Asking Tim Sweeney about NVIDIA and more

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

  1. Reverend

    Banned

    Joined:
    Jan 31, 2002
    Messages:
    3,266
    Likes Received:
    24
    Right after witnessing the Half Life 2 benchmarks and presentation, I was left wondering about a couple of things from a developer's point of view, and that was their view of NVIDIA, the GeForceFX family of products as well as the strategic "The Way It's Meant To Be Played" marketing arrangements NVIDIA has struck with a number of developers (including companies like Epic and EIDOS).

    And so I ask Epic's Tim Sweeney. My email is a little long but I'll quote myself in full :

    Valve's Gabe Newell have recently said some things about NVIDIA, their drivers and their GeForceFX products that are not only less than complimentary but also quite damning insofar as NVIDIA driver "optimizations" go.

    While the performance of NVIDIA's GeForceFX products/drivers in Half Life 2 (as presented by Valve) is not surprising to a site like Beyond3D, since we have personally conducted various specific synthetic DX9 tests/benchmarks that show that NVIDIA's GeForceFX is lacking in performance compared to ATI's DX9 offerings, we're wondering what your thoughts are on NVIDIA at the moment.

    Just as we proved ( in our article ) that NVIDIA's GeForceFX is having problems (comparatively speaking, in relation to ATI's DX9 parts) with an actual game utilizing many DX9 features, we'd like your thoughts on how and what you think NVIDIA may have gone "wrong" in their GeForceFX/NV3x architecture in relation to the general gaming development scenario.

    Obviously, we do believe that NVIDIA's inclusion of full 32-bit floating point in their GeForceFX part is "A Good Thing" going forward, but we are also left wondering if this has indeed turned out to be a major mistake on their part in view of ATI's DX9 offerings as well as the "minimum specifications" of floating point precision in DX9.... among other things too, besides precision performance issues.

    What are your thoughts on the general 3D industry landscape as it is currently shaping out to be, not just in terms of hardware architecture differences as a result of architecture decisions taken long before actual hardware became available, but also the various attempts NVIDIA has taken to try and close the obvious DX9 performance gap between their GeForceFX products and those of ATI's DX9 products as well as NVIDIA's business-savvy decision to arrange marketing strategies with various developers and publishers? Is this strategy one to be expected of the 3D and gaming industry at large, and will it ultimately benefit purchasers of games?


    The following is Tim Sweeney's reply where, while he doesn't answer point-by-point, it is quite illuminating :<blockquote>Hi Anthony,

    Well, you could run hardware vendor funded benchmarks and interview people like Gabe and I whose companies have signed big marketing deals with hardware companies. Or you could just write some DirectX9 shader code and draw your own independent conclusions. How's your C++ skills? :)

    -Tim
    </blockquote>What is Tim Sweeney suggesting?
     
  2. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    ha ha that's a good one
     
  3. Reverend

    Banned

    Joined:
    Jan 31, 2002
    Messages:
    3,266
    Likes Received:
    24
    I thought Tim was being quite honest with me with his answer. I think he wanted to tell me this :

    No, Tim didn't email me with the above-quoted... it's my personal translation of his email reply to me!
     
  4. Humus

    Humus Crazy coder
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    3,217
    Likes Received:
    77
    Location:
    Stockholm, Sweden
    I read his reply to mean "Sure, nVidia's hardware is slow, but since we have a marketing deal with nVidia we don't want to come out a state this. But anyone with some DX9 coding skills can prove it for themselves."
     
  5. tEd

    tEd Casual Member
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    2,094
    Likes Received:
    58
    Location:
    switzerland
    scary

    truth is out there but don't expect it from a developer who signed a big marketing deal with a IHV

    :(
     
  6. nelg

    Veteran

    Joined:
    Jan 26, 2003
    Messages:
    1,557
    Likes Received:
    42
    Location:
    Toronto
    I will second that opinion.
     
  7. Babel-17

    Veteran Regular

    Joined:
    Apr 24, 2002
    Messages:
    1,004
    Likes Received:
    245
    Hehe, and let's not forget soundcard support. :)
     
  8. Sabastian

    Regular

    Joined:
    Feb 6, 2002
    Messages:
    991
    Likes Received:
    1
    Location:
    Canada
    Clearly that is the point that Tim implies. Regardless though I think that this plan nvidia is using is going to fail and cost them piles of cash at the same time. Unless these deals are to not only make nvidia cards run faster but make the competitions cards perform worse then they should, nvidia is throwing lots of money out of the window for the failing FX series cards. "The Way its meant to be played" campaign could be a total failure and end up as mere damage control.
     
  9. digitalwanderer

    digitalwanderer Dangerously Mirthful
    Legend

    