NVIDIA GF100 & Friends speculation
- Thread starter Arty
- Start date
onethreehill
Newcomer
GTX 460 Tested
Core Clock 900MHz
Shader Clock 1800MHz
Effective Memory Clock 4400MHz
http://www.coolaler.com/showthread.php?t=242673
Core Clock 900MHz
Shader Clock 1800MHz
Effective Memory Clock 4400MHz
http://www.coolaler.com/showthread.php?t=242673
Which essentially means that they have killed off the market for their top end chips unless of course they can come out with a substantially faster revision for those too. In any case the important comparison may indeed be with the HD 6850 considering ATI's near clockwork release schedule thus far.
Silent_Buddha
Legend
Which doesn't matter to Nvidia. They've already probably sold most of their original batch of Fermi chips.
Nvidia could care less if their IHVs are left holding unsellable inventory. They already have the money in pocket.
And I'm sure Nvidia would prefer to sell a less costly chip with higher margins in the consumer space. Then push Fermi chips into the Quadro and Tesla lines where the margins are astronomical.
They have benchmarks with GTX 480 higher than 5870, so its job is done. Retire it and move on is probably what they are thinking, and keep just enough in the channel to make sure they can continue to lay claim to fastest single GPU.
Regards,
SB
That isn't reasonable. In order to survive, nVidia needs to cultivate and maintain its relationship with its relatively small number of customers. Piss them off, and they could easily find themselves with a fraction of the sales the next go-around. It is most definitely in nVidia's best interest not to leave board vendors holding unsellable product.Nvidia could care less if their IHVs are left holding unsellable inventory. They already have the money in pocket.
It's not reasonable but it is reality. What you think happens with GTX465 sales now?
I'm not sure that we have the information available to make firm conclusions on that. We'll see. But in any case my point was slightly different, that nVidia most certainly does care that their small number of customers aren't left with unsellable product. Either that or they're stupid, and their business success to date indicates that isn't the case.
Exactly. I wonder about performance of a full blown GF104 with high clocks. Would it be enough to beat GF100? I guess it could come close to 5870, at least.
1751,25 × 384 = 1401 × 480.
So you'd need a full GF104 with shaders running around 1750MHz to be roughly equivalent to GF100 (geometry, cache, and bandwidth aside).
I doubt NVIDIA could make many of those, let alone within 300W; especially when they don't seem to be able to (or optimistically, to be willing to) make a 384-SP variant.
Well GF104 would have also 4 more TMUs
Non-384 parts are reaching already 900Mhz with a little voltage tweak.
And we still dont know how much power an overclocked GTX460 needs. In FulMark it doesnt exceed 81ºC, to it might be good.
GF100 is more limitedy by texturing performance, than by its FLOPs. GF104 has similar per-clock texturing performance, but it allows higher clock-speed...
It's possible to say, that GF104 has much higher effectivity in terms of per-flop gaming performance, than GF100 (not only because its ALU:TEX ratio).
It's possible to say, that GF104 has much higher effectivity in terms of per-flop gaming performance, than GF100 (not only because its ALU:TEX ratio).
Which doesn't matter to Nvidia. They've already probably sold most of their original batch of Fermi chips.
Nvidia could care less if their IHVs are left holding unsellable inventory. They already have the money in pocket.
And I'm sure Nvidia would prefer to sell a less costly chip with higher margins in the consumer space. Then push Fermi chips into the Quadro and Tesla lines where the margins are astronomical.
Well im sure a certain individual might be jumping up and down screaming 'I told you so'. But I can't fault that as I would probably do the same in his shoes.
Well I guess the cards will sell eventually given enough time. Im sure there are enough people who will buy them because they have higher numbers than the cards which seem to be replacing them. Also I guess since the workstation line of cards already seem to be salvage models it won't really effect them too badly in the medium term.
Well, that may not last long if Dave is being Dave and has already got his refresh models coming into the pipeline soon. Failing that, theres always the dreaded price cut and the HD 58xx models have been out long enough for non recoverable costs to have been paid back. They also seem to have been building up quite a following in the absence of any real Nvidia competition. I suspect the shoe might be on the other foot here in terms of momentum and for a change ATI may be in the drivers seat for once.
Silent_Buddha
Legend
I wouldn't count on that, ATI faced worse competition when R300 was around and Nvidia didn't lose many fans from that. If they can get back on the track, I'm sure they'll be fine.
I dread the day Nvidia or ATI gets into enough trouble that one of them can no longer compete. That would be a bad day for 3D hardware no matter which side of the fence you sit on.
Regards,
SB
For many people the alternatives were ATI or wait. There was/is a massive upgrade process going on for many enthusiast gamers who may have skipped Vista or want to skip away from Vista. So the momentum is still on ATIs side because of it, and the Steam survey reflects that. Even now Cypress is still outselling the Fermi based cards in-spite of the typical 2:1 sales disparity. Thats what im talking about when I say that ATI has the better momentum at the moment.
Im sure there'll be three players in the market before that happens. Intel, AMD and Nvidia so the loss won't be quite as bad as you'd think. Its Karma really. When Nvidia bested 3DFX there looked to be only one big player, but then ATI came along and...
Similar threads
- Deleted member 2197
- Architecture and Products
