NVIDIA GT200 Rumours & Speculation Thread

Status
Not open for further replies.
Okay, did I just see "accelerated encoding"?

Now is that some snarky, subtle answer to the paid CUDA accelerated encoder, or just something else :D

Hopefully this is done on the ALUs instead of dedicated hardware where we just went through the same format/capability concerns with GT200.

Maybe it means the AVIVO Transcoder might actually start using the GPUs and not just be really fast software solution using CPU?
 
G92b won't support GDDR4/5. Next? :p (would be nice if the 9900GTX-or-whatever-it's-called used 1200MHz GDDR3 and exposed it fully though... OC versions could always use 1300MHz chips)
 
When did being faster, more power efficient, better at video, more feature rich and using more advanced processes, supporting a higher revision of DirectX and using new memory technology count as desperation?

I can't even see how you've remotely got a case there.
IMHO
1. performance leadership - that I don't know, but they claim the new AMD chip is faster than the old Nv chip. NV will wait and give'em price/performance easily? "best price at every price point, except the top-level" , thats sad :(
Still this point may be valid, but after so many fiascos AMD have to prove before they get credit.

2. technology leadership
55nm vs 65nm - so what? NV30 anyone? R600?
DX10.1 - so what? Who uses this? Assasins creed says it all, no devrel, no Dx10.1.
And if GT2x0 is 10.0 part, well, thats it, 10.1 goes the way of 8.1 (even madonion supported 8.1, so this time its even worse, it will go dodo)
GDDR5 - so what? if your card is not faster, then type-of-memory is nothing. Besides, which ram is cheaper ddr3 or ddr5 ? price/performance?

3. Best perf/watt - really? I've seen many claims but how many independant tests confirmed these?
Heck, a week ago Anandtech made comparison of IGPs - and guess what RS780 was measured to use more power than NV one. After many many claims that PowerPlay is better.

4. Best HVD - same as p.3 . Tests I've seen show NV on-par trailing with margin <10%, who says G92b won't make it closer? If you can't see it with bare untrained eye its not existing. (valid for fps too)

5. Maybe the only point I have nothing against, but also I don't care for any of these features.

In short, after years of being pro-ati and pro-amd I'm tired of waiting and giving credit. Until they deliver cards and these have better price/performance (you see, I don't even want'em to have the fastest card...), all their slides are nothing. Arguments like dx10.1, better HD etc are void.
/IMHO
 
Last edited by a moderator:
IMHO, the "RV770 advantages vs G92" listed in the slide look like desparation

Considering G92 first came out in, what, October 2007, and RV770 will come out in mid-June 2008 it looks sad for ATI to be boasting about the superiority of their products over the 8 month old G92. ATI's answer to G92 was RV670 (actually G92 was more like NVidia's response to RV670, but NVidia released it earlier), because HD3870 and 8800GT had similar price points upon their release.

HD3870 was a very nice piece and I'm thinking it did much better than R600's, but still not good enough because well, 8800GT was faster and AMD couldn't still even match the ancient 90nm 8800GTX.

Obviously we can only speculate about RV770's performance but whatever it is, actually it does look like that the RV770's competitor is the G92, looking at the price points. 4850 is $220 and it looks like it'll do better than a G92, which is as cheap as $160. Only, 4870 is sort of an unique card because when it comes out it'll be the only card to be priced at that segment ($350), and if it does perform accordingly it'll be the only choice for people with $350 to spend on a graphics card (which are more numerous than $400+ people)
 
Considering G92 first came out in, what, October 2007, and RV770 will come out in mid-June 2008 it looks sad for ATI to be boasting about the superiority of their products over the 8 month old G92. ATI's answer to G92 was RV670 (actually G92 was more like NVidia's response to RV670, but NVidia released it earlier), because HD3870 and 8800GT had similar price points upon their release.

But what do you suppose to do? What would you do?
 
What ATI is doing (4870 and 4850) is, or at least looks good. My complain is in that slide they talk about being superior to a 8-month GPU, which will have been the previous generation of NVidia GPUs by the time 4870 and 4850 are released. Performance leadership at every price point? What about $450 or $600?

