View Single Post
Old 01-24-10, 09:23 PM   #86
Registered User
Join Date: Jul 2003
Posts: 1,526
Default Re: 5870/5970 vs. GF100 (Fermi)

Originally Posted by Slytat View Post
What double standard?

First of all, when I do comparisons, I compare on a GPU vs GPU basis.

1 Fermi is likely going to be slower than the 5970 but the 5970 is essentially 2 cards so, unlike yourself, I compare them GPU vs GPU and since the 5970 has 2, I'm going to compare 2 x Fermi or the GTX395 (which will most likely have some kind of gimping).

If you are going to compare the Fermi to an ATI card, it should be the 5870. This belief that Fermi has to beat the 5970 in order to be a success is ridiculous and it certainly wasn't applied to ATI when they released the 5xxx series as even their 2 GPU card doesn't beat SLI by that much and their 5870 can't even best the aging 295 let alone 285s in SLI. I think it is you who is indulging in a double standard here mate, not me.

And the main point that ATI intentionally keeps the relative size/power and production cost of their last few generations of GPU's,limited on purpose to allow the use of the same GPU on several different models of cards has nothing to do with it right?,which includes cards starting as low as 250$ using Cypress GPU's(HD5830).

Absolutely nothing stops ATI from building GPU's just as large and using just as many transistors as Nvidia does for their high end parts,and sticking to a single GPU design for their highest end models,just like Nvidia does,but it seems that a least in the last 2 generations,they've found a better way to serve several price points and address several markets with the same GPU.

And the latest example is the most obvious one....Like it or hate it,ATI's HD5800/5900 cards were released on time and kicking ass and taking the speed crown,while Nvidia does nothing but talk how great Fermi will be,and it's horribly late and using a lot of power,so it's not hard to see which strategy is working better so far.

So with that in mind,comparing a single Cypress,which was still built to be economical to produce and use reasonable amounts of power and easily allow the possibility of a dual GPU card,against a strategy from it's direct competitor which tried to stuff as many transistors as possible into a single GPU die,uses a lot more power and isn't cheap to make,isn't really a fair comparison either.

Ask yourself this....When will you see Fermi being used in a 250$ video card,just like ATI can use Cypress in a card selling for the same price,and still make money on it?....That's right,probably not happening anytime soon.
shadow001 is offline   Reply With Quote