PC Mag on Graphics

[Geo]

Mar 2, 2004

"ATI's chips process all texture data in a 24-bit floating-point format, while the nVidia GeForce FX line processes data in either 16-bit or 32-bit floating-point. This difference has created some headaches for software developers. As a result, software written for ATI cards often runs slower on nVidia hardware."

Damn that ATI for sabotaging NV cards.

Commenting on NV's compiler drivers:

"Additionally, tweaks like these can be somewhat fragile. If a game is patched, or a mod created for a game, it might not work properly with this type of on-the-fly optimization. The result may be incorrectly rendered images. This would require the release of a new driver to patch the problem. There have been reports of compatibility issues with a few game modifications."

Really, I missed those reports --what are they? (Not saying they don't exist; just hadn't heard).

 

[Pete]

ATI's chips process all texture data in a 24-bit floating-point format

Is that always true, or true with qualifications?

 

[Dave Baumann]

It not a correct statement at all. They don't know what they are talking about.

 

[Quitch]

Isn't that second quote from the mag almost an exact copy of some quote regarding the latest 3DMark03 build? Possibly even an nVidia quote?

 

[Pete]

Thought so, Dave. My time spent here hasn't been in vain!

Quitch, that second quote sounds very similar to what The Carmack said in a .plan file a while ago, and what The Newell said at ATi's Shader Days. The mod issue is new to me, though.

 

[2senile]

Mar 2, 2004

As a result, software written for ATI cards often runs slower on nVidia hardware."

Damn that ATI for sabotaging NV cards.

Shouldn't it say "As a result, software written for DirectX 9 Pixel Shader 2.0 always runs slower on nVidia hardware."??

I'm not trying to be biased; it just seems to me that ATi are being used as a scapegoat. Of course, I could easily be misunderstanding things & would be grateful if somebody could explain to me.

This article on 3DCentre seems to take the same stance.

http://www.3dcenter.org/artikel/2004/01-10_english.php

"If the shader program was coded in a radeon-optimized format , the CineFX architecture will additionally be slowed down. The way instructions are coded in favour of the Radeons causes execution interruptions on Nvidia hardware. The Radeons are not so vulnerable, admittedly it's because of their simpler shader implementation."

 

[Quitch]

May we know the name of this PC mag?

 

[Bouncing Zabaglione Bros.]

I guess that's one way of turning around a statement which is that unless you code specifically for GFFX, your shaders will be slow on Nvidia hardware. I wouldn't be surprised if this particular kind of wording has come from Nvidia at some point.

How about:

"If the shader program was not coded in a GFFX-optimized format , the CineFX architecture will additionally be slowed down. Unless the instructions are coded in favour of the GFFX, this causes execution interruptions on Nvidia hardware. The Radeons are not so vulnerable, admittedly it's because of their simpler shader implementation."

As for the "simplier shader implementation" I guess they are trying to say that because ATI only supports the 24 bit standard at good speed and quality, it must be "simpler" than the more complex Nvidia implementation that offers FP16 speed at low quality, or FP32 high quality at low speed.

 

[kemosabe]

May we know the name of this PC mag?

PC Magazine is the name. http://www.pcmag.com/

 

[T2k]

BTW, PC Magazine is not a bad magazine.
I'm subscribing a dozen similar magazine and I definitely like it - of course, it's not about the hottest graphics card or similar stories.

PC Magazine is always a good read.

 

[digitalwanderer]

PC Magazine is always a good read.

Is that the one Dvorak has a column in? I always liked Dvorak.

 

[cthellis42]

Yeah, Dvorak's in there. (Usually with a personal column and sort of a "summing up" of a lot of different things.) He can be boneheaded from time to time as well, but he's usually interesting to read.

 

[digitalwanderer]

Yeah, Dvorak's in there. (Usually with a personal column and sort of a "summing up" of a lot of different things.) He can be boneheaded from time to time as well, but he's usually interesting to read.

Well back when I used to really like him I didn't know anywhere near as much about PCs as I do now so I can't really agree/disagree, I just remember I used to look forward to reading his columns working midnight shifts at US Steel baby-sitting a VAX cluster because I liked his writings style and humor as well as some of his insites.

I remember PC Magazine, they didn't used to suck...I actually liked it.

 

[Tim Murray]

You know what I miss?

Byte.

 

[T2k]

PC Magazine is always a good read.

Is that the one Dvorak has a column in? I always liked Dvorak.

Yeah, it's him. I like him, still (even though I know much more about PC/Mac/whatever than him )... as well as the other columnists.

 

[T2k]

You know what I miss?

Byte.

Hehe...

 

[digitalwanderer]

You know what I miss?

Byte.

Was that the old VIC-20 mag or was that "Compute"?

 

[Geo]

Yes, "PC Mag" = PC MAGAZINE, one of the grand old ladies of the field of consumer PC magazines. Their Editor's Choice awards use to drive most of my buying decions until, oh, the mid-90's. Dvorak has two columns in it each month.

My other common abbreviation is "Max PC" = MAXIMUM PC (formerly BOOT).

Sorry for the confusion.

 

[Geo]

PC Magazine is always a good read.

Is that the one Dvorak has a column in? I always liked Dvorak.

Don't let it go to your head, but you remind me of him a little at times. That sort of Happy Warrior thing. Doesn't surprise me you recognize a kindred spirit.

 

[Xmas]

As for the "simplier shader implementation" I guess they are trying to say that because ATI only supports the 24 bit standard at good speed and quality, it must be "simpler" than the more complex Nvidia implementation that offers FP16 speed at low quality, or FP32 high quality at low speed.

No, they are not.