GTS 640MB 2.0? I tried doing a Google search to find out more about it, but couldn't find anything. What were the differences between that and the original GTS 640MB?
Nvidia GT200b rumours and speculation thread
- Thread starter nicolasb
- Start date
GTS 640MB 2.0? I tried doing a Google search to find out more about it, but couldn't find anything. What were the differences between that and the original GTS 640MB?
It was very short-lived. It had another SIMD enabled, bringing the total SP count to 112 (16 less than full G80, 8800 GTX and Ultra SKUs).
link
According to Computerbase Nvidias Jens Neuschäfer confirmed a new GTX260 card with 216 ALUs.
According to Neuschäfer they won't rename the card because the modifications are not enough to justify a new name.
For some reason this reminds me of the 8800 GTS history:
According to Neuschäfer they won't rename the card because the modifications are not enough to justify a new name.
For some reason this reminds me of the 8800 GTS history:
- 8800 GTS standard (G80 based)
- 8800 GTS 2.0 (G80 based)
- 8800 GTS 512 (G92 based)
another web page about the new gt260
http://translate.google.com/transla...1219284176d9812.html&prev=/search?q=expreview
http://translate.google.com/transla...1219284176d9812.html&prev=/search?q=expreview
DegustatoR
Legend
Well, maybe they should really _read_ reviews then?
Is that an excuse for NV making a bigger mess of their namingscheme than before? Or is this their way of simplifying their product range? Just name everything GTX260 no matter how many SPs or TMUs it's got. That simplifies it alright.
Is that an excuse for NV making a bigger mess of their namingscheme than before? Or is this their way of simplifying their product range? Just name everything GTX260 no matter how many SPs or TMUs it's got. That simplifies it alright.
Sure does. And hopefully the confusion helps shift a few legacy units.
DegustatoR
Legend
They've done exactly the same thing almost every generation. And every generation someone is saying 'oh my god how could they what will the buyers do?' And every generation it turns out that videocard buyers aren't as stupid as someone think they are.
GTX260 'new' will eventually replace GTX260 'old' (or they both will be replaced by something new). The performance difference with another cluster operational won't be anything to shout about anyway. The number of SPs operational will be written in the prices and on the boxes and on the websites.
I really fail to see a huge problem here.
Sure it's ugly. But introducing GTX270 which essentially has the same performance as GTX260 is even uglier.
Shadowmage
Newcomer
The point of the extra TPC is to beat ATI's 4870, which NV's current GTX260 can not accomplish. Of course the performance difference is something to "shout at", or else NV wouldn't bother doing so: after all, disabling 2 TPCs will increase yields compared to only disabling one.
Kaldskryke
Newcomer
I'm hoping someone here can explain to me why introducing a new model name is a big issue, or perhaps why re-using the GTX260 moniker is worth confusing customers over.
Can one feasibly have a new GTX260 and an old GTX260 in SLI? If not, the name re-use seems even worse.
As far as I can tell, this new card is directly between the current GTX260 and GTX280 in terms of processing units, and I can't see why they wouldn't go with GTX270. There has to be a real reason, right? Is it because they're saving the number for a 55nm variant?
DegustatoR
Legend
GTX260 beats HD4870 512 MB as it is.
The point here is presumably that they don't have enough parts with 2 TPCs broken now and that there are no reasons for them to use such parts for any retail card.
Otherwise they wouldn't do such an 'upgrade' because GTX260 is doing fine even with 2 TPCs disabled.
DegustatoR
Legend
Because the number of processing units isn't the only measure of performance.
If it'd have frequencies in-between GTX260 and GTX280 and an increase in memory bandwidth then yes, it would be feasible to use GTX270 name.
But the clocks remain unchanged and that means that it'll be quite a bit closer to GTX260 then to GTX280. Using GTX270 name for such card is even more misleading to me than naming it GTX260 too.
It does what now?
Last time I checked, they were trading blows almost equally, but in general HD4870 taking small edge over GTX260 (this is based on several reviews, not one, of course)
It all depends on which reviews you (wanna) read. Or better said, which benchmarks you (wanna) see... and which color glasses you're wearing... NV green or AMD green...
Fact of the matter is that in my homecountry the cheapest HD4870 costs around €190 while the cheapest GTX260 costs €236. Tie that price to the performance and that makes the HD4870 the clear winner in my book... at least in Holland it does.
DegustatoR
Legend
Last time I checked they were mostly comparable with GTX260 leading in newer games in high resolutions due to insufficient VRAM on 4870.
You can always pick some benchmarks where 4870 would be faster of course.
The point is -- GTX260 is doing OK as it is, no need to upgrade it if it'll hurt your yields. So i take it as if this upgrade won't hurt their yields much.
GTX 290 X2: 216 SPs, 72 TMUs, 24 ROPs, 384-bit 2.0GHz GDDR5, 648MHz Core, 1620MHz Shader, $599. vs 4870X2.
GTX 290: 240 SPs, 80 TMUs, 24 ROPs, 384-bit 2.3GHz GDDR5, 702MHz Core, 1755MHz Shader, $399. vs 4850X2.
GTX 270: 216 SPs, 72 TMUs, 20 ROPs, 320-bit 2.0GHz GDDR5, 648MHz Core, 1620MHz Shader, $299. vs 4870.
GTX 250: 168 SPs, 56 TMUs, 16 ROPs, 256-bit 2.0GHz GDDR3, 594MHz Core, 1485MHz Shader, $199. vs 4850.
IMHO.
I don't fundamentally disagree, and regarding the *range* of the clock speeds your guess is as good as mine. The reason why I speculated that they'd use the top bin clock-wise for the X2 while not activating all the clusters is that they risk being bottlenecked by other 'fixed' parts of the chip otherwise, especially at 1920x1200 where a lot of people still benchmark SLI/CrossFire-like configs despite the fact nobody sane (except avid Crysis players) would buy this without a 30" monitor.[...]
IMHO.
At the same time, they need to make sure to take noticeably less than 150W per card all the time (also allowing for some overclocking headroom for both end-users and AIB SKUs). I think they can do that at the top bin with 9C, but I could be horribly wrong.
Shadowmage
Newcomer
I'm echoing the sentiments of the other people posting in this thread: 4870 in the worst case, most pro-NV reviews, ties the GTX 260 overall, and in the best case, beats the GTX 260 by around 5%.
Here's the Anandtech review, for example: http://www.anandtech.com/video/showdoc.aspx?i=3341&p=13
Also, NV needs to successfully differentiate their GTX 260 and 280 products. A 9 TPC vs a 10 TPC product might have too small of a performance differential for users to justify spending an extra $150 for the GTX 280.
The only reason NV is doing this is to beat ATI in performance at all price segments; else, there is no reason for NV to do this. It's obvious that there will be a slight reduction in yields, but it's currently their easiest way to close the performance gap against the 4870.
