When Tuesday does the G70 NDA expire?

[Megadrive1988]

Yep, I raised that some issue with Brent/Kyle over in [H] forums and they basically shot me down and said that's how they do it over there.

As Dave points out in his review, pixel fillrate is 6.8 Gigapixels / second. Texel fillrate is 10.3 Gigapixels / second.

that makes a lot more sense.

 

[Blastman]

Everyone is using their own demo, so the results are not comparable from site to site.

Well, you’d think there would be some consistency to the benches as they’re all running the same demo. amdzone also had BF2 benches which are similar to those of the other 3 sites I linked -- with the X850 XTPE comfortably ahead of the 6800U.

Any review where they have the X850 XTPE beating the 6800U in only 1/8 game benchmarks (12x10-4AA/16AF) is bias in the extreme. I can only conclude FS are hand picking benchmarks (games) to favor one IHV. That FS review looks like it was paid for by NV one way or another.

 

[Edge]

> "The real conclusion here to draw is that the RSX/PS3 won't have it, which is bad for Sony."

Hardly the deciding factor in the console race at all.

 

[mercury]

Having a looking FSAA 4xAA and 16xAF .. ATI's 850XT PE whoops Nvidia G70's ass.

Even at optimal filtering(i take it's CP controlled) the ATI is faster.

For a new card . .I would've expected better.

Relying on a single benchmark, and a single test within that benchmark (GT2) is not a reliable way to compare the two cards.

Ok .. noted. Except that.

http://www.beyond3d.com/previews/nvidia/g70/index.php?p=12

Far Cry, 4x FSAA & 8x AF (FPS) 	640x480 	800x600 	1024x768 	1280x1024 	1600x1200
7800 GTX 	                        80.3 	  80.2 	      80.1 	    77.2 	     62.1

http://www.beyond3d.com/reviews/ati/r480/index.php?p=9

X850 XT PE (FPS) 	640x480 	800x600 	1024x768 	1280x1024 	1600x1200
4x FSAA + 8x AF 	   108.1 	   108.3 	   106.1 	   87.4 	   67.0

ATI still whooping G70 ass

But......
http://www.firingsquad.com/hardware/nvidia_geforce_7800_gtx/page13.asp

 

[digitalwanderer]

But......
http://www.firingsquad.com/hardware/nvidia_geforce_7800_gtx/page13.asp

In that link you provided they're testing against an X800XL, the benchies shown above were with an X850 XT PE....a bit better ATi card.

 

[Richard]

But......
http://www.firingsquad.com/hardware/nvidia_geforce_7800_gtx/page13.asp

In that link you provided they're testing against an X800XL, the benchies shown above were with an X850 XT PE....a bit better ATi card.

DW, take a better look.

Anyway, different demo, different test ram, different operating system service pack, and most likely very different methodology.

 

[Geeforcer]

Having a looking FSAA 4xAA and 16xAF .. ATI's 850XT PE whoops Nvidia G70's ass.

Even at optimal filtering(i take it's CP controlled) the ATI is faster.

For a new card . .I would've expected better.

Relying on a single benchmark, and a single test within that benchmark (GT2) is not a reliable way to compare the two cards.

Ok .. noted. Except that.

http://www.beyond3d.com/previews/nvidia/g70/index.php?p=12

Far Cry, 4x FSAA & 8x AF (FPS) 	640x480 	800x600 	1024x768 	1280x1024 	1600x1200
7800 GTX 	                        80.3 	  80.2 	      80.1 	    77.2 	     62.1

http://www.beyond3d.com/reviews/ati/r480/index.php?p=9

X850 XT PE (FPS) 	640x480 	800x600 	1024x768 	1280x1024 	1600x1200
4x FSAA + 8x AF 	   108.1 	   108.3 	   106.1 	   87.4 	   67.0

ATI still whooping G70 ass

From X850 review:

Far Cry is a DirectX9 title that uses a number of DirectX9 and DirectX8 Shaders. The 1.3 patch also utilises PS2.x and PS3.0 shader paths, making use of their longer instruction length capabilities to collapse some rendering passes. The test we are using is a custom Firingsquad demo and the highest shader profile available is used, which means PS2.b for the X800's and PS2.0 for the 9800.

