Multiple GPU cores

[Sergio]

I don't know if this has been posted already.

As the trend seems to be set towards dual core CPU's, do you think that the GPU market will follow, or will it stay for a while delivering SLI solutions?

 

[hoom]

Been said before.
Basically a 16 pipe GPU is already a 4 core chip.
R9700 was a double core.

 

[Luminescent]

That goes for the pixel pipelines, but what about the vertex pipes? Are they all on one core, separate from pixel pipelines or what?

 

[KimB]

Once again, multi-core CPU's are only going that way because it's a clean way to introduce parallelism on a software level without changing the underlying instruction set. GPU's are already highly parallel, and are very efficient at it, so multi-core would buy you exactly nothing.

 

[Wunderchu]

Once again, multi-core CPU's are only going that way because it's a clean way to introduce parallelism on a software level without changing the underlying instruction set. GPU's are already highly parallel, and are very efficient at it, so multi-core would buy you exactly nothing.

exactly

 

[Rolf N]

Once again, multi-core CPU's are only going that way because it's a clean way to introduce parallelism on a software level without changing the underlying instruction set. GPU's are already highly parallel, and are very efficient at it, so multi-core would buy you exactly nothing.

Even worse. It adds redundant transistors (cull/setup/interpolation) that could have been spent on other things, and complicates memory management. And unless you're brutally clever, the duplicated vertex fetch/VS units will be wasted, too.

 

[Fodder]

That goes for the pixel pipelines, but what about the vertex pipes? Are they all on one core, separate from pixel pipelines or what?

They're seperate from the pixel pipes, but are still 'parallel-ise-able'.

WTF is the word I'm after there?'

 

[ndoogoo]

If the Cell processor is all it is made out be, then my ultimate graphic card would be a PCIe board with 512MB RAM and 16 Cell processors.

No need to buy new hardware for new features, just download the upgrade software.

Totally programable (make your own graphic card via sofware)

Use its processing power for other thing besides graphics.


Would probably never happen.

 

[Arun]

If the Cell processor is all it is made out be, then my ultimate graphic card would be a PCIe board with 512MB RAM and 16 Cell processors.
No need to buy new hardware for new features, just download the upgrade software.
Totally programable (make your own graphic card via sofware)

Nooooooooooo!
Quick, someone hide this before Dave sees this!

Uttar

 

[Tim Murray]

If the Cell processor is all it is made out be, then my ultimate graphic card would be a PCIe board with 512MB RAM and 16 Cell processors.
No need to buy new hardware for new features, just download the upgrade software.
Totally programable (make your own graphic card via sofware)

Nooooooooooo!
Quick, someone hide this before Dave sees this!

Uttar

Utty, the noobs are making me cry. *cries*

 

[KimB]

If the Cell processor is all it is made out be, then my ultimate graphic card would be a PCIe board with 512MB RAM and 16 Cell processors.

And it'd be a fantastic competitor to the RIVA TNT.

 

[olivier]

If the Cell processor is all it is made out be, then my ultimate graphic card would be a PCIe board with 512MB RAM and 16 Cell processors.

And it'd be a fantastic competitor to the RIVA TNT.

cell is powerfull dont forget !!!! so i think you are a little hard with it ... it should be attt least powerfull like a radeon 7000 and with the feature of a voodoo 1 !!

 

[Rolf N]

cell is powerfull dont forget !!!! so i think you are a little hard with it ... it should be attt least powerfull like a radeon 7000 and with the feature of a voodoo 1 !!

Why should Cell be any more powerful than any other stream processor with the same transistor budget?

Because it has a cute name? Because IBM designers can ignore the laws of physics while others cannot?

 

[Wunderchu]

Why should Cell be any more powerful than any other stream processor with the same transistor budget?

Because it has a cute name? Because IBM designers can ignore the laws of physics while others cannot?

"We've done a lot of work in the design center for proof of concept," he said. "I think the original 'Cell' vision was not (for) any one product."

Kahle said it was too soon to talk about some of the production specifics of "Cell," like the manufacturing process that would be used to make the chip or how soon it will be coming off production lines in volume.

But he said the chip would address many of the problems inherent in chip-making today, such as the difficulty of producing processors with smaller and smaller features, while keeping down their power requirements and heat output.

He also said he was spending 20% to 30% of his time thinking about products to follow up on "Cell," which is built to be reconfigured easily and without extensive redesign of the hardware itself.

"We're being fairly general purpose about it," he said.

[source: http://www.beyond3d.com/forum/viewtopic.php?t=15304&postdays=0&postorder=asc&start=240 ]

 

[Rolf N]

That doesn't answer my question.
I don't care about how hard they work on application flexibility or heat distribution.
I want to know why some people believe that a given amount of transistors can produce higher arithmetic throughput than would otherwise be imaginable just because these transistors form a design known as Cell. That's as irrational as it gets IMO.

This Cell hysteria basically implies that e.g. you cannot generally build a 32 bit floating point multiplier in less than x transistors, and you cannot run it faster than at y Hz (with some correlation between x and y), but if you happen to work on a design called Cell you suddenly can.

 

[Simon F]

This Cell hysteria basically implies that e.g. you cannot generally build a 32 bit floating point multiplier in less than x transistors, and you cannot run it faster than at y Hz (with some correlation between x and y), but if you happen to work on a design called Cell you suddenly can.

That's because the cell architecture is partly kept in hyperspace.

 

[Mariner]

This Cell hysteria basically implies that e.g. you cannot generally build a 32 bit floating point multiplier in less than x transistors, and you cannot run it faster than at y Hz (with some correlation between x and y), but if you happen to work on a design called Cell you suddenly can.

That's because the cell architecture is partly kept in hyperspace.

Aha - so Cell is, in fact, a 'Mind' then.

 

[Simon F]

Aha - so Cell is, in fact, a 'Mind' then.

"In anticipation of a new console's rival"

 

[Rolf N]

This Cell hysteria basically implies that e.g. you cannot generally build a 32 bit floating point multiplier in less than x transistors, and you cannot run it faster than at y Hz (with some correlation between x and y), but if you happen to work on a design called Cell you suddenly can.

That's because the cell architecture is partly kept in hyperspace.

While that's a bit unlikely, it is at least a rational explanation. Exactly what I'm looking for

 

[ndoogoo]

That's because the cell architecture is partly kept in hyperspace.

Well at least it will not be lonely, it has the PowerVR5 for company