I really don't want to buy a 256-bit "high-end" card only to replace it 6 months later with something significantly faster in RV770 or G100.
NVIDIA GT200 Rumours & Speculation Thread
- Thread starter Arun
- Start date
- Status
Uhmm I wouldn't be one bit suprised if RV770 would be also a "256bit high end" GPU LOL. Buswidth doesn't define the total amount of bandwidth alone and bus- as well as bandwidth don't define final performance either of a GPU. More important is if the overall balance of each "pipeline" in a GPU is perfectly balanced and it gets sufficient bandwidth for its resource handling.
Me either. So I won't bite on the carrot of promised performance over my GTXs. :smile:
I knew I was going to get caught up on that. Should've said "only 256-bit with less than GDDR5".
Unknown Soldier
Veteran
Hmm such a high CPU score with C2Q clocked only @2,67Ghz? Does anybody think this score could be true?
While I don't think the score is real, what makes you think the CPU is a Q6600? Why can't it be the 2.66Ghz from the Quad-Core Intel Xeon Processor 5400 Series?
http://compare.intel.com/pcc/showchart.aspx?mmID=33081&familyID=5&culture=en-US
US
Last edited by a moderator:
First, nobody mentioned a Q6600, if anything that CPU is either a Q6700/QX6700/Q9450/Xeon 5430. Neither of them produces the supposed CPU score at that clock-rate. You could argue that in a Skulltrail mobo with dual-Xeons you could get that, but then the score would be too low for dual 2.66 GHz CPUs. Second, why on earth are we still considering that thingie...it's so fake it makes porn-star orgasms seem like the real deal
Second, why on earth are we still considering that thingie...it's so fake it makes porn-star orgasms seem like the real deal
I knew there was something fishy about Peter North and Ron Jeremy! Bastards have been using yogurt all along...
G92(A) vs GT200 ~ 7900GTX vs 8800 GTX...
But the competition does not sleep and R700XT is also supposed to be a more than significant upgrade over RV670XT.
Well, if it`s true it`ll be great
But is it only your personal presumption or dou you have some info from reliable source? As i suppose there are not working samples of GT200 at present because of release date about Q4/08.PS. I think 9-series name should be used for GT200 not G92.
DegustatoR
Legend
Doubt that. There is no room in 9x00 moniker for all the G1xx chips currently in development -)
It's just that it SHOULD, but then again, all the other GF9x00 cards (namely 9600) should be named 8x00 for that to work
No rumour suggests 2X number of SPUs...all I've seen points to 96/480(depends on how you like to count them), which means a 50% increase over there. I've also seen rumours suggesting that they'll up the number of TMUs but not the number of ROPs, which may or may not be a good idea.
Considering how good the G92 based products perform, 16 ROPs seems to be sufficient for now, especially if they're going to be running at over 1GHz like the RV770 rumors suggest
I have thought about rumours from Chiphell Forum
Here you are
http://bbs.chiphell.com/redirect.php?tid=17279&goto=lastpost#lastpost
It seems ATI has made very powerful GPU and NVIDIA didn`t expect it and at all i think. It seems very strange. Rv770 which is supposed to be at least 50-60% faster than Rv670 against NVIDIAs G92 which isn`t faster than old GF8 series.
ATI has a tiny flagship GPU (192mm^2) that is based on 1 year old tech and you think Nvidia would be surprised by something that's 50% faster? I'd hope that they have more faith in their competitor than that.....
I`ve said AT LEAST 50% faster. Rumoured Rv770 die size is about 250mm2 and it`s even smaller than G92 but as you see it has 2x more Shaders and TMUs than Rv670! ATI has done a great job. Moreover NVIDIAs GPUs are still in 65nm and GT200 will probably use this process too. Another important thing is Rv770 will be launched in May so NVIDIA won`t have real competitor if GF9800GTX is only OCed GF8800GTS 512 and GF9800GT i only OCed GF8800GT.
#1 There is a reason for the R700 thread, please post there and don't try to start something in the GT200 thread.
#2 As much as I like the rumored specs, let's not forget that these are still only rumors. We have heard a ridiculous amount of rumors, some of them complete opposites of each other, for both the R700/RV770 and the G100/GT200.
I recall a public statement from a NV employee that was quite clear that they are expecting AMD to be on an uptake. If memory serves well that appeared before the RV670 was released. I wouldn't say that NV is "surprised" in that sense, but I'm willing to believe that they're forced to push quite a few things at a much more aggressive pace in order to keep their position.
Recent rumours suggest a shrink for G92 at 55nm. If that one hasn't been in older roadmaps it's already one sign that could point to the above assumptions.
Since we are in the "GF-Next" thread, the real question is what changes NV has made to the architecture for the next "generation". Personally I'd expect G94 to have some elements that might point in the direction they went, yet those could be a rather small portion of any possible changes.
The rumours seem to indicate the figure 480/96 (just like what morgorth said depending on how you count them) which isnt twice the ALU power of an RV670. More like a 50% increase in terms of units. If it was twice, it would be 100% increase. But then theres the current rumours mostly from Fudzilla about RV770 being clocked at 1GHz or so.
I guess G80/G92 (not sure about G94), are all limited by its setup more than anything else. Then maybe bandwidth and so on. I would be disappointed if this GT200 is nothing more than G92 @ 55nm. Unless theres more to it than just it being a die shrink.
I guess G80/G92 (not sure about G94), are all limited by its setup more than anything else. Then maybe bandwidth and so on. I would be disappointed if this GT200 is nothing more than G92 @ 55nm. Unless theres more to it than just it being a die shrink.
- Status
