Radeon 9800 Pro preview

[martrox]

Just to be a little bit evil (that's my name no? )

Are those drivers WHQL? Are they publicly available?

*remind me of something *

Seems to me that this is a preproduction "review" as as such doesn't fall within Dave's stated use of publicly available drivers........

Don't you just hate when people have to grab at straws......

 

[Joe DeFuria]

Are those drivers WHQL? Are they publicly available?

From the review:

The 7.84 drivers supplied by ATI are not yet WHQL certified; however, they are WHQL candidates and have been sent to Microsoft for certification, so assuming the pass then everything in this driver should be to WHQL specification.

 

[Evildeus]

In fact, i really don't f*** mind, just showing the one way thinking of some people sometimes and tease a bit

 

[Clashman]

Has anyone tested Truform with this? Is it still software or is it back in hardware again? Inquiring minds want to know.

 

[Ante P]

Has anyone tested Truform with this? Is it still software or is it back in hardware again? Inquiring minds want to know.

http://www.nordichardware.com/reviews/graphiccard/2003/Radeon9800Pro/index.php?ez=16

 

[Clashman]

Has anyone tested Truform with this? Is it still software or is it back in hardware again? Inquiring minds want to know.

http://www.nordichardware.com/reviews/graphiccard/2003/Radeon9800Pro/index.php?ez=16

Must have missed it the first time.

Reading through....

Doh!

 

[demalion]

Hey, ATI isn't sending this comparison chart around, are they?

"Maximum Texels Filtered per Pixel: 768 texels/pixel vs. 512 texels/pixel"

What are they counting here?

My guess was some sort of "tap count" bastardized idea for trilinear filtering at maximum aniso, so 64 tap * 4 * 2 = 512 for nv30, and 96 tap * 8 * 1 = 768...

My only problem now is figuring out fully how to do the math to get 96 tap notation for the R3x0...?

 

[Althornin]

My guess was some sort of "tap count" bastardized idea for trilinear filtering at maximum aniso, so 64 tap * 4 * 2 = 512 for nv30, and 96 tap * 8 * 1 = 768...

My only problem now is figuring out fully how to do the math to get 96 tap notation for the R3x0...?

Could it involve AA level somehow?
6 (AA) * 16 (AF) * 8 (Tirlinear) = 768

 

[demalion]

My guess was some sort of "tap count" bastardized idea for trilinear filtering at maximum aniso, so 64 tap * 4 * 2 = 512 for nv30, and 96 tap * 8 * 1 = 768...

My only problem now is figuring out fully how to do the math to get 96 tap notation for the R3x0...?

Could it involve AA level somehow?
6 (AA) * 16 (AF) * 8 (Tirlinear) = 768

Yeah, my math is bad since each TMU isn't trilinear capable for the nv30. I guess since I don't have much confidence in their math, I'm being sloppy.

8 (AA) * 8 (AF) * 8 (Trilinear) = 512, looks like you're right.

 

[Lincoln Bauman]

I'm curious whether the driver bugs that manifest themselves on the 9700 are still present with the 9800? (Morrowind Z-Buffer errors, flickering shadows in Mafia, etc?)

Lincoln

 

[Reverend]

No. As I said in the review (do people actually read it?? ) I forced the installation of the 7.84's on to the Radeon 9700 PRO.

I think many folks click on the review index, sees the number of "pages" or sections and jump straight to the benchmarks and/or conclusion.

 

[Dave Baumann]

Yeah. That particular article is averaging about 10 out of the 23 pages (which isn't actually too bad).

 

[Natoma]

Well, the Radeon 9800 Pro w/256MB DDR-II has shown up on Pricewatch. Have you received a sample Dave?

Linky

 

[sireric]

I'm curious whether the driver bugs that manifest themselves on the 9700 are still present with the 9800? (Morrowind Z-Buffer errors, flickering shadows in Mafia, etc?)

Lincoln

Those were ALL fixed in the Cat 3.1 drivers -- Thank you.

 

[Brent]

Did HardOCP list what AF mode they were benching with for the FX board?

I used Balanced in all modes, non AF and with AF

but now looking at it I think I'm realizing that Application for non AF testing may be the best match with r300's default trilinear

i'll definitely be looking at all this now this weekend since i got the gffx again

 

[Randell]

I'm curious whether the driver bugs that manifest themselves on the 9700 are still present with the 9800? (Morrowind Z-Buffer errors, flickering shadows in Mafia, etc?)

Lincoln

both fixed in the Cat 3.1 driver release

 

[Lincoln Bauman]

Those were ALL fixed in the Cat 3.1 drivers

Thank you, I had not been following the ATI driver releases as I do not own a 9700.
I will most likely own a 9800 when they become available however.

Lincoln

 

[Slides]

So, the 9800 is basically an overclocked 9700?

 

[RussSchultz]

It seems like it, though there are apparently some minor efficiency improvements here and there. (See the comparison's clock for clock, the 9800 seems to be a few percent faster).

 

[Ante P]

So, the 9800 is basically an overclocked 9700?

An overclocked 9700 Pro "and then some"

running at identical speeds the 9800 is 5-10 % faster in most tests