R420 Benchmarks

[kyparrish]

bogus

 

[Gagabiji]

Originally posted by: SneakyStuff
haha there is no WAY that's true, the 6800u would never lose at EVERY test. I think i'll wait until a REAL benchmark comes out eh? 😉

How do you know? It could very well beat the stuffing out of the 6800Ultra, but then again, IT could get the stuffing beat out of it. Only time will tell.

 

[Acanthus]

Its just a wild guess, nothing to see here.

 

[modedepe]

wtf dreaming up their own numbers for this? lame

 

[edmundoab]

Looks ATi biased benchies

 

[Viper96720]

repost btw though now it's in the right place
http://forums.anandtech.com/messageview.cfm?catid=27&threadid=1297064&enterthread=y

 

[Lonyo]

Originally posted by: Jeff7181

Originally posted by: Lonyo
I know they cannot be used as anything even vaguely indicative, but you cannot say "lets compare the NV50 to NV40 by extrapolating".
The R420 is supposedly based on the R3xx cards, so it does make limited sense to extrapolate figures (looking at core clock/pipeline/mem clock increases), but impossible to do unless you have a linear change or know the curve with which performance chances in relation to clocks/pipelines.

It's not as absurd as looking at NV50 performance by extrapolating NV40 numbers, since we don't know NV50 specs and it probably won't be based on the same core.

Still a pointless exercise though, unless they know something the rest of us doesn't.

Ummm, we do know NV40 specs... the NDA was lifted over a week ago. Where's you been? 😀

Only thing we don't know is the exact clock speed they'll be released at.

I said we don't know NV50 specs 😛

 

[Gamingphreek]

That benchmark is load of crap. I mean cmon almost 1.5 times the scores. COME ON!! 16 vs 12 pipelines.
-Kevin

 

[Jeff7181]

Originally posted by: Lonyo

Originally posted by: Jeff7181

Originally posted by: Lonyo
I know they cannot be used as anything even vaguely indicative, but you cannot say "lets compare the NV50 to NV40 by extrapolating".
The R420 is supposedly based on the R3xx cards, so it does make limited sense to extrapolate figures (looking at core clock/pipeline/mem clock increases), but impossible to do unless you have a linear change or know the curve with which performance chances in relation to clocks/pipelines.

It's not as absurd as looking at NV50 performance by extrapolating NV40 numbers, since we don't know NV50 specs and it probably won't be based on the same core.

Still a pointless exercise though, unless they know something the rest of us doesn't.

Ummm, we do know NV40 specs... the NDA was lifted over a week ago. Where's you been? 😀

Only thing we don't know is the exact clock speed they'll be released at.

I said we don't know NV50 specs 😛

Hmm, guess I read it wrong the first time, sorry.