Maybe, as I wrote, it makes sense to compare 4850 to G92's but generally comparing the upcoming RV7xx to one generation old NVidia GPU's doesn't make a lot of sense to me.
 
They couldn't; but that doesn't remove the fact that comparing your up and coming products with 8-month old GPUs is going to make you look like you're desparate.
 
What else should they do then? Not compare it to the competition at all? Now that would look even worse and you'd have all fanboys all over it.... The most recent GPU they can compare it to is G92. Doesn't matter if it's 8 months old or 2 months old.
 
I am sure nVidia does the same... any slides they have shown probable compare themselves to AMD's current products the old HD3800 series, but I would not consider nVidia to be desperate.

It's funny as nVidia has the mindshare of people (excellent PR), they are showing superiority if they compare to older products and AMD is desperate when they compare to older products. Both can only show what is announced. Both can change shipped clocks untli the product is announced.
 
They couldn't; but that doesn't remove the fact that comparing your up and coming products with 8-month old GPUs is going to make you look like you're desparate.
Lets go but a few short days back shall we? What are they comparing against? An 8 month old competetor? Would that be an 8 month old competitor thats not even close to being in the same price bracket? Desparate?
 
That's just a benchmark trying to show how fast their new card is, not only relative to HD3800 but also the PS3 and a CPU as well. And they give exact numbers, which is much more different from preparing a slide which roughly says "we're better than a year old products." And it doesn't have a marketing-hype style approach, it's only numbers. When you try to show off your new GPU's speed (how "faster" it is, not just faster) you can only do that the way NVidia did, because yeah, AMD hasn't released its new GPUs either.

Saying "we're three times faster than an 8-month old GPU" doesn't look that desperate to me. However, "we're better than our rival's 8-month old GPU's at every price point, and we have this this this (marketing hype, although I do believe that some of the features have value), all compared to a 8-month old GPU" looks like you're giving everything you can to make people believe that you're better than G92.

I have nothing against AMD and nothing for NVidia. I've grown past fanboyism during 3dfx-NVidia wars.

AMD did a fair job with 3870, and it looks like its new line of products are going to be nice, and sell good as well (especially if you consider the 4870 will be the only product in its price range). It's just that that slide LOOKS somewhat desparate. I know they aren't.
 
Lets go but a few short days back shall we? What are they comparing against? An 8 month old competetor? Would that be an 8 month old competitor thats not even close to being in the same price bracket? Desparate?

What would you have them do ? AMD/ATI only goes that far for price/performance by their own choosing, and since there will be no GT2xx mainstream version for some time (i seriously doubt they'll be out before the end of the year)...
 
Obviously there is nothing wrong with comparing GT200 against RV670, as long as you take into consideration that they are aimed at differing market segments.

Just as obviously, there is nothing wrong with comparing RV770 against G92 in mind of the fact that, initially, at least, they are likely to be competing products.

Both are perfectly valid comparisons as long as you take into account the various caveats involved and, to be honest, I can't see any reason why either shouldn't stand!
 
GT200 is very good


Last year, completed the design and silicon, dieshot it is a great cross

Second-generation US framework, improve control of the thread, the new tex unit, strong performance

10TPC, 30TPA, die size large, 512 bit Corssbar MC in particular for local, ROP will not only increase the number of the design is also new

Double-precision floating-point hardware support, can do many things outside of graphics, CUDA plans to expand in

Comparison of regret that the process is subject to restrictions, Flops ultimately failed to break 1 T, and the initial yield is not high

Properties not said that the Internet may Chuide a bit excessive, but I can guarantee that even the most discerning enthusiast, to see the results of this card also will scream

20080528aa723666bbd65a9kl6.jpg

fmnt9.jpg
http://we.pcinlife.com/thread-940503-1-1.html#zoom

Even on this small picture, the ten clusters can be seen. Lets hope the bigger version will be soon online...
 
Last edited by a moderator:
Status
Not open for further replies.
Back
Top