PDA

View Full Version : ATi GPGPU's using R580?



turtle
08-31-2006, 02:54 PM
Pay close attention to the X server log output from ATI's Linux driver, as we did recently with a FireGL V7350, and you'll see mention of support for what we believe is a brand new class of product featuring a modern graphics processor. Listed in recent drivers is a SKU called FireStream 2U, apparently powered by ATI R580 and given away by the following strings: FireStream 2U (R580 724E), FireStream 2U (R580 724F).

Using R580 and with unique device IDs of 724E and 724F, at least as far as the Linux driver is concerned and confirmed on ATI's vendor ID page (http://www.ati.com/developer/vendorid.html), these FireStream products appear to be distinct boards with R580, but with a different purpose than 3D graphics acceleration.

First consider the ATI SIGGRAPH '06 presentation on what they call a Data Parallel Virtual Machine (http://www.ati.com/developer/siggraph06/dpvm_sketch_siggraph.pdf), using a recently released 'close to the metal' API, built specifically for general purpose GPU computing (GPGPU). Then consider that ATI have spent significant resources on GPGPU since the inception of the R5-series of GPUs, lastly bearing in mind that stream processing is a class of problem that exploits massive parellism of a device, and it seems logical that FireStream is a product using a 3D GPU but is not for the explicit purpose of drawing pixels.

We therefore surmise that FireStream products from ATI will come in significantly different configurations to those you're used to seeing for desktop discrete graphics, possibly without any display outputs at all, and almost certainly with different PCI Express slot configurations for use in platforms providing 1x, 2x, 4x and 8x slots -- in addition to PEG16X of course. We invisage stream processing clusters with multiple FireStream boards per system, possibly with multiple GPUs per board, designed to exploit their DPVM efforts and accelerate parallel processing problems that fit a modern GPU, as well as their recent Havok FX relationship for the mainstream. Also, thinking about spy shots of upcoming Radeon desktop boards with NVIDIA-like inter-board connectors, we see no reason why future FireStream/DVPM can't also exploit those 'Crossfire' links for concurrent high-bandwidth connection of multiple FireStream boards, for further performance increases and inter-GPU communication outside of the PCI Express bus.

Whatever the final configuration of the product and how it operates, it seems clear to us that FireStream is a coming reality and the first product of its kind from either of the main desktop GPU IHVs to address a growing and very significant market, and from ATI's perspective exploiting as much as possible the architectural traits they've built into R5-series hardware from day one. More on FireStream as we get it.



http://beyond3d.com/

cky2k6
08-31-2006, 03:00 PM
i wonder why amd bought ati again...

turtle
08-31-2006, 03:06 PM
Yeah, I was thinking the same thing. It would be interesting if this was for Torrenza...

Lightman
08-31-2006, 03:33 PM
I wonder what R600 will can in GPGPU area!
New era is comming, CPU as part of bigger system with close connected GPGPU as DSP and at the same time normal GPU.
Like in old good times of Commodore AMIGA with 5 (fully programmable) co-processors in one system ;)

n-sanity
08-31-2006, 03:47 PM
Back to the basics it is.

turtle
08-31-2006, 03:48 PM
Like in old good times of Commodore AMIGA with 5 (fully programmable) co-processors in one system ;)

Before math co-processors were intergrated onto the die of the cpu... ;)

The more things change, the more history repeats itsself.

Lightman
08-31-2006, 04:00 PM
Before math co-processors were intergrated onto the die of the cpu... ;)

The more things change, the more history repeats itsself.


That's right!
Two less with MC68040/060 because of integration, but still there was blitter, copper and sound chip :D

Now I want to use GPU for rendering, folding an physics (later I will want more :cool: )

turtle
08-31-2006, 04:22 PM
lol...I recall an old-timer discussing his beloved Agnus, Denise, and Paula with me. He also told me Copper was quite exciting for it's time.

I have to admit, i've been slightly jaded having been introduced to computers in the i486dx era. :p:

nn_step
08-31-2006, 05:20 PM
this has a bit to do with AMD and what AMD wants really soon...

turtle
09-01-2006, 01:18 AM
Mike Houston (a pretty interesting fellow whom seems to beta test API's amongst other things) shared a couple of interesting tidbits at B3D/GPGPU on why the API is geared towards DX9...At least right now (or I think that's what he was trying to get across):


"DX10 still doesn't have all the features, like scatter, but should reduce much of the overhead seen in DX9. DX10 also doesn't expose all the raw memory formats that can be used to effectively change caching behavior"



In case you were wondering what the unknown product was:




...FireSTREAM 2U "Enterprise Stream Processor." It didn't say anywhere that I can't talk about it, so I'll share the insanity. 1 gpu (48 shader processors), 1GB ram, 512bit memory bus. Frequencies unstated, but since it's 2U probably not that hight (you can't even have a decent-sized fan, nevermind a sizable heatsink). Another question is whether the 512bit bus is 8x64bit (current bus is 8x32bit) or 16x32bit. Most likely it's the former. However, having more channels would have allowed more random reads to be processed. With 128bit reads, however, it's probably not too important.

man, ati is pushing ahead. if i was looking for long-term investments, I'd buy their parent company's stock. AMD's chips are gonna have a huge advantage incorporating this technology, and Intel will be playing major catch-up.

Pretty interesting stuff.

[cTx]Philosophy
09-01-2006, 01:52 AM
Interesting read..
Im guessing this is phase one of amds approach of their all in one design, just testing it out in a pci slot?

perkam
09-01-2006, 04:24 AM
Mike Houston (a pretty interesting fellow whom seems to beta test API's amongst other things) shared a couple of interesting tidbits at B3D/GPGPU on why the API is geared towards DX9...At least right now (or I think that's what he was trying to get across):





In case you were wondering what the unknown product was:



Pretty interesting stuff.Sounds like X1950GX2 to me.

Perkam