r420 benchmarks (not real..extrapolated results)

Hellbinder said:
Evildeus said:
Didn't find those extrapolation really good appart for the XT.
So then you must not find the Nv40 numbers to be *really good* either.?
No, i did find them really good. But as some people on this board (hint hint ;)) are saying that R420 pro >> than NV40U, then i'm disapointed ;)
 
Hellbinder said:
Besides I am one hell of a lot more reliable that some of you give me credit for. As some who actually pay attention could easily attest to.
I can take a hint, and I'll happily give credit where credit is due. :)

Hellbinder is a hell of a lot more reliable than I'm seeing him get credit for here, a LOT! 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.

Now granted he can be a bit enthusiastic at times defending/evangelizing his fave IHV, but I think a lot of folks (myself included) tend to do that from time to time.

Personally I like the info HB gives and I find it very reliable, once you figure out how to tell the actual info from him just pulling your leg. ;)

DaveBaumann said:
IMO. NV40 vs R420 = "scrap".

I believe Dave is using "scrap" to me a fight, and a good one at that.
 
Evildeus said:
digitalwanderer said:
Now granted he can be a bit enthusiastic at times defending/evangelizing his fave IHV, but I think a lot of folks (myself included) tend to do that from time to time.
A bit? It's an euphemism :?:
The Dig hugs his shiny new NF7-S and winks slyly.

BTW-I know "an euphemism" is technically correct, but doesn't "a euphemism" work better off the tongue?
 
Evildeus said:
I always thought the a/an rules worked off pronunciation, not spelling. Hence an history if you're French and have a fear of the letter H, an STD, an FTP, and on the flipside, a euphemism and a ukelele. I'm probably wrong though. :)
 
dont believe in any of these

ok first the x800 series arent even released so the scores of the final products will obviosuly differ. but the question is by how much will it differ from this fake post?
my theory is that based upon the final Products of either company, 6800U will own x800xt. why? coz NV always had higher clocked core and memory than ATI. so no matter what core speed x800xt will come up with, 6800U might just be a tad higher. so if it is 16pipline vs 16pipline, 6800U will hav this round.
as for x800pro......no chance in hell agaisnt 6800U.
first, it should be lower clocked and has less pipelines than the XT, so by that i dont c a way for it to succeed 6800U. it could compete with 6800NU though.

another thing i dont get is how come ati claims SM.2 will run faster than SM.3 when NV claims SM.3 will run faster under some areas.

some minor factors could be take into account is the lower colour precision from atixr420(24bit) against NV40(32bit). i mean come on, it's imbarassing they still only support 24bit for a next gen card. o_O
 
Re: dont believe in any of these

ultragpu said:
some minor factors could be take into account is the lower colour precision from atixr420(24bit) against NV40(32bit). i mean come on, it's imbarassing they still only support 24bit for a next gen card. o_O

Can you tell us wht this is embarassing? What differences have you seen between FP24 and FP32?
 
DaveBaumann said:
IMO. NV40 vs R420 = "scrap".

Can you be a little clearer as I'm wearing my idiot hat today? :D

The attempt for comparisons are scrap because both products are not out to really compare thus=scrap? Or actual product to product equates to one product being scrap?

Please :D
 
this matters

Can you tell us wht this is embarassing? What differences have you seen between FP24 and FP32?

here is the Interview with Developers - Multiple Pixel Shader Precision modes
Tim Sweeney, Epic Games
For games shipping in 2003-2004, the multiple precision modes make some sense. It was a stopgap to allow apps to scale from DirectX8 to DirectX9 dealing with precision limits.

Long-term (looking out 12+ months), 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.

gee, what would the x800 users do by then? apparently they wouldnt last very long whenever games fully supports SM.3 come out. 6800U is definately more future proof. and dont forget Tim S. did say "6800 is the first card which can run SM.3 games (etc.third generation Unreal technology) at pretty decent frames."
 
Again, where are the differences? Has Tim shown anyone why he requires IEEE 32-bit, or does he just think that needs to be the case? Tim also happens to be talking about work that Epic has acknowledged that won't here until after 2006 - what will people be using then?
 
Re: this matters

ultragpu said:
gee, what would the x800 users do by then?

Same thing the 6800 users will do.

Ditch their cards for something that actually gets more than 15 FPS in games that come out 2006...

"6800 is the first card which can run SM.3 games (etc.third generation Unreal technology) at pretty decent frames."

Don't forget that in the next room, R420 was running the same 3rd generation Unreal technology, and some report at better frame rates.
 
about the demo

Don't forget that in the next room, R420 was running the same 3rd generation Unreal technology, and some report at better frame rates.
no actually i heard they tested both cards on the demo but r420 was somehow slower but not by much.
Ditch their cards for something that actually gets more than 15 FPS in games that come out 2006...
how come people would assume nv40 runs it only around 15fps when the man said it can run it withpretty decent frames
15fps LMAO is no where near "pretty decent frames" so i would assume 6800U would run it at least above 30fps.

plus there could be some chances that games using PS.3 would come out in 2005. u never know 8)
 
Re: about the demo

15fps LMAO is no where near "pretty decent frames" so i would assume 6800U would run it at least above 30fps.

BTW... the NV40 was running at 500 Mhz, not the 400 Mhz shipping clock. (Dunno what R420 was running at, or shipping at for that matter.)

And I would assume R420 will run it at whatever frames NV40 is. The point stands that in 2006, both of these cards will be closer to the "bottom" of what you want, than to the top.

Do you think a NV30 will be a better card than R300? After all, NV30 has FP32 support...

plus there could be some chances that games using PS.3 would come out in 2005. u never know 8)

And make a difference like in Far Cry? (Zero difference) Probably...but who cares?
 
Re: about the demo

Joe DeFuria said:
Do you think a NV30 will be a better card than R300? After all, NV30 has FP32 support...
Hey, didn't it have higher clockspeeds too? It must have been way better than that POS R300! :oops:























;) :LOL:
 
SM.3 a must!

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.

and yes, r420 must hav lost major quality while running the demo otherwise it wont make any sense. they could just cover the facts and didnt release to the public. by the way, far cry is not a full PS.3 game unlike unreal engine 3. and what i meant before was any games fully supports or requires SM.3 hardware will not run good on r420 coz it will loose major image quality. so basically if any of them comes out in 2005, then r420 will simply loose grounds.
 
Tim also was on record once saying that he thought 32 bit textures in a 16 bit screenmode were a good idea (as a step up from 16/16), which is a really stupid thing actually.
Tim also somewhere down the line thought that negative LOD bias is a nice way to improve texture quality, while in reality this, too, is a really stupid thing.

He's certainly done a lot of respectable work. But I can't get rid of the feeling that he's a bit "far out" sometimes. Ie I agree with Walt ;)
 
nv30 vs r300

Hey, didn't it have higher clockspeeds too? It must have been way better than that POS R300!

i only compare their clock speed when both x800xt and 6800U are 16piplines, 256bit so niether of them hav advantages at those 2 areas. the case of nv30 vs r300 is different however. nv30 has to use more core speed to compensate it's low 128bit and 4*2 pipline.
 
Re: nv30 vs r300

ultragpu said:
i only compare their clock speed when both x800xt and 6800U are 16piplines, 256bit so niether of them hav advantages at those 2 areas. the case of nv30 vs r300 is different however. nv30 has to use more core speed to compensate it's low 128bit and 4*2 pipline.
You don't suppose the fact that the two GPUs will use entirely different architectures will influence the comparison between the two, do you? :|
 
Back
Top