here is the link to that 7800 scores 7800 thread

[tyx]

OK pls don't flame me for double posting :(
I would have posted this there but its locked now.

http://www.hardspell.com/news/showcont.asp?news_id=14305

*me hopes its not true

[Dazz]

Doesn't look so impressive would of expected alittle higher due to the extra 8 pipes it uses looks like it maybe bandwith limited now. Besides the 7800 is really a refresh from the NV40/45 You can't consider the NV45 as a refresh as it only brings PCI-E and thats it.

Also 7800pts is pretty low considering my SLi MSi 6800 both @ 430/700MHz got just shy of 9000pts so when you think about it it should be on par with my cards 12x 430MHz aka 24 pipes and 700MHz 256bit x 2 1.4Ghz which is pretty much bang on with the 7800 Specs.

Although come to think of it 3D Mark2003 and 2005 makes heavy use of Vertexes and the G70 only has 8 so only an extra 2 units which mayvery well be the problem. As for games i think it will show little improvment one thing i have noticed with games is that they are becoming more and more CPU limited. CPU's are pretty much at a stand still really they an't getting much higher. Can't count Dual cores as no games support SMP. Once games support SMP then we will see it pulling away on dual cores as AI and pysics and being done on the 2nd core so more processing power for gfx is used on the other core.

[Clay]

OK pls don't flame me for double posting :(
I would have posted this there but its locked now.

http://www.hardspell.com/news/showcont.asp?news_id=14305

*me hopes its not true
Let's not forget how ridiculously easy it is to fake screenshots. The burden of proof is very much alive at this point.

For example...

http://img196.echo.cx/img196/4269/fake0at.jpg

[Dazz]

|God that looks horrable thats what happens when you save a jpeg from a jpeg :D

[vX]

If it's true, then this IS a refrsh product, regardless of what others seem to say. A x850xt oc'd can do 6000-6500 so this isn't a huge leap. IF these scores are real.

[Clay]

|God that looks horrable thats what happens when you save a jpeg from a jpeg :DExactly, artifact city. :)

[serenity]

This is a refresh product not a completely new architecture. I'm not dissappointed with the score.

[SH64]

Can't say i'm not disappointed with that score .. i seriously hope its not true :|

This is a refresh product not a completely new architecture. I'm not dissappointed with the score.

Its NOT a refresh .. its a whole new generation called GeForce 7 series thats supposed to use CineFX & Intellisample HCT 4.0 technologies.

[MUYA]

Been told by ppl, in CPU limited situations (nearly all the time) the g70 will be 50-80% faster in high res with AA and AF applied. Didn't say which aa and af mode though.

[Clay]

Been told by ppl, in CPU limited situations (nearly all the time) the g70 will be 50-80% faster in high res with AA and AF applied. Didn't say which aa and af mode though.You've usually been spot on with your sources in the past so this is good to hear. :cool:

[MUYA]

You've usually been spot on with your sources in the past so this is good to hear. :cool:
Hopefully they are right or...wrong...in terms of better numbers. But I get an indication that faster CPUs are needed to keep up with all these new GPUS round the corner.

I certainly hope those figures were with at least 4X AA and 16X AF at 1600 by 1200. It should be but cannot confirm.

I remember as I take a step back thought folks, amazing that that gf 256 (was it?) could do 100 fps was it in quake3 without AA and AF back in 1999? :p

[MUYA]

Keep it on topic Red_star

[serenity]

I certainly hope those figures were with at least 4X AA and 16X AF at 1600 by 1200. It should be but cannot confirm.
From that screen capture it looks like the resolution is default (1024x768) and thats why most people are dissappointed.

[Demirug]

As the 3dMark is much more vertexshader limited as normal games this result do not necessary show the real improvement from the last gen.

[DMA]

As the 3dMark is much more vertexshader limited as normal games this result do not necessary show the real improvement from the last gen.

