihilat0r]

[Manicdan]

SSD's ftw? Yes, hardocp does charts. You can still have half the time in the teens half the time in the 80's, resulting in a 50fps average with a miserable experience .

exactly why its nice to use a chart and see what does and dosnt change between setups of multigpu or changing brands.

i also think its up to the user to determine the best settings for their games for their rig. if somehow turning on 8xAA on a 5970 cut the fps in half, leave it at 4x and turn on the ultra high textures (you get the idea). reviews are great, but they never give 100% of the picture. (i still hate how many people dis the 2900, which is a beastly card as long as you dont need AA, but it handles high rez and lots of effects quite well)

[zerazax]

Just need some damn games worth playing these days...

[LordEC911]

Not in public.

Hmmm... guess it depends on your definition of public.
I first heard about it reading on some forums.

[Andrew LB]

Sounds interesting. Do you happen to have a source? Or were you at the deep dive meeting yourself?

TSMC recently announced to share holders and many financial news websites that ALL yield problems with manufacturing .40nm complex graphics chips have been resolved. This is further proven by how ATi only now is able to deliver large quantities of chips to their 3rd party manufacturers and cards for consumers.

I've seen news articles from tech websites posted here as well.

[***Deimos***]

Deep Dive meeting has over, and we already know how powerful GF100 is, and IT IS! Performance is better then expected, most peoples here will search in own words and feel to shy for craps in thread. We can only laugh now, because Nvidia did it! Perf in Unigine DX11 benchmark is spectacular, Radeon HD 5870 is there like low-end toy for kids against GF100.

PS. Yeah, GF100 has 512 CC, and GPUs are in massproduction now, plenty of manufacturers have first GF100 inhouse.

nVidia did it!? What... 'cause I dont see anything in stores yet.

Also, I'm sure you love playing Unigine for many hours, but its not the type of game most people play. Performance in benchmarks can be misleading and quite different between a variety of games.

Actually, all he has posted has turned out true so far. (Which is 3DFinity or whatever, the deep dive, the NDA date.) And seeing as how he said the power consumtion for GTX380 was also huge, it's probably in the ~250W or else it wouldn't have been all that much. A 250W card should beat a 188W card.

R600 used more power (and had more bandwidth) than G80. Are you suggesting 8800GTX users were stupid, and that X2900XT was faster?

Likewise, the nVidia fx 5800 Ultra used a lot of power too. Performance was good as long as it was DX8. In contrast, the fx DX9 performance was spectacularly miserable at best. Anybody here remember the poor rendering quality and low performance?

High power usage just means better home heating.

[Tim]

Deep Dive meeting has over, and we already know how powerful GF100 is, and IT IS! Performance is better then expected, most peoples here will search in own words and feel to shy for craps in thread. We can only laugh now, because Nvidia did it! Perf in Unigine DX11 benchmark is spectacular, Radeon HD 5870 is there like low-end toy for kids against GF100.

PS. Yeah, GF100 has 512 CC, and GPUs are in massproduction now, plenty of manufacturers have first GF100 inhouse.

If true, sounds much better then expected. And 512 CCs, that's good too.

I really hope this will be another G80.

[annihilat0r]

http://forum.donanimhaber.com/m_36999404/tm.htm

It says that "volume production" will start in the 3rd week of February, some cards will be available in March but not a lot.

I think what this article means by volume production is the placement of the GPUs on the PCB's, and what JHH meant by volume production is that the wafers are in the oven in TSMC.

[Andrew LB]

nVidia did it!? What... 'cause I dont see anything in stores yet.

Also, I'm sure you love playing Unigine for many hours, but its not the type of game most people play. Performance in benchmarks can be misleading and quite different between a variety of games.

While synthetic benchmarks do occasionally favor one particular hardware design over another, especially with a little influence from said particular company.... the facts in the situation tend to be quitte stubborn when it comes to Unigine.

First off, ATi (now AMD) has been listed on their website as a development since I first heard about Unigine a few years ago. And after just recently visiting their website, it seems they finally added nVidia... albeit to the very bottom of their "partners" list.

http://unigine.com/company/partners/

Also, I believe everyone here knows that ATi has been way ahead of nVidia on making DirectX 11 hardware, which Unigine used for the majority of the development cycle since we all know nVidia only recently had Fermi samples running well enough to show off in a company controlled demo. Do you honestly think nVidia would have handed out any of these early Fermi cards 6 months ago to a small 3rd party development team?

R600 used more power (and had more bandwidth) than G80. Are you suggesting 8800GTX users were stupid, and that X2900XT was faster?

The R600 was just a crappy design by a graphics company hurting financially and only saved by AMD buying them up.

Likewise, the nVidia fx 5800 Ultra used a lot of power too. Performance was good as long as it was DX8. In contrast, the fx DX9 performance was spectacularly miserable at best. Anybody here remember the poor rendering quality and low performance?

