ATI R600
- Thread starter Megadrive1988
- Start date
jimmyjames123
Regular
Read more carefully what DaveB wrote in that linked thread. He is saying essentially the opposite of what you are saying.
jimmyjames123
Regular
DaveB seemed quite convinced that he was right. I guess it's DB's word vs HB's word huh? 
We'll just have to wait and see...
We'll just have to wait and see...Megadrive1988
Veteran
well, you might very well be right.
however a cross between R500 and R600 would not be the same as NV2A being a cross between NV20 and NV25. the R500 and R600 are two different generations. The NV25 was simply a beefed up refresh of NV20.
thus, NV2A was between NV20 and its refresh. I suppose one could agrue that NV2A incorporated elements of the NV30 (some geometry features, as once mentioned) but I don't see much evidence of NV30 in NV2A. I suppose some shader length things, but.. well nevermind.
with that said, I am more than willing to believe Dave B over HB on the R500 vs R600 issue.
R600 makes more sense. Microsoft NEEDS the absolute latest ATI technology they can get their hands on, in the fight against PS3.
From the information here, it seems that NV2A also featured NV30's double pumped Z/Stencil, had more FX units per pipeline and NVIDIA also themselve made note that NV2A would feature some of NV3x's vertex processing capabilities (but I don't know what that was).
Megadrive1988
Veteran
R480 it is.
question is, is R420 ==> R480 like R300 ==> R350
question is, is R420 ==> R480 like R300 ==> R350
That's true, with no color writes NV2A can easily achieve a real (measured) fillrate above 1 GPixel/s (where theoretical maximum would be 933 MPixel/s). It's pretty simple to achieve that..one have just to disable color writes and switch on multisampling..DaveBaumann said:
ciao,
Marco
"this" speaks about the current part R420, while "that" speaks of what could have been with the R400.
So, now the quesitons are.
-When is the R480 going to be released?
-What process will it use?
-What will ATI have learned by then or has already learned -- as relevent?
Personally, here is my take. Say they couldn't work the current process they wanted with a reasonable die size. This seems to be a big issue with them. The next was power. So lets say they've advanced these a fair bit.
Say, they can get more transistors in there and keep the power acceptable.
Now, what are they going to do with this?
Well the R480 could be a fairly radical departure from the R420. Maybe, they've learned how cram enough transistors in there that a high speed, more flexible part is possible. Remember, Orton said Nvidia was about 10%-15% larger die, could they squeeze things a bit tighter? Improve yields on the current process to absorb any hit that they might incur with their current setup.
The R400 work wasn't ditched. Merely delayed till it was feasible.
So, now the quesitons are.
-When is the R480 going to be released?
-What process will it use?
-What will ATI have learned by then or has already learned -- as relevent?
Personally, here is my take. Say they couldn't work the current process they wanted with a reasonable die size. This seems to be a big issue with them. The next was power. So lets say they've advanced these a fair bit.
Say, they can get more transistors in there and keep the power acceptable.
Now, what are they going to do with this?
Well the R480 could be a fairly radical departure from the R420. Maybe, they've learned how cram enough transistors in there that a high speed, more flexible part is possible. Remember, Orton said Nvidia was about 10%-15% larger die, could they squeeze things a bit tighter? Improve yields on the current process to absorb any hit that they might incur with their current setup.
The R400 work wasn't ditched. Merely delayed till it was feasible.
Similar threads