So thats what Baumann@B3D ment when he said that 3dmark-05 has limitations.. :o
I've never been able to understand his hints. :D

[BrianG]

Been told by ppl, in CPU limited situations (nearly all the time) the g70 will be 50-80% faster in high res with AA and AF applied. Didn't say which aa and af mode though.
And this is a good point. We probably won't see the performance advantage of a newer architecture until you apply things like AA and AF. If 3DMark indeed has a vertex input limitation hampering higher scores by the new GPUs, then the benchmark can not be a valid measure of generational improvements.

I am still far more interested in Doom3 scores with AA and AF at higher resolutions. More specifically, I want to see how RoE performs on the new GPUs versus the current generations. Half Life 2 and Far Cry, a little, but mainly things like HDR performance.

If these scores translate to real performance, I am still a little disapointed. But I understand that the raw fillrate numbers can not continue to go up as incrementally as the effective fillrate, meaning performance now has to be measured with AA and AF applied. Which brings is the other problem of apples to apples comparison because every GPU has subtle nuances in implementation and IQ. Gone are the days of direct comparison.

The G70 and R520 are going to be the most difficult generation to benchmark comparatively. I have sympathy for those that have the task of cross platform performance comparisons, especially Crossfire vs. SLI. The range of hardware will be huge as now you have motherboard factored in as a key component.

[harl]

As the 3dMark is much more vertexshader limited as normal games this result do not necessary show the real improvement from the last gen.

what about SLI scores?
are in AFR mode?

Becasue in the other modes both cards must do all the scene geometry
(so this modes are Vertex shader limited)

[JoKeRr]

well, at least this score makes ppl with 6800U GT or NU or 6600s feel better b/c they're not that far behind atm, right???

[Demirug]

what about SLI scores?
are in AFR mode?

Becasue in the other modes both cards must do all the scene geometry
(so this modes are Vertex shader limited)

That is the reason why AFR is the default SLI Mode.

[nVidiaGuru]

I lost all faith in 3dmark when i upgraded from an athlonxp to a 64 and my score went down 200 points, but my games performance took the fu(k off, night and day difference. Then i upgraded to windows x64 and all my games once again responded nicely with less drops in fps, but my 3dmark still lost another 50-100 points....3dmark is worthless, when game performance goes up and scores go down we are in no way getting an accurate score to how well our machines are doing. Sure its the graphics card its measuring, but games dont use just the graphics card. The sooner we put futuremark out of business the sooner we get better gaming video cards

[Dazz]

Hell yeah i want to see how it haddles HDR lighting it's the step forward to cinimatic gaming.

Hey BrianG noticed you have a Athlon64 X2 4800+ nice i should be getting my 4400+ some time next week also. I just got myXmystique card today also Dolby Live man is reall nice picked up for 50% less then the RRP heck i cost half as much as an Auidgy 2 Zs lol :D

[ChrisRay]

I dunno. That score looks reasonable and consistent with the specs I have seen on the G70. 3dmark05 is insanely vertex dependent when it comes to rendering power. There are several ways of testing this. You can disable vertex engines on current NV4x softwares. The most interesting thing is. The higher the pixel performance you have the quicker the software becomes vertex limited. I did a test a while back which showed vertex scaling in 3dmark to give you a better idea of how sensitive the software is to vertex unit performance verses pixel performance.

http://www.nvnews.net/vbulletin/showthread.php?t=39276&highlight=3dmark05

http://members.cox.net/omega1979/nvnews/vertex.png

Given the fact that two 6800GTS in SLI have equivalent pixel fillrate and much higher vertex processing than 7800GTX in AFR mode. . These results really dont surprise me. This is excluding of course any changes done to texturing ALU's ect. Which are always optimal on NV4x without AF enabled.

[MUYA]

From that screen capture it looks like the resolution is default (1024x768) and thats why most people are dissappointed.
I wasn't refering to the 3D Mark scores

[OWA]

well, at least this score makes ppl with 6800U GT or NU or 6600s feel better b/c they're not that far behind atm, right???
Not me. I was sure I'd upgrade to a newer card but if it's just marginally better I won't do it and I've been itching to upgrade for a while since nVidia skipped their refresh.

[JoKeRr]

did u guys notice on the greece website that's selling MSI 7800GTX, it had Vertex and pixel shader 4.0?? Guess that's a typo? maybe not??