GTX465 (I think it's 465 after all?) is according to so far speculations a castrated GF100 with 11SMs/4 ROP partitions. If true why would something like that have a chip with a different codename?
NVIDIA GF100 & Friends speculation
- Thread starter Arty
- Start date
GTX465 (I think it's 465 after all?) is according to so far speculations a castrated GF100 with 11SMs/4 ROP partitions. If true why would something like that have a chip with a different codename?
Why? 5830 isn't based on 530mm² GPU.
Pretty much imho. Still, I think gtx465 should beat it. This card should scale down gracefully with unit count, unlike the HD5830 which has a rop / usable memory bandwidth problem. Could get close to GTX285 performance as a guess.
That's what I got from it as well. Maybe the close model numbers indicate they are very close in performance, though it doesn't fit the rumours that GF104 is actually faster than this downscaled GF100 (though I said I'm sceptical about this so it would fit my expectations...)
Well if you look for some more recent speculations, they say that GTX460 is after all GF104...
Maybe GTX 460 and GTX 465 are not different names for the same product, but actually different products, the former based on GF104, and the latter based on a salvaged GF100.
After all, isn't it more natural to assume that different names refer to different things? Granted, this is NVIDIA, but they don't usually rename products before launching them.
After all, isn't it more natural to assume that different names refer to different things? Granted, this is NVIDIA, but they don't usually rename products before launching them.
Well, ive thought about that, but i dont belive it would be just a xx5 separating the products.. But this is nVIDIA, you never know
Well, ive thought about that, but i dont belive it would be just a xx5 separating the products.. But this is nVIDIA, you never know
Well, it's happened before: GTX 275, 280, 285...
True but, correct if im wrong, those essentially the same chip, just with a respin/die shrink... a castrated GF100 and GF104 should be two different chips, despite based on the same architecture..
Sure, but there's good reason to believe that performance should be close, and in the end, it matters a lot more than what piece of silicon is under the heat-sink.
My (wild) guess here is that NVIDIA originally planned a GTX 480, (470) and 460 all based on GF100 as well as a GTS 450 based on GF104. Then, they found out that the GTX 460 wouldn't be very much faster than the GTS 450, so they renamed the former to GTX 465 and the latter to GTX 460, in order to be more consistent.
My (wild) guess here is that NVIDIA originally planned a GTX 480, (470) and 460 all based on GF100 as well as a GTS 450 based on GF104. Then, they found out that the GTX 460 wouldn't be very much faster than the GTS 450, so they renamed the former to GTX 465 and the latter to GTX 460, in order to be more consistent.
thechosentwo
Newcomer
Yes it is
If the 530mm² GPU yoou are talking about is a joke, then AMD's offering must be a bigger joke...as their GPU's perform worse.
Or was is purely circulear logic (aka falllacy) you delivered?
Meh, i still dont buy the idea
Besides, reports say that GF104 smokes GTX460... Unless nVIDIA is artificially crippling GF104 to not canibalize GTX460 sales...
In the end, i think that we never had so much fun with GPU speculation as with Fermi
The point in question was that GF100 is a very large and very costly chip to make. Selling it at the same price point as the HD 5830 can't be good for NVIDIA. Of course for end-users, power consumption aside, it doesn't make any difference.
More interestingly, that is another rumour source which claims GTX 460 is now GF104.
With fewer shader cores but a higher clock speed and more TMUs, it'll be interesting to see just how far below (and above?) the 465 that card will score.
Looks quite expensive to me. I've got my doubts GTX460 can compete with HD5850, and I have a really hard time believing GF106 (whatever it will be) is as fast as a HD5770...
Similar threads
