r420 benchmarks (not real..extrapolated results)

digitalwanderer said:
If he seems wrong or out there on some of his calls it's because he tosses in some bullshit every now and again just to keep people guessing/keep from breaking any NDAs.

Do you realize how little sense this actually makes?
 
Natoma said:
Oh god......

Please let this all be over on Monday. PLEASE!!!!!! For the love of Jeebus! o_O

Can you please keep your religous views out of these posts . I do not believe in Jeebus nor Do i want you to hear you .
 
If you mean this coming Monday then I wouldn't count on it. Just a hunch, of course. :)
 
Re: SM.3 a must!

ultragpu said:
remember what Tim Sweeney has mentioned:
Long-term, everything's got to be 32-bit IEEE floating point. With the third generation Unreal technology, we expect to require 32-bit IEEE everywhere, and any hardware that doesn't support that will either suffer major quality loss or won't work at all.

Lets remember what a Pixar Renderman developer once told me: "floats aren't precise enough, we need to use doubles".

So looks like you 32b floating point is useless and won't be able to scale to games with toy story like graphics.

Long-term, everythings got to be 64 or 80b IEEE Floating point compliant and therefore support all denormals, NaNs, Inf and all interactions between said special numbers.

Aaron Spink
speaking for myself inc.
 
MuFu said:
If you mean this coming Monday then I wouldn't count on it. Just a hunch, of course. :)

yeah the pclab guy who was at shader day said he couldn't reveal any details until early may(4. or 5. probably) so i don't think there will be anything this monday.
 
jvd said:
Natoma said:
Oh god......

Please let this all be over on Monday. PLEASE!!!!!! For the love of Jeebus! o_O

Can you please keep your religous views out of these posts . I do not believe in Jeebus nor Do i want you to hear you .

:oops:

Heretic!! :devilish: Ahhh Condamn you to hell!
 
MuFu said:
If you mean this coming Monday then I wouldn't count on it. Just a hunch, of course. :)

Sigh............... Why couldn't The InqTM be right on this occasion. The speculation has gone past the useful and the somewhat informed to the absurd. :(

Here's my speculation to add to the fetid stinking brew I suppose:

R420 Pro: Faster than the NV40 Ultra
R420 XT: Faster than the R420 Pro

NV40: Faster than the R420 Pro
NV40 Ultra: Faster than the NV40

Did I get it all or am I missing something? Mebbe I need lessons from Fuad. :rolleyes:
 
I'm still puzzled by the nv490 having originally 475Mhz ( plus some headroom you would guess , 10-20-30Mhz ? ) sent out and then later the preview samples being at 400Mhz with only a 20-30-40Mhz overclock.

Normally cards get better with later cores, not worse.

Was the original nv40 sent out a 12 pipeline job that could do 475Mhz similar to the X800Pro ( if that is 12 pieplines ? ). If it was that it does seem that 475Mhz 12 pipeline cards are similar to 16 pipelines at 400.

Is the Hellbinder / Snipingwaste rumour of 800+Mhz GDDR3 still going to appear on the R420 cards, if so you'd expect it to have a big edge. Of course Hellbinder could be wrong, sorry let me correct that for DW, deliberately incorrect ;) on this one.
 
Now I am trying to figure out the *scrap* comment. Becuase I really dont see how from a performance perspective you could consider it a "Scrap" as in a Close hard fight. Especially in shader intensive games and with AA+AF. Not unless He knows something about a much faster version of the Nv40 on the immidiate horizon.

(Not that the i am implying that X800 is *always* faster in every case.)
 
dizietsma said:
I'm still puzzled by the nv490 having originally 475Mhz ( plus some headroom you would guess , 10-20-30Mhz ? ) sent out and then later the preview samples being at 400Mhz with only a 20-30-40Mhz overclock.

Normally cards get better with later cores, not worse.

Was the original nv40 sent out a 12 pipeline job that could do 475Mhz similar to the X800Pro ( if that is 12 pieplines ? ). If it was that it does seem that 475Mhz 12 pipeline cards are similar to 16 pipelines at 400.

Is the Hellbinder / Snipingwaste rumour of 800+Mhz GDDR3 still going to appear on the R420 cards, if so you'd expect it to have a big edge. Of course Hellbinder could be wrong, sorry let me correct that for DW, deliberately incorrect ;) on this one.
No that 800mhz Ram Was always just a Rumor. I have never tried to imply it was anything but a Rumor.
 
Hellbinder said:
Now I am trying to figure out the *scrap* comment. Becuase I really dont see how from a performance perspective you could consider it a "Scrap" as in a Close hard fight. Especially in shader intensive games and with AA+AF.
Are you talking about 6xAA and 8xAA? Most probably the X800 should be faster there because Nv40 uses a mixture of MSAA and SSAA in those modes. But how about 4xAA? Is that X800 faster there, too?
 
It just occured to me that he could be refering to the Nv40 *being* "Scrap" Vs the R420. Which would make much more sense to me actually. Especially looking the the serious dismemberment the Nv40 is going to get at the hands of the X800XT.
 
I haven't followed the thread from the beginning to be honest, but since the hybrid modes were mentioned, I don't think it wouldn't be welcomed from the userbase to see either hybrid MS/SSAA or just garden variety OGSS support in ATI's upcoming products. Especially if I look back at the ever repeating initiatives from users for supersampling support for R3xx and the endless debates over it.

Obviously it wouldn't be present for performance reasons either.
 
In one of the threads someone mentioned LCD owners wanting
an R420. Does that mean it has sub-pixel aware MSAA?
 
Back
Top