X1950xtx

atticus1398 said:
oh yeah, will ATI be venturing into the Physics Card realm? Feel free to pm any juicy AMD etc. info to me.:)
ATi and Nvidia have already experimented with using regular off the shelf GPU's as stand-alone physicis cores when using two of the cards in SLI/Crossfire with limited success. Initialy the ageia card was much better at stand alone physics caculations, but give Nvidia/Ati some time and they'll be able to match Ageia..

Eventualy, imo the physics card will be incorporated into your average GPU or at least onto the stand-alone GPU PCB board, much like the same thing that happened to DVD-decoders..

First they were on PCI cards that you had to buy seperate. Then they got incorporated into the GPU itself.

Forgot about this one X1800XL AIW.
Was in the closet.
wtf do you have hobby of just stockpiling $400+ video cards? (peep)
 
Last edited:
atticus1398 said:
whats the deal? I know you get "reference samples" but why let them pile up dust? Do they eventually ask for them back? Or are you going to one day make the worlds first 30 card crossfire system?

I just forgot about this one.
 
sephiroth said:
ATi and Nvidia have already experimented with using regular off the shelf GPU's as stand-alone physicis cores when using two of the cards in SLI/Crossfire with limited success. Initialy the ageia card was much better at stand alone physics caculations, but give Nvidia/Ati some time and they'll be able to match Ageia..

Eventualy, imo the physics card will be incorporated into your average GPU or at least onto the stand-alone GPU PCB board, much like the same thing that happened to DVD-decoders..

First they were on PCI cards that you had to buy seperate. Then they got incorporated into the GPU itself.


wtf do you have hobby of just stockpiling $400+ video cards? (peep)
Yeah, Ageia will be history. Their cards are too expensive and the benefit (right now anyway) is too small to justify the expense. Also even if you have a card it can severely slow the game while the extra bodies ar being rendered just look at GRAW.
I think that ATI's version with two cards in crossfire and a third card for physics will be a nice setup. What's cool is you can run two high end cards for graphics and a lower end part you may allready own like an X1600 and instead of throwing it away you can use it for physics.
You can allready run a Folding at Home client that uses the X1900 to run calculations.
 
yes, but most boards only have enough space for two GPUs and one pci. Some still run sound cards. They will need to make a bigger board with at least 1 more pci slot, IMO.
 
atticus1398 said:
yes, but most boards only have enough space for two GPUs and one pci. Some still run sound cards. They will need to make a bigger board with at least 1 more pci slot, IMO.
They are coming, oh yes they are.
Check this out. 45nm! Will hit 4GHz easily. Short pipe high clock oh yeah. Looks like Intel changed their minds on the 1333 Kentsfield FSB.
Oh well just clock that b**** up :)
Wolfdale.png
 
Last edited:
Hmmm supposedly the R600 has a 512 bit external bus.


"If the 512 memory ring turns to be the real thing, we are talking about 128 GB/s of memory bandwidth with GDDR4 clocked at 2000MHz. We also learned that the R600 may use memory faster than 2000MHz as it will be available by Q1. If ATI keeps pushing the chip we might get even faster GDDR4 chips at production time.
Even the PCB of the R600 will be super complicated, as you need a lot of wires to make 512 bit memory to work. Overall it has the potential to beat Nvidia's G80, but yet again it will come at least three months after Nvidia. The G80's memory works at 384 bit as Nvidia pretty much dis-unified everything in G80 from shaders to memory controllers. Nvidia likes to make rules and probably could not get more than 384 bit wide controller in the chip, as the G80 is still a 90 nanometre chip.
Its a shame that we will need to wait at least until February to see it in action. "
 
Get your 3rd video card ready for physics.
"Windows Vista will support Physics solutions of all forms. Physics engines can execute on CPU, GPU or a custom hardware. With the support for HLSL and DirectX 10, Windows Vista has an awesome platform to enable Physics on the GPU. Multiple GPU support (LDA Linked Display Adapters e.g. SLI and Crossfire) enable game engines to distribute their graphics and physics loads across multiple GPUs."


1160761917gFElmj6gzG_1_1_l.jpg
 
Last edited:
yashooa said:
Hmmm supposedly the R600 has a 512 bit external bus.


"If the 512 memory ring turns to be the real thing, we are talking about 128 GB/s of memory bandwidth with GDDR4 clocked at 2000MHz. We also learned that the R600 may use memory faster than 2000MHz as it will be available by Q1. If ATI keeps pushing the chip we might get even faster GDDR4 chips at production time.
Even the PCB of the R600 will be super complicated, as you need a lot of wires to make 512 bit memory to work. Overall it has the potential to beat Nvidia's G80, but yet again it will come at least three months after Nvidia. The G80's memory works at 384 bit as Nvidia pretty much dis-unified everything in G80 from shaders to memory controllers. Nvidia likes to make rules and probably could not get more than 384 bit wide controller in the chip, as the G80 is still a 90 nanometre chip.
Its a shame that we will need to wait at least until February to see it in action. "


the stuff about 512bit ringbus is true, but that quote has the G80's facts wrong. G80, as I said before, uses Twice the amount of unified shadders (128) than RD600, and nvidia couldn't get more than 384bit unless they want to make the PCB board something like 1.5' long.. they're still using parallel memory controllers. while ATi's ringbus is using serial technology(lots of fewer traces).
 
sephiroth said:
the stuff about 512bit ringbus is true, but that quote has the G80's facts wrong. G80, as I said before, uses Twice the amount of unified shadders (128) than RD600, and nvidia couldn't get more than 384bit unless they want to make the PCB board something like 1.5' long.. they're still using parallel memory controllers. while ATi's ringbus is using serial technology(lots of fewer traces).
I suppose we will see when it comes out. Some are saying Nvidia is not doing unified but hey, either one will haul ass.
 
Last edited:

New Threads and Articles

Back