What do you expect for R650

[Geeforcer]

I am sure that at that point they have sunk quite a bit of money into R520 development and by releasing it, however short-lived it ended up being, they were able to get some of it back.

 

[Jawed]

They had to get R520 working so that they could be sure other stuff would also work, I expect.

Jawed

 

[Unknown Soldier]

R650 has more texture power

Our sources confirmed that R650 will have a slightly redesigned core and that the main marchitecture difference will be the texture capability of the GPU.

The R650 will have more texture power, but we don’t know if this comes from the higher clock speed of the GPU or if ATI actually plans to redesign the whole core. The end result will be more texture power and faster rendering.

We will try to get more details soon.

http://www.fudzilla.com/index.php?option=com_content&task=view&id=1074&Itemid=1

Hmm.. more than the 16 basic pipeline configuration?

US

 

[Shtal]

Our sources confirmed that R650 will have a slightly redesigned core and that the main marchitecture difference will be the texture capability of the GPU.

Still reminds R520---->R580 3:1 ratio shader power

It seems to me that ATI will keep everything same besides texture unit count for R650. "could be something more"

R520/R580 situation, they just added shader power, that's it. "Nothing else".

So for R600----->R650 will be texture power. "Nothing else"

 

[Shtal]

Can someone speculate something for me; lets assume that the only differentness be in R650 is texture power and clock speed. And also lets assume that R650 will have 64TMU's

R600 16 TMU's vs. R650 64TMU's
R600 740MHz core vs. R650 840MHz "100MHz GPU differentness"
R600 MEMORY 825MHz-GDDR3 vs. R650 MEMORY 1000MHz-GDDR4 "175MHz differentness"

Based on this specs differentness, how much in percentage will it boost performance in Todays games? and will it match G80-GTX/Ultra or NOT - OR is it still be limited on 64 shader units vs. Nvidia 128 shader units.

[Edit: I don't expect for nobody to come up in percentage "Just estimate guess"]

 

[Geeforcer]

I don't remember if it was covered in one of the R600 threads or not, but what was ATI's way of managing huge pin-out in the face of shrinking die size?

 

[Geeforcer]

Can someone speculate something for me; lets assume that the only differentness be in R650 is texture power and clock speed. And also lets assume that R650 will have 64TMU's

R600 16 TMU's vs. R650 64TMU's
R600 740MHz core vs. R650 840MHz "100MHz GPU differentness"
R600 MEMORY 825MHz-GDDR3 vs. R650 MEMORY 1000MHz-GDDR4 "175MHz differentness"

Based on this specs differentness, how much in percentage will it boost performance in Todays games? and will it match G80-GTX/Ultra or NOT - OR is it still be limited on 64 shader units vs. Nvidia 128 shader units.

I am more interested in how it would do against G80 with 1955MHz shader core, 1.21 gigahertz RAM and 88 TMUs.

 

[aeryon]

I am more interested in how it would do against G80 with 1955MHz shader core, 1.21 gigahertz RAM and 88 TMUs.

short answer, R650 will fail like R600 failed against G80...

 

[Shtal]

short answer, R650 will fail like R600 failed against G80...

Could you follow me in details please, step by step, I just don't understand how R650 will fail....

Lets say ROP's and Shader count will remain same ---->only texture + clock speed will be increase.

 

[DegustatoR]

I am more interested in how it would do against G80 with 1955MHz shader core, 1.21 gigahertz RAM and 88 TMUs.

88 TMUs?..

 

[Geeforcer]

Yes, 88 TMUs. While the rest of the specs are in a flux, using exactly 88 TMUs will result in a very future-proof design.

 

[aeryon]

Could you follow me in details please, step by step, I just don't understand how R650 will fail....

Lets say ROP's and Shader count will remain same ---->only texture + clock speed will be increase.

no need to go into deep details to see the big picture. if R650 is only a 65nm speed bumped version of R600 and even if R650 will also include some minor tweaks (and please don't argue with big changes, time is simply too short), I cannot see how it can offer 2x more real gaming performance than R600 to compete with G92... after we can always argue about details and specific benchs where R600/650 macroarchi should be faster than G80/92, but at this time it's pointless.

 

[silent_guy]

Based on this specs differentness, how much in percentage will it boost performance in Todays games?

42% !

(Shtal, there is absolutely no way to give a reasonable answer to your question...)

 

[DegustatoR]

Yes, 88 TMUs. While the rest of the specs are in a flux, using exactly 88 TMUs will result in a very future-proof design.

Really strange number if you ask me. 96 or 64 maybe? 64 is effectively double of what G80 has, and i think this will be pretty much enough for G90.

 

[Shtal]

42% !

(Shtal, there is absolutely no way to give a reasonable answer to your question...)

Let me ask you simple question: What makes G80GTX faster then R600XT ?

 

[DegustatoR]

no need to go into deep details to see the big picture. if R650 is only a 65nm speed bumped version of R600 and even if R650 will also include some minor tweaks (and please don't argue with big changes, time is simply too short), I cannot see how it can offer 2x more real gaming performance than R600 to compete with G92... after we can always argue about details and specific benchs where R600/650 macroarchi should be faster than G80/92, but at this time it's pointless.

We have to assume that r650 was in development for autumn'07 release and r600's delays hasn't affected it in any way. So time isn't really an issue here, design decisions are.

 

[silent_guy]

Let me ask you simple question: What makes G80GTX faster then R600XT ?

A bug in R600? Texture limited? Inefficient ring bus?

Pretty much everybody thought that R600 would outperform a GTX, given the MC width and shader specs. Yet all of us were seriously wrong. That being the case, how do you expect anyone to come up with a meaningful percentage? Would it make any of us wiser if I really meant 42%? (That's average of course: 39.5% for Doom3 and 44.7% for CoD.)

 

[aeryon]

We have to assume that r650 was in development for autumn'07 release and r600's delays hasn't affected it in any way. So time isn't really an issue here, design decisions are.

R650 is heavily based on R600 and no one will argue against that. So the only thing we can assume for sure is that R650 work started after R600 nearly finished debugging process. IMHO, R650 started really in January when R600 A13 silicon came back. If R650 comes next September, it gives 9 months for die shrink + small tweaks, which seems right to me (but I may be wrong)

 

[Shtal]

If "silent_guy" could not see how R600 is limited in a lot of situations vs. G80GTX I'm very surprise.

I don't expect for nobody to come up in percentage by how much R650 be faster. All I would like to know what is limitation is and how R650 will be based on this estimate change.

 

[seahawk]

If we knew. That is the main topic for future analysis. It could be driver issues (I doubt that though), it can be a bug or a design decision.

I can bei an error in the ROPs or that they had to drop color compression and did it on purpose. The question is unanswered at the moment.