The LAST R600 Rumours & Speculation Thread
- Thread starter Geo
- Start date
- Status
Unless there was a need to fast track driver development, in light of the direct competition's perceived performance and/or features.
zgemboandislic
Newcomer
If they don't manage to get the damn thing out with or before Vista, I think the ATI part of AMD can start updating their CVs and start looking for a job at NV.
If they don't manage to get the damn thing out with or before Vista, I think the ATI part of AMD can start updating their CVs and start looking for a job at NV.
Riiiight.
Hector: "Dirk, how's that $5.4B investment with ATI working out?"
Dirk: "Well, we're several months in and they're late on their current project."
Hector: "Well, screw it, let's pull the plug then."
Somehow, I'm not seeing that. . . :smile:
zgemboandislic
Newcomer
Lol. Ok, ok, you're right. But some heads MUST roll!
One thing I don't understand is why they are trying to push 80nm when there is obviously enough room for improvment with 90nm, look at the 8800GTX. Couldn't they develop a 80nm and a 90nm part at the same time, or is it too costly?
One thing I don't understand is why they are trying to push 80nm when there is obviously enough room for improvment with 90nm, look at the 8800GTX. Couldn't they develop a 80nm and a 90nm part at the same time, or is it too costly?
The delay of R600 is not because 80nm shrink....
90nm tech has limit to Max about ~700Million transistors count. "If ATI used 90nm instead 80nm it would be harder to finalized the GPU"
Although I still doubt a release later then March, I also do doubt that it would be a software issue that would keep AMD/ATI from releasing the card.
If we believe that the last Spin of R600 acutally fullfilled expectations, then there are imho 2 things that could be a problem. The PCB and the cooling solution. They surely can not come out with Dustbuster Mach2 Turbo or something like that and the 8800GTX is ok when it comes to noise so being much louder is no option. And the thing needs to be 2-Slot and not longer then 30cms incl. cooling solution.
So for me I see it that way :
Most likely : March release still stands
likely : the last R600 spin was not what is was hopped to be
less likely : cooling problem or PCB problem
unlikely : Software problem
If we believe that the last Spin of R600 acutally fullfilled expectations, then there are imho 2 things that could be a problem. The PCB and the cooling solution. They surely can not come out with Dustbuster Mach2 Turbo or something like that and the 8800GTX is ok when it comes to noise so being much louder is no option. And the thing needs to be 2-Slot and not longer then 30cms incl. cooling solution.
So for me I see it that way :
Most likely : March release still stands
likely : the last R600 spin was not what is was hopped to be
less likely : cooling problem or PCB problem
unlikely : Software problem
zgemboandislic
Newcomer
So in other words, NV once again prove that they are the ultimate leader when it comes to performance per transistor.
How'd you come to that conclusion, quite funny.
Yup. NV30 just wouldn't have been possible on 150nm and with 100M transistors. It was just too ambitious and stuff!
More seriously, there's no such problem if your architecture is scalable. In the NV30 days, it certainly wasn't, though.
Uttar
What a bunch of bullshit from the INQ. That site should just not be qouted anymore, please.
I wont even read that story, let alone lend it the slightest shred of credence.
Taking all bets that upon release of R600, the very FIRST paragraph INQs' magic tale of incredible coincidence will contain nothing other than a spool about 3dmegamarks (or 'penismarks' as he calls it. Shouldn't that be e-penis? Oh well).
I hope it's another Ruby, Ruby is sexy...fishwomen and fairies with hairy wings protruding from their backs aren't nearly as sexy as a readhead in skintight leather with no sag.
Unless there was a need to fast track driver development, in light of the direct competition's perceived performance and/or features.
Perceived or estimated? There's not an inch of public available info on how G80 behaves under D3D10 and how good or bad it's drivers are, albeit weeks before the launch funky details about the chip were floating all over the net.
If and mark IF with block letters, another last re-spin was necessary for R600 as some recent rumors seem to indicate, then my imagination cannot reach as far to think of anything that's software related.
As for the features notion above, care to elaborate? It takes a lot more than a couple of months to add anything, unless we're talking about simplistic software improvements due to the programmable nature of the underlying hardware. Other than that D3D10 requirements are so steep that IHVs didn't have the luxury to add much more than the basics and they don't usually add an algorithm overnight.
Dunno about the delay but I can easily believe to this:
I'm really looking forward to see R600 in action with some amazing demo that the ATI demo team is surely capable of (nvidia underdelivered big time on the demos front..)
zgemboandislic
Newcomer
ATI usually has more impressive looking demos. Even today when I link to the Toy Shop demo, some folks are still wondering if that one was really written for R520 LOL
Definitely, Toy Shop demo is still the most impressive demo I have ever seen, forget about the smoke and water demos!
zgemboandislic
Newcomer
Well they were right last time, when R520 had a re-spin...I sure hope they are not right this time
I'm sure "Negotiated the $5.4 billion buy-out of the company I was CEO of at the right time to an over-anxious buyer" makes a fine addition to Orton's CV! Just the type of guy nVidia might need in the not to distant future.If they don't manage to get the damn thing out with or before Vista, I think the ATI part of AMD can start updating their CVs and start looking for a job at NV.
And as for the latest R600 delay, wouldn't it only mean a couple of extra weeks vs. the previously planned GDDR4-based introduction? (Sorry if I'm not totally up to snuff on the R600 tidings!)
- Status