R520 3DMark 2005 ............ 10K +

[ANova]

Agreed. There was ~ 65% clock increase to achieve that score so a 32 pipe R520 even at lower clocks could do 10K. I have a hard time believing that a 24-pipe R520 could do it though unless there were massive improvements to each "pipe".

Would be possible if it were a 24 pipe part running at 600-650 MHz.

 

[neliz]

[quote="ANova"
Would be possible if it were a 24 pipe part running at 600-650 MHz.[/quote]

Which frankly isn't that much higher than current gen clockspeeds

 

[mmp121]

Remember R520 pipelines are 1.3x the powa of R420's mere straws.

Jawed

Hmm if the R520 has 16 pipelines which are 1.3 as powerful as 'normal' pipelines... and a X850XT scores about 6000 in 3Dm2k5... then a R520 would score 7800 points.

your 7800 # is based on the assumption that R520 is a 16 pipe part.

6000 / 16 => 375 per pipe for R480
1.3 * 375 => 487.5 per pipe for R520
487.5 * 16 pipe R520 => 7800 3DMark05 score
487.5 * 24 pipe R520 => 11700 3DMark05 score
487.5 * 32 pipe R520 => 15600 3DMark05 score

Not bad. Now if the R520 hitting 10K in 3DMark05 "demo" was a 32 pipe part:

10000 / 32 => 312.5 per pipe for R520

Looks like R520 is not living up to its "potential"

Numbers are fun to play with. Lets wait and see what reality brings.