The LAST R600 Rumours & Speculation Thread

Status
Not open for further replies.
When it comes to the figures I suffer nowadays from rumour fatigue and have terrible trouble getting excited one way or another, 750 or 1Ghz or 2Ghz or umptythrumpty Ghz . The numbers just pass before my eyes no I have no real idea how fast it all means in fps.

I think I would be most excited by a really low cost high end part that shook the pricing world apart, but no chance of that of course. :(
 
I'd prefer better use of existing hardware but this is the wrong thread for that :arrow:

I completely agree on the rumour fatigue thing though.
 
Why is 512-bit so impressive? NVIDIA had it since the GeForce 7950 GX2. :p
3Dfx since V5-6000 (and profi Mercury system from 1998 was even 1536bit :devilish: ). Impressive numbers, but effective bandwidth of this solutions is much lower...
 
Notice how the gap between the XTX and the XT seems to be huge? just like the GTX and the GTS. Coincidence?

But the XT sure does have much more bandwidth than the GTX. By about ~30gb/s.

So much for clock domains and 1ghz rumours from hexus. :LOL:
Guess we can expect more leaks later on this month and into the next.

Would you consider the GTS with the memory bandwith of the GTX as bandwith limited ?
 
Look, all props to VR-Zone for nailing G80, but remember the rule of thumb. . . their "ATI sources" generally seem to turn out to be NV sources at the end of the day.

It doesn't jangle any alarms that they are claiming ATI is launching with A12 while they already have results from an A15?
 
It doesn't jangle any alarms that they are claiming ATI is launching with A12 while they already have results from an A15?
Not necessarily. It could be that they are in full production for the equivalent of an X1900 product, and that the chip to be used in the equivalent of the X1950 product (which is functionally identical but higher-clocked) is now sampling.
 
Probably, but who knows if it's a 65nm part, it may just be a final respin on 80nm that they are unwilling to wait the extra time to complete, so they're mass producing an early revision first just to get it out there. Remember R520 and the crappy metal spin batch that was rumored to have ended up in at least the first batch of XL's (Which I think was FUD, but I'm sure you know what I'm talking about)? Perhaps this time they're just going to get the darn thing out the door and will give us the final spin with a nice ol' 50 attached to it. Hmm...X2800 to compete with G80 in Q2, X2850 to compete with G81 in Q3, and X2900 (65nm/R680) to compete with G90/G85 (65nm) going into the end of the year? Sounds plausible enough, although...

Still, I do smell FUD. Wasn't A12 supposedly NOT going to be the release version according to several sources? Either that's flat out wrong, or plans seemed to have changed.

Still, even stranger enough, there was that rumor from Level505 (please don't hit me) that they were going to release the current revision, and revise the chip and release a better version a few months later. As much as we'd like to hate that site, that lines up more or less (except for time frame and only GDDR3 on the early model) with this current rumor. I suppose it's possible plans (launch dates, versions launched) have changed, and l505 could've been right, and now they will make a GDDR4 version with the first chip. It may be or have been a fluid situation depending on how quickly they could get what turned out to be A15 produced, amongst other things. The fact UFO seems to have been planned, cancelled, planned again, cancelled, and now on track again lends credence to that.

DAAMIT, I'm confused.
 
Last edited by a moderator:
Wouldn't A12 date from ~ October?

Also, A12 has been reported at 1Ghz, so it looks like confusion/misinformation all round, as usual.

Jawed
 
Wouldn't A12 date from ~ October?

Also, A12 has been reported at 1Ghz, so it looks like confusion/misinformation all round, as usual.

Jawed
A12 is old yes.. It would be the first samples that actually went to partners.. and I thought they were lucky to high 1Ghz on (with water?)
A15 means it's months away from A12. so if A12 is a 750Mhz variant that is capable of just outrunning G80 and G81 is a speed bump.. then a 1Ghz A15 (with 2Ghzmem) would.. once again be just on-par with a G81..

So what ever changes in the graphics world?
 
3Dfx since V5-6000 (and profi Mercury system from 1998 was even 1536bit :devilish: ). Impressive numbers, but effective bandwidth of this solutions is much lower...

The V5 6000 was slated for $599 MSRP would it had launched if memory serves well. Highly OT, but a K2 that appeared only ~8 months after the projected release of the V5 6k had an initial pricetag of $150. Skim through the review here and you might understand why I sigh every time I hear of TBDRs:

http://www.3dcenter.de/artikel/voodoo5-6000/index6.php
 
This might shock you, but I would suspect the PDC is bigger or as big as the register file. It is actually 256KiB, I think - and not 128KiB like it was suspected earlier based on extrapolation from some PDFs. This is because each half-cluster (8-wide ALU) has its own 16KiB PDC.
I thought Nvidia said the parallel data cache is 16KB at the Supercomputing conference. My thinking is you shouldn't just add up the parallel data caches from the 8 thread processing clusters and call them a 128KB cache because data isn't shared between clusters. At least I don't think it is. Mike Houston seems to think it's 16 KB as well.
 
According to our sources but still a rumor revealed R600 as X2800 series. There are 2 variants for XTX; 1GB and 512MB. X2800XTX runs at 750MHz core, 2.2GHz memory, 512-bit memory interface and GDDR4 http://www.vr-zone.com/?i=4510

If R600 will have at least 24 ROP's, and 4:1 ratio = 128 shaders, and @750Mhz core. "And at least 32 texture units"
It should blow G80 away!
 
If R600 will have at least 24 ROP's, and 4:1 ratio = 128 shaders, and @750Mhz core. "And at least 32 texture units"
It should blow G80 away!

In pure theory what you're describing with an equal or higher shader unit clock domain as G80, I'd suspect it would take a very long hard breath to move G80 a couple of millimeters out of it's pace, but I doubt it would blow it away.
 
Status
Not open for further replies.
Back
Top