Xenos - invention of the BackBuffer Processing Unit?

[Tacitblue]

Same 256 GB/sec figure here:

http://pc.watch.impress.co.jp/docs/2005/0520/kaigai02l.gif

Seperate chip, but architecturally an internal bus.

 

[ralexand]

Yes, Dave. We need you.

 

[Fafalada]

Its quite the opposite with the local SPUs

-_-
There's nothing freaking opposite about it - I could write a software rasterizer and render backbuffer entirely in SPU local memories. But apparently that somehow doesn't qualify as the same kind of process because ...
...
...
because of some circular reasoning you posted back there?

The only real difference between the two examples is size of embeded memory (10MB vs 2MB).

People seem to try to dismiss the advantages of Xenos's EDRAM just because an MS PR person made the wrong choice of counting it into the system bandwith.

I would never dismiss it, I'm way too spoiled with working on eDram chips over last couple of year Although granted, Xenos doesn't come with one of my favourite eDram features (free render to texture), but still.
All I'm saying is that if MS keeps parading 256GB/s figure around someone should have paraded the 150GB/s for GS way back, and consequently local store bandwith on Cell as well

 

[rabidrabbit]

Firing Squad has an interview with Ati about xbox360.
http://www.firingsquad.com/features/xbox_360_interview/

 

[Shifty Geezer]

thanks for the link radibrabbit. The image of the block diagram shows 256GB/s to eDRAM...

but the actual interview says what we've all been saying...

ATI: The 2-terabit (256GB/sec) number comes from within the EDRAM, that’s the kind of bandwidth inside that RAM, inside the chip, the daughter die. But between the parent and daughter die there’s a 236Gbit connection on a bus that’s running in excess of 2GHz. It has more than one bit obviously between them.

The GHz clockspeed figure sounds screwy. ATI say...

There’s also a very high speed connection between the parent and daughter die, in case you do get have to get data back and forth, and that connection is, well there’s a 2GHz wide bus connection between them.

2GHz of wide bus? Or a bus that is '2GHz wide' whatever that means? Because he says beteen main GPU and daughter die is 236 Gbit. Is that Gigabits*2 GHz =

472 Giga-gigabits bandwidth?! (what's the next one up from Tera?)

Or is it 236 Gb/s = 30 GB/s? Which means a 15 bit wide bus at 2 GHz? :?

Or is the 2 GHz number marketting nonsense. 2 GHz equivalent? Actually 500 MHz @ 48 bit bus?

Ugh! Still can't make sense of the numbers.

 

[Gubbi]

2GHz of wide bus? Or a bus that is '2GHz wide' whatever that means?

It means that there is a bus, it runs at 2GHz and it is not narrow.

Maybe it's quad pumped from the 500MHz that the GPU runs at. Who knows?

Since there's asymmetrical read and write bandwidth, it's probably separate uni-directional read and write lanes (similar to the FLEX I/O in CELL).

This means that the write channel is 128bits wide, and the read channel 64 bits wide (From the 32/12 GB/s bandwidth figures) for a total of 192.
No idea what the 236 bit-width means. If we assume differential signalling this results in 472 wires between the GPU and the EDRAM.

Cheers
Gubbi

 

[Shifty Geezer]

2GHz of wide bus? Or a bus that is '2GHz wide' whatever that means?

It means that there is a bus, it runs at 2GHz and it is not narrow.

Then how do they get 236 GBits/s then? 236 G/2G = 118 bit...okay. So we're talking a 128 bit bus @ 2 GHz. Is that wide? Seems rather average to me

 

[London Geezer]

2GHz of wide bus? Or a bus that is '2GHz wide' whatever that means?

It means that there is a bus, it runs at 2GHz and it is not narrow.

Then how do they get 236 GBits/s then? 236 G/2G = 118 bit...okay. So we're talking a 128 bit bus @ 2 GHz. Is that wide? Seems rather average to me

128 bit is just fine.

 

[Dave Baumann]

Might actually just be 64bit or less as it is DDR (or more than one data bit per cycle)

 

[ralexand]

Firing Squad has an interview with Ati about xbox360.
http://www.firingsquad.com/features/xbox_360_interview/

Thanks. Some interesting things in there.

The power of the platform [pauses] we’re going to be the most powerful platform out there, we’ve got a lot of innovation in there, we’re not just a PC chip.

And concerning fluid reality:

Well, what we’ve been trying to achieve in this particular go around is, well, realism, you can see it in games like Half-Life 2 where the walls, the environment, it all looks really good and you get a good sense of realism. But the next big hurdle is this fluid reality. The idea that characters in motion, lets say humans in motion, the joints look natural as they move along. That’s involves a lot of vertex processing, and with this unified shader we can put all these shaders towards vertex processing. Cloth, as it’s flying in the wind, like a flag for example, when it drops down on top of something, how that looks as it ripples.

Fur and feathers, the wind blows through them, grass, all that is where this idea of where fluid reality comes from. I’d say we’ve had static quality up until now, but now the fluid rhythm, the motion quality is this next realism that we’re really bringing.

Again, lots of power to devote to vertex processing, generating lots of pixels, to drive HD. HD is the platform of choice for this

That's definitely a good target.