So we are to automatically assume Fermi is going to be another flop like the FX? Are you forgetting that once we got basic architecture specs for FX, everyone at B3D and other well informed sites all agreed was going to suck. Especially since it took about a year and countless "tape outs" just to get a chip stable enough to show the public a demo.

I think people with ample skills at using reason can see that it was only logical for nVidia to delay Fermi because of the huge problem of .40nm production at TSMC, something that plagued ATi for the past 5 months. So instead of releasing a product that had extremely high production costs and very low yields which was almost impossible to purchase.... nVidia decided to not release specs or performance data so they and their 3rd party card manufacturers could sell off warehouses full of product.

I don't think nVidia is losing much sleep at night knowing that being #2 in performance for a few months is not as bad as the huge costs of full scale manufacturing at 40% yields AND having the performance crown while only having a small number of those costly cards available for the public to purchase.

If true, sounds much better then expected. And 512 CCs, that's good too.

I really hope this will be another G80.

After seeing that Unigine and Rocket Sled video on Youtube and the incredible framerates and detail with tessalation running.... I have a feeling you wont be dissapointed.

.... the people who spent the past few months beating their chests over the HD5xxx cards performance will just grab some tissues and sulk for the next year or two until ATi gets the crown again for a short time. I'm sure all the sane ATi and nVidia fans will be glad to have the endless thread crapping, lies by Charlie, and other nonsense just go away for a while. I sure as heck will.

[DarthBeavis]

This just in!! Breaking News!!!

Attendees at PDXLAN will be amoung the FIRST LAN attendees to see GF100 in action!!! That's right, the NEXT, NOT EVEN RELEASED Video Card from NVIDIA!!!! Taking it a step further, CLUB SLI members can win one of the first ones AUTOGRAPHED by the CEO of NVIDIA!! ZOMG!!!!! To win one, it's all part of a contest taking place ONLY at PDXLAN 15!!!

All contest details at www.pdxlan.com - Last day to get tickets to PDXLAN is Tuesday night!! Register at pdxlan.com!!

[annihilat0r]

nice to see some marketing/publicism about the GeForce brand itself again

[DarthBeavis]

nice to see some marketing/publicism about the GeForce brand itself again

More to come . . .I will be involved a bit

[saaya]

I thought nvidia didn't do hard-launches anymore?

hehehe, well, theres always hope...

in the past it was announce and launch on the same day
then it was announce first and launch later
in recent years that become: announce, then wait, then launch, then wait, then sell



nvidia announced fermi at GTC, they will launch it in february, 4 months later... im really curious when it will actually sell

If there going with 256 CC then that be it self shows there in bad shape. Think about it, that's half the CC turned off on a die that's very large and that cost alot to get. The only reason I can think of for them doing this is trying to find a way not to make money but trying to find a way to break even and not be in the red. They will be selling cards with a 256 CC at a loss but its better then trashing them. Even a 320 CC will be hard to sell to break even on with the low yields and the large size of the die.

yeah, they basically buy good PR by selling cards at a loss then... wouldnt be a first though afaik

Deep Dive meeting has over, and we already know how powerful GF100 is, and IT IS! Performance is better then expected, most peoples here will search in own words and feel to shy for craps in thread. We can only laugh now, because Nvidia did it! Perf in Unigine DX11 benchmark is spectacular, Radeon HD 5870 is there like low-end toy for kids against GF100.

PS. Yeah, GF100 has 512 CC, and GPUs are in massproduction now, plenty of manufacturers have first GF100 inhouse.

cool, sounds like the event was pretty nice if they got you all excited like this
why did they do this behind closed doors and didnt make it public though?
im still sceptical over the claims they threw at you and others during that presentation...

[annihilat0r]

hehehe, well, theres always hope...

in the past it was announce and launch on the same day
then it was announce first and launch later
in recent years that become: announce, then wait, then launch, then wait, then sell



nvidia announced fermi at GTC, they will launch it in february, 4 months later... im really curious when it will actually sell

yeah, they basically buy good PR by selling cards at a loss then... wouldnt be a first though afaik

cool, sounds like the event was pretty nice if they got you all excited like this
why did they do this behind closed doors and didnt make it public though?
im still sceptical over the claims they threw at you and others during that presentation...

Saaya, I never saw you agree with anybody about anything that is Nvidia related. When Fermi shows up, regardless of its price or performance, you will be automatically wrong about around 500 things. If it is a good product, you'll be wrong about an additional 2500.

[neliz]

Well, TSMC does seem to have less yield issues now. Put it another way, at least ONE party seems to be happy with the clocks they're reaching on some products.

[blob]

nvidia's deep dive meeting:
It was all technical. But no clock speeds, power consumption or benchmarks were being shown…yet at least," the source confirmed at the time of writing.

source

[ajaidev]

thanks blob, sadly it was just a white-paper oral...


