Overclock.net banner
Status
Not open for further replies.

[vr-zone] Fermi in trouble?

35K views 429 replies 117 participants last post by  Burn 
#1 ·
Quote:


Fermi in trouble?

We all know it is constantly being delayed - but at the end of all those delays, what we were expecting was a stellar product from Nvidia. However, with today's press release, certain inconvenient details are revealed. Let's forget about the delays for now, and just consider the product itself.

The first Fermi GPU - GF100 - as we know for a while now, is a 3 billion transistor giant, taking a die size of around 500 mm2. Compare this with the 2.15 billion transistor, 330 mm2 Cypress on the same 40nm TSMC process, and you would be expecting a different class of product. Unfortunately, the details revealed today about cast an uncertain shadow over this basic assumption.

The first thing worth noticing is a complete and total absence of Single Precision performance figures or any comparison to direct competition - i.e. ATI's GPGPUs. It is clear that Fermi's real performance advantage would be Double Precision performance - had it hit the right clock speeds.

However, today's press release suggests Nvidia have missed target speeds by a lot. To be fair, Tesla products do clock lower, though not by much. In fact, GTX 280 and Tesla C1060 were clocked the same. Even taking a generous increase for Geforce products, things are still uncertain. As a result, DP performance is rated at between 520 GFlops and 630 GFlops. Suddenly, ATI Radeon HD 5870 - which wasn't even supposed to be a direct competitor - is performing right on par with 544 GFlops against Fermi's supposed strong point.

Consider Single Precision - far more important for gaming graphics, and things turn rather ugly. GF100's target speeds were reported to be...

Source
 
See less See more
#4 ·
Speculative speculation is speculative.

Very interesting.

This is the scary part.
Possibly paying premium ATI prices for many more months.

Quote:


And we have not factored in the fact that GF100 is nowhere to be seen, and are unlikely to be on shelves in quantity for at least 4-5 months. Any further delays, and we will be looking at new products from AMD.

 
#5 ·
very interesting. the Q1 10 is good, but everything else is bad. vr is fairly well known for presenting the news... this cannot be good for nvidia =/
 
#6 ·
Quote:

Originally Posted by grunion View Post
Speculative speculation is speculative.

Very interesting.
Speculation?

http://www.nvidia.com/object/io_1258360868914.html

Quote:

  • Tesla C2050 & C2070 GPU Computing Processors
    • Single GPU PCI-Express Gen-2 cards for workstation configurations
    • Up to 3GB and 6GB (respectively) on-board GDDR5 memoryi
    • Double precision performance in the range of 520GFlops - 630 GFlops

5870 gets 544 GFLOPS of double precision, can't argue those numbers.

Looks like nVidia right back where they started. They have a massive die, a much more expensive product, and the performance simply isn't there.

Best case scenario it looks like we're going to get another HD 4890 vs GTX 280 battle again, instead of the GTX 285 vs HD 3450 that the nVidia fanboys were predicting.
 
#7 ·
Can't be good for those who want ATI cards either, due to pricing.
 
#8 ·
Quote:

Originally Posted by nathris View Post
Speculation?

http://www.nvidia.com/object/io_1258360868914.html

5870 gets 544 GFLOPS of double precision, can't argue those numbers.

Looks like nVidia right back where they started. They have a massive die, a much more expensive product, and the performance simply isn't there.

Best case scenario it looks like we're going to get another HD 4890 vs GTX 280 battle again, instead of the GTX 285 vs HD 3450 that the nVidia fanboys were predicting.
Speculation pertaining to the gaming performance.
 
#12 ·
i hope they can pull it together... if not... well, AMD will be raking in the profits, and i wouldn't be surprised if they left the inflated MSRP of the 5850 and 5870 as they stand
 
#14 ·
Quote:

Originally Posted by nub View Post
Quick question. How much time would it take for Nvidia to do a die shrink and add dx10.1 or dx11 to the 260, 275, etc? Is such a die shrink already planned for those gpu's?
You have to make a new core for DX10.1 and DX11. For 3870 it was easy because those feature were already with 2900XT but disabled and the die shrink could take 4-5 months from start.
 
#15 ·
Quote:

Originally Posted by ZealotKi11er View Post
You have to make a new core for DX10.1 and DX11. For 3870 it was easy because those feature were already with 2900XT but disabled and the die shrink could take 4-5 months from start.

It took them nearly 2 years to come out with a DX10.1 version of the 8800GS.
 
#16 ·
Quote:


Originally Posted by sLowEnd
View Post

Is this article using FLOPs to estimate gaming performance, or am I mistaken?

no problems with estimating gaming potential using flops. after all, its what is used to measure processing power. i agree you really cant say THAT much about how its directly proportional but people were expecting it to blow everything else out of the water... this does not look good at all for nvidia.
 
#17 ·
Quote:


Originally Posted by nathris
View Post

5870 gets 544 GFLOPS of double precision, can't argue those numbers.

But you CAN say that the 5870 which is rated at 2.7TFlops SP is typically beaten by the 1.8TFlop SP GTX295. So I think it would be safe to say that 520GFlops DP on Fermi will be considerably faster than 544GFlops DP on RV870.

Quote:


Originally Posted by forcifer
View Post

no problems with estimating gaming potential using flops.


Considering the 1.8TFlop GTX295 is faster than the 2.7TFlop 5870 the majority of the time...
 
#24 ·
Quote:


Originally Posted by sLowEnd
View Post

Really, man?
Are you trying to start a flame war?
That topic has been argued to hell and back. It has no place in this thread.

What am I supposed to say? GT220s are definitely current generation GPUs?

When the second half of Nvidia's business model doesn't involve renaming something and then selling it for more, I'll stop ragging on them. The first half that involved the creation of the GTX 200s is quite deserving of merit.

In relation to Fermi: I am so disappointed. I do not need a replacement CPU.
 
#25 ·
Quote:


Originally Posted by iSlayer
View Post

In relation to Fermi: I am so disappointed. I do not need a replacement CPU.

Correct me if I'm wrong... but I thought we don't have any performance data on Fermi? Not hitting target clocks is not a huge deal, the performance impact is minimal. The 5870 at 725 vs 850MHz is a small difference in performance to say the least.
 
#26 ·
Quote:


Originally Posted by 003
View Post

Correct me if I'm wrong... but I thought we don't have any performance data on Fermi? Not hitting target clocks is not a huge deal, the performance impact is minimal. The 5870 at 725 vs 850MHz is a small difference in performance to say the least.

Fermi is basically a Quasi-Workstation GPU that can function as a CPU. It's not designed to me a godly and hyper-powered gaming GPU. In short, not what I want.
 
Status
Not open for further replies.
You have insufficient privileges to reply here.
Top