Asking Tim Sweeney about NVIDIA and more

Discussion in 'Beyond3D News' started by Reverend, Sep 29, 2003.

  1. jimbob0i0

    Newcomer

    Joined:
    Jul 26, 2003
    Messages:
    115
    Likes Received:
    0
    Location:
    London, UK
    Why is it that all those bashing ATI and B3D and defending NV are posting as guests? :roll: .... semi-rhetorical question for those missing sarcasm there....


    As for this I would accept Reverend's words when he suggests that Tim has a 'dislike' of 3DMark03 for various reasons....

    If Tim suggests coding our own DX9 shaders to test with (good ol' Humus!) then given current knowledge of how the GFFX performs with those it is sensible to consider Rev's 'translation' as a logical way of thinking.

    As for HL2 and Gabe.... well we will have the ability to test for ourselves to see how reliable that 'IHV-sponsored' benchmark was in about 24 hours (or just under ;)
     
  2. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    I can tell who the annomous posters is. :D
     
  3. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    Perhaps it's time for an open source benchmarking suite created by the enthusiast community. Then there want be any doubts about the benchmark favouring certain hardware as the code will be available for all to see.

    No. I don't know what I am talking about and I'd be surprised if there wasn't such a project in existence already.
     
  4. Simon F

    Simon F Tea maker
    Moderator Veteran

    Joined:
    Feb 8, 2002
    Messages:
    4,560
    Likes Received:
    157
    Location:
    In the Island of Sodor, where the steam trains lie
    Perhaps it's less of a issue of HW architecture but more of an issue of poor driver support software? (Just trying to offer up some other possibilities for the lack of performance. <shrug> God knows why I am, given the Anti-Kyro FUD slides that were released).
     
  5. digitalwanderer

    digitalwanderer Dangerously Mirthful
    Legend

    Joined:
    Feb 19, 2002
    Messages:
    17,267
    Likes Received:
    1,783
    Location:
    Winfield, IN USA
    I really did find Mr.Sweeney's statement to basically say, "Find out yourself, trust no one."...which is pretty much what I thought has been going on anyways.

    I think it'll be a lot easier for us to all draw conclusions about Half-life2's performance after the benchmark is released tomorrow, then we'll all be able to see for ourselves what the image quality/speed quality difference between ATi & nVidia is. 8)
     
  6. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    Just to note that this is the best reply ever :D

    Tim is a wise man :wink:
     
  7. Snyder

    Regular

    Joined:
    Feb 10, 2002
    Messages:
    609
    Likes Received:
    8
    Location:
    Vienna, Austria
    I too thought he meant sth. like this, and with just a slight hint of "Please don't drag me into this mess..." :)
     
  8. Tim Murray

    Tim Murray the Windom Earle of mobile SOCs
    Veteran

    Joined:
    May 25, 2003
    Messages:
    3,278
    Likes Received:
    66
    Location:
    Mountain View, CA
    Interesting. A jab at Futuremark rather than Gabe and the HL2 benches first shown at an ATI event? I'm surprised.

    Or maybe I'm not, given some things I've heard recently. I guess the test will be whether or not any NVIDIA drivers are FutureMark-certified.
     
  9. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    and to quote myself, after reading the thread (obviously I am not the guest from page 2 :d ) and is there a thread of how fast Humus's demo runs on the current DX9 cards? (as this is the practical part of the message to be had from Tim) + I think I finally need to register here :oops:
     
  10. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0


    Don't worry, people who really READ your articles know you are objective.

    Those slides are the nvidiots last desperate hope to grab onto something to mudsling ATI and anyone that defends that stand. :roll:

    Want an independent benchmark? Try the new ShaderMark2.0 then...as far as I know, they guy coded it by itself, and if I'm wrong, somebody please tell me.

    Thnx.
     
  11. Tim Murray

    Tim Murray the Windom Earle of mobile SOCs
    Veteran

    Joined:
    May 25, 2003
    Messages:
    3,278
    Likes Received:
    66
    Location:
    Mountain View, CA
    Is ShaderMark 2.0 publicly available now?
     