Using stupid logic one can say if per shader unit the performance of a GF100 to a G200 shader unit is min 5%-10% more we can assume that at similar speeds the GF100 even with 448 shaders can beat a GTX295 without any problem and a 512 shader version can give a additional min 15%-20% boost.

This is off-course done by using stupid logic no idea on the b/w or anything else...

EDIT: I also got this in mail:-


"Dear Ajai dev,



You’re Invited: Join us for a webinar on Wednesday January 13, 2010, 10am – 11am pst

Do Science with Tesla Today. Be First to do Mad Science with Fermi.

NVIDIA’s next generation CUDA architecture, code named “Fermi” is the most advanced GPU computing architecture ever built. Join us for a live webinar to learn about the new Tesla GPU Compute solutions built on Fermi and the dramatic performance capabilities they offer customers who are tackling the most difficult, compute-intensive problems. In addition you will learn about our limited time offer, the Mad Science Promotion, whereby you may qualify for a promotional upgrade to a new NVIDIA Fermi-based Tesla product when you purchase a NVIDIA® Tesla™ C1060 GPU Computing Processor or a S1070 1U GPU Computing System today.

Who Should Join: Researchers, Scientists and other Professionals looking to accelerate compute-intensive applications with the GPU

Presenter: Sumit Gupta, Sr Product Manager, NVIDIA GPU Compute Solutions

When: Wednesday January 13, 2010, 10am – 11am pst""

[LordEC911]

Also, I believe everyone here knows that ATi has been way ahead of nVidia on making DirectX 11 hardware, which Unigine used for the majority of the development cycle since we all know nVidia only recently had Fermi samples running well enough to show off in a company controlled demo. Do you honestly think nVidia would have handed out any of these early Fermi cards 6 months ago to a small 3rd party development team?

Maybe not 6 months ago, they didn't have anything 6 months ago...
Now about 2 months ago, they definitely did.

The R600 was just a crappy design by a graphics company hurting financially and only saved by AMD buying them up.

Hmmm... ATi was hurting financially? They were actually doing very good at the time. Pretty sure everyone agrees AMD bought them at the wrong time.

I think people with ample skills at using reason can see that it was only logical for nVidia to delay Fermi because of the huge problem of .40nm production at TSMC, something that plagued ATi for the past 5 months. So instead of releasing a product that had extremely high production costs and very low yields which was almost impossible to purchase.... nVidia decided to not release specs or performance data so they and their 3rd party card manufacturers could sell off warehouses full of product.

Actually, the only thing that was limiting AMD/ATi was capacity, not yields.
Ahhh... so you are one of the few that think Nvidia ran hotlots, risky, and did 2 respins for "fun." Good one!

I don't think nVidia is losing much sleep at night knowing that being #2 in performance for a few months is not as bad as the huge costs of full scale manufacturing at 40% yields AND having the performance crown while only having a small number of those costly cards available for the public to purchase.

Nvidia would be in heaven if they were even close to 40% yields.

.... the people who spent the past few months beating their chests over the HD5xxx cards performance will just grab some tissues and sulk for the next year or two until ATi gets the crown again for a short time. I'm sure all the sane ATi and nVidia fans will be glad to have the endless thread crapping, lies by Charlie, and other nonsense just go away for a while. I sure as heck will.

Hmmm... you are sure doing a good job here. Thanks for the facts and non-bias comments. As much as some hate Charlie, I'm not sure where you can say he lied. Only recent thing I can remember is that he got some very wrong info on the whole Lucid ordeal.

[To(V)bo Co(V)bo]

Saaya, I never saw you agree with anybody about anything that is Nvidia related. When Fermi shows up, regardless of its price or performance, you will be automatically wrong about around 500 things. If it is a good product, you'll be wrong about an additional 2500.

Whatr you talking about Saaya has agreed with me plenty times. We both agree that nvidias business model is doomed. You cant COMPETE in the gpu market, if you aint making gpu's period. Fermi is a trickle down product that has no business being in a desktop. Fermi is designed only for being a gpgpu cruncher. The only way nvidia can truly compete is if they split their designs and make dedicated ppu's and dedicated gpu's. You cant have one gpu that works both ways efficiently!

[eric66]

nvidia's deep dive meeting:
It was all technical. But no clock speeds, power consumption or benchmarks were being shown…yet at least," the source confirmed at the time of writing.

source

so that zed x guy was lying then ?

[Nedjo]

but you know what?
people are spending that on cpus for years, and a highend vga is 10x more worth it than a highend cpu imo...
actually im surprised highend vgas havent been selling for more for the past years...

it is phenomena of its own that's OK when Intel sells CPUs for the price of $560+ and $1000+, and when VGA manufacturer offers killer card with the processing unit that is way more complex, and with bunch of GRR5 memory chips for 300-400 USD everyone is complaining how expensive it is!

so I'm complitely with you on this one!

[Nedjo]

because a quick spike could simply be from a HDD loading issue

i prefer charts (hardocp does that right?)

finally something that we both agree on