From 7800 review:

Far Cry is a DirectX9 title that uses a number of DirectX9 and DirectX8 Shaders. The 1.31 patch also utilises PS2.x and PS3.0 shader paths, making use of their longer instruction length capabilities to collapse some rendering passes. The test we are using is a custom Firingsquad demo and the highest shader profile available is used, which on these boards result in the Shader Model 3.0 path.

 

[pat777]

Maybe Dave was testing different levels on those different reviews.

 

[Unknown Soldier]

Geeforcer .. so??

The GTX is a new Card .. the X850XT PE an older card.

If it was the older 6800Ultra I could understand.

Also note the low fps at low res.

Pat777 .. I don't think so .. that benchmark would've been standard FS benchmark.

The fact that FC patch 1.30 and 1.31 patches were used shouldn't make a difference .. unless changes in the changelog indicated so.

Actually i'm surprised Dave didn't use the Patch 1.32 .. but then again that's to fix a bug using the R520.

US

 

[ondaedg]

Geeforcer .. so??

The GTX is a new Card .. the X850XT PE an older card.

If it was the older 6800Ultra I could understand.

Also note the low fps at low res.

Pat777 .. I don't think so .. that benchmark would've been standard FS benchmark.

The fact that FC patch 1.30 and 1.31 patches were used shouldn't make a difference .. unless changes in the changelog indicated so.

Actually i'm surprised Dave didn't use the Patch 1.32 .. but then again that's to fix a bug using the R520.

US

The low fps at low res is strange. That is typically due to driver inefficiency if I am not mistaking. It could also be that they are no longer replacing shaders...

 

[_xxx_]

As I get it, 7800 was running SM3.0 path and X850 SM2.0b path? What's the speed difference on nV between the two paths?

ondaedg: the 7800 does not replace shaders like the previous nV chips, that's been confirmed in a few reviews.

 

[ANova]

As I get it, 7800 was running SM3.0 path and X850 SM2.0b path? What's the speed difference on nV between the two paths?

+ ~2-5 fps when running SM3 over SM2 on nvidia hardware.

 

[pharma]

Xbitlabs benchies including both 850 XT PE and 800 XL:

http://www.xbitlabs.com/articles/video/display/geforce-7800_3.html

 

[Pete]

ondaedg: the 7800 does not replace shaders like the previous nV chips, that's been confirmed in a few reviews.

Has it been confirmed or merely repeated?

 

[Geo]

ondaedg: the 7800 does not replace shaders like the previous nV chips, that's been confirmed in a few reviews.

Has it been confirmed or merely repeated?

Yeah, I've kinda wondered about that too. At least the firmness with which it has been stated that it *never* does. That sounded to me a bit like an engineer saying "we won't need shader replacements much anymore" turning into "never" by the time marketing got done with it.

 

[DemoCoder]

They need a whole new slew of different replacements due to the ALU changes, and the shader compiler needs more work too.

 

[Pete]

I was thinking of Chalnoth's saying that the one Doom 3 shader that touches most of the scene is optimized to the tune of 20% higher framerates (50 to 60fps). I'm guessing that optimization would remain for G70, though it may also be able to be further optimized to account for the revamped shaders.

 

[Geo]

They need a whole new slew of different replacements due to the ALU changes, and the shader compiler needs more work too.

Heh. So "We Have No Shader Replacements!" because the old ones suck now and we haven't had a chance to do new ones yet.

 

[ondaedg]

They need a whole new slew of different replacements due to the ALU changes, and the shader compiler needs more work too.

Heh. So "We Have No Shader Replacements!" because the old ones suck now and we haven't had a chance to do new ones yet.

Hehe. "We are not doing shader replacements because the new ones aren't ready yet!"

 

[KimB]

Well, most games that use shaders use too many to make shader replacements viable.