  12. Humus

    Humus Crazy coder
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    3,217
    Likes Received:
    77
    Location:
    Stockholm, Sweden
    Well, I didn't miss that point, but it seems I didn't comment on that. Gabe is biased as well as Tim since they have marketing deals. Gabe had the facts on his side so he could go public with high performance numbers on ATI and low on nVidia. Tim know very well that he can't prove that the GFFX performs better than the 9800, so he remains silent. Publicly stating anything might hurt the relationsship nVidia, so it's pretty wise to stay out of the discussion. For Gabe on the other hand it had more pluses than minuses to public.
     
  13. Bouncing Zabaglione Bros.

    Legend

    Joined:
    Jun 24, 2003
    Messages:
    6,363
    Likes Received:
    82
    Nvidia have been working on Nv3x drivers for more than a year, and they still can't improve NV3x performance to match R3x0 levels without massive cheating and dropping IQ down to sub-DX9 levels. NV3x is not fixable. Can there be any doubt after all this time and all the evidence we've seen?

    NV3x was designed to give lip service to DX9 while Nvidia flogged cards on the back of Doom 3. Same as always when Nvidia implement new tech - it's unusable for the first couple of generations if you want more than a slideshow, but it gives them a few nice marketing check boxes.
     
  14. Dave Baumann

    Dave Baumann Gamerscore Wh...
    Moderator Legend

    Joined:
    Jan 29, 2002
    Messages:
    14,079
    Likes Received:
    648
    Location:
    O Canada!
    The results from Shadermark clearly show that they are actually making some reasonable performance increases to the shader performance with later drivers, and things no longer seem to be 3 times slower than the R300's, but rather much closer now. I am a little stumped as to why it has taken so long though.
     
  15. Bouncing Zabaglione Bros.

    Legend

    Joined:
    Jun 24, 2003
    Messages:
    6,363
    Likes Received:
    82
    The shadermark thread here http://www.beyond3d.com/forum/viewtopic.php?t=8193&start=20 doesn't make NV35 look very good even in partial precision.
     
  16. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    Oh, for heavens sake

    ... Just write the damn code, will ya?

    Sheesh.
     
  17. Evildeus

    Veteran

    Joined:
    May 24, 2002
    Messages:
    2,657
    Likes Received:
    2
    But much better than 2/3 times slower....
     
  18. Rugor

    Newcomer

    Joined:
    May 27, 2003
    Messages:
    221
    Likes Received:
    0
    It's funny, but people seem to have forgotten one very simple explanation for what they meant by "Influence benchmarks in ATI's favor."

    It could have simply meant to try to push for shader-heavy DX9 benchmarks that write directly to the API as written. Things like telling benchmark coders that PP hints aren't part of DX9 and don't belong in a benchmark. That would be influence but not something I'd worry about.

    Nvidia could reasonably do the same thing if they tried to push for shadow-heavy rather than shader-heavy benchmarks.

    I think it was a piece of simple advice to follow any IHV's standard practice and try to get benchmark coders to work with your product's strengths.
     
  19. Joe DeFuria

    Legend

    Joined:
    Feb 6, 2002
    Messages:
    5,994
    Likes Received:
    70
    Just doing some rough calculations here....

    Based on Shadermark, I would say that the 5900, running with the Det 52.xx drivers, using PS 2.x shaders is finally performing more or less how we would have expected it to, based on it's architecture. (That is, a 4x2 vs. an 8x1).

    Adjusting the results to normalize based on core clock: (That is, CLOCK FOR CLOCK):

    The 5900 Ultra is on average about 55% slower than the 9800 Pro in full floating point precision mode.

    The 5900 Ultra is on average about 40% slower than the 9800 Pro in partial precision mode.

    Just for sake of completeness, (and to avoid accustaions of biased reporting...) the "absolute scores" (not clock for clock adjusted:)

    The 5900 Ultra is on average about 47% slower than the 9800 Pro in full precision.

    The 5900 Ultra is on average about 30% slower than the 9800 Pro in partial precision.


    One big question mark, of course, is the NV35's inability to run several of the ShaderMark tests at all....
     
  20. Bouncing Zabaglione Bros.

    Legend

    Joined:
    Jun 24, 2003
    Messages:
    6,363
    Likes Received:
    82
    Looking at Joe's figures above, the phrase "too little, too late" comes to mind.
     
Loading...

Share This Page

  • About Us

    Beyond3D has been around for over a decade and prides itself on being the best place on the web for in-depth, technically-driven discussion and analysis of 3D graphics hardware. If you love pixels and transistors, you've come to the right place!

    Beyond3D is proudly published by GPU Tools Ltd.
Loading...