Is GDDR based on nVidia-patented technology? ( -> GDDR2 for the NV30 )

[Uttar]

Hello everyone,

I fear I don't have many facts to back what i'm going to say. But oh well... Only time will tell if i'm truly right.

Anand recently said the following about GDDR:
Then there's ATI's new graphics memory standard - GDDR3. Despite the name, GDDR3 is fundamentally a DDR-II based architecture (e.g. the technology still allows for four transfers per clock). The difference is that GDDR3 implements a different clocking scheme and different I/O.

Since this article isn't focused on the technology behind GDDR3 we won't go into great details explaining the differences but GDDR3 uses unidirectional strobes and open drain 1.8V for I/O (as opposed to differential clocking and SSTL for DDR-II); to put it simply, GDDR3 changes the way data gets in and out of DDR-II devices. The benefits of these changes are mostly in simplifying graphics card design while optimizing for the high frequencies that will be needed by the end of next year.


Note that the bold is my personal addition.
Now, the big question is... what's the different I/O?

I think i can answer that for you.
http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=/netahtml/PTO/search-bool.html&r=1&f=G&l=50&co1=AND&d=PG01&s1=nVidia.AS.&OS=AN/nVidia&RS=AN/nVidia

This is a nVidia patent for a optimized-for-graphics DMA system. In other words, that's exactly what GDDR2 is about.


However, while ATI expects GDDR-3 to be widely available in 2004, i expect NV30 secret memory optimization ALL come from that patent and GDDR-2.

And that makes perfect sense compared to the rumor saying the NV30 will have 500Mhz Samsung DDR2.
Because Samsung is the company nVidia designed GGDR-2 with. And nVidia recently said 500Mhz ( the maximum for GDDR-2 ) in 2002 is exciting.

Strange, eh?
I guess we'll see if i was right at Comdex :)


Uttar

[Chalnoth]

DMA deals with transferring information from system memory without CPU intervention. I don't think it has anything to do with the video chip<->video memory interface.

[StealthHawk]

Originally posted by Uttar
Anand recently said the following about GDDR:
[i]GDDR3 is fundamentally a DDR-II based architecture (e.g. the technology still allows for four transfers per clock). The difference is that GDDR3 implements a different clocking scheme and different I/O.

uh....since when does ddrII do 4 transfers per clock?

[Uttar]

Hmm, after rereading the document, Chalnoth is right. Could it be that DMA, with R300/NV30 great power, became a bottleneck? And that this might give a very slight performance advantage if the R300/NV30 would be exactly the same without that?
Or am i wrong again?

And, err, it's a good question about what the heck Anand meant by that. He probably just made a big mistake...


Uttar

[Chalnoth]

DMA's been around for ages, at least as long as PCI, probably a fair bit longer.

With graphics cards, the biggest reason for DMA is for texture access when AGP memory is not in use. Everything else the CPU pretty much sends anyway (Well, perhaps it's also beneficial for display lists that aren't stored in video memory...).

[Uttar]

Originally posted by Chalnoth
DMA's been around for ages, at least as long as PCI, probably a fair bit longer.

With graphics cards, the biggest reason for DMA is for texture access when AGP memory is not in use. Everything else the CPU pretty much sends anyway (Well, perhaps it's also beneficial for display lists that aren't stored in video memory...).

Actually, i think this patent is about an OTPIMIZED DMA system.

Remember that this is for input/output AFAIK. So... Could this be a system nVidia is developing for companies like Pixar due to a complain from some people that retrieving info from the GPU to system memory takes WAY too long?
I think that was an article at thetechreport, but i'm not sure...


Uttar