Nvidia GT300 core: Speculation

[trinibwoy]

Well, if you make it wider, than you need even more decoders to able to fill a bigger warp with useful work to do

Who said anything about bigger warps?

 

[nAo]

Who said anything about bigger warps?

Not that simple, there's a reason (actually more than one..) why NVIDIA hw logical SIMD width doesn't match physical SIMD width.

 

[MfA]

You could simply run more warps of the same program ... trade flexibility and branch granularity for less control circuitry. Are there any annotated die graphs to show how much area they could save in this way? It seems to me that even for NVIDIA it's not really an issue.

 

[nAo]

You could simply run more warps of the same program ... trade flexibility and branch granularity for less control circuitry. Are there any annotated die graphs to show how much area they could save in this way? It seems to me that even for NVIDIA it's not really an issue.

Umh, something doesn't compute here. If it didn't pose problems why aren't doing it already instead of 'artificially' increase their SIMD width? For instance don't they need 2 clock cycles to schedule an instruction?

 

[MfA]

They double pump the SIMD array so they can easier deal with latency of the instructions ... but that's a completely orthogonal issue (you'd still be doing that even if you ran multiple warps with the same instruction unit). As for why they aren't doing it already, I personally don't think it makes sense for them to do it ... as I said you lose flexibility (if you don't have enough WARPs to run part of the SIMD lies idle) and the branch granularity increases (branch paths get shared between warps).

I don't think the control circuitry necessary for 16 wide scalar SIMD vs. a wider SIMD is really what is keeping their flops/mm2 down, that's why I asked for the annotated die micrograph.

 

[Sxotty]

Why would I? You stopped addressing the arguments and just resorted to stuff like this ... I'll just sit here and gloat now

It was a joke about misunderstanding what people are saying. A situation that arises quite often in forums.

I still think you meant that physX becoming popular would be bad b/c it would become entrenched in a dominant position and lead to less innovation due to pressuring others out of GPU market. If that is not what you meant feel free to correct my understanding.

 

[Richard]

Saying that something's better is the same as saying that something's worse. So you're essentially saying the same thing.

You need to re-read my post: saying something is worse is different from saying something sucks.

Chalnoth: wrt the wrappers. I tried a couple, unfortunately EF2000 was one of the earliest titles using GLIDE and the game ran in a weird DOS/win32 mode which isn't compatible with today's Windows. You can run the game in DOS or Windows or this weird mode. GLIDE is only supported in this last one.

Btw, I don't believe a wrapper will be really necessary as PhysX is more abandonment-proof (that's a mouthful) since it runs, however slowly, on X86.

 

[trinibwoy]

Not that simple, there's a reason (actually more than one..) why NVIDIA hw logical SIMD width doesn't match physical SIMD width.

I'm not discounting any possibility. Who knows, maybe instruction issue now runs at the shader clock.

 

[Davros]

@Richard EF2000 didnt work with a wrapper because the game talked directly to the card and didnt use a driver

 

[-The_Mask-]

According to CJ the GT300 is faster then HD5870, but not faster then HD5870X2
Link: http://translate.google.nl/translat...ring.tweakers.net/forum/view_message/32639927

 

[loekf]

Since when does the definition of first include a > 6 week lead?

Fuad also claims GT300 taped out and he's referring to a 6 weeks fab time.

My goodness... does he have any clue ?

Usually a development lot takes 6-7 weeks to get through a fab if you pay lots of money. Production lots are more in the order of 12 weeks.

Allow for packaging 2-3 weeks.

If the green team just did their first tapeout, validation, qualification and testing takes time.

 

[trinibwoy]

I don't think Fuad is referring to the tape-out there. He's probably referring to the first lot back from the fab. According to Charlie tape out was a few weeks ago. And because it's Charlie that's probably the worst case scenario

 

[neliz]

Paper launch in November. Retail availability Q1'10.

If you found the Cypress performance underwhelming, prepare for another card that won't make you sell your GTX295.

 

[digitalwanderer]

Are you ready?

 

[Arty]

Paper launch in November. Retail availability Q1'10.

If you found the Cypress performance underwhelming, prepare for another card that won't make you sell your GTX295.

Wowsies, that scenario makes it sound terrible. When was the last time a product was paper launched two months ahead of availability?

As for performance if it is slower than the GTX295 then that would put it about 20% faster than Cypress as a best case scenario.

 

[digitalwanderer]

Wowsies, that scenario makes it sound terrible. When was the last time a product was paper launched two months ahead of availability?

Was the FX paper launched that early even?

 

[neliz]

As for performance if it is slower than the GTX295 then that would put it about 20% faster than Cypress as a best case scenario.

20% sounds about right.

 

[Domell]

According to some latest rumours GT300 performance is about 10-20% better than 2XGT200.

http://translate.google.com/transla...splay.php?fid=4&sl=zh-CN&tl=en&hl=en&ie=UTF-8

So 384-bit in GT300 and 3 times faster SP (~3 Tflops?) than GT200 ones? Sounds good for me because i think it is pretty enough 384-bit with GDDR5.

 

[neliz]

According to some latest rumours GT300 performance is about 10-20% better than 2XGT200.

Did you quote this too?

Some scores seem to refer to the general-purpose computing performance rather than gaming performance

 

[Domell]

That`s right but there is something about "several times better performance than opponents" so there is very possible that it could give at least 2X better gaming performance than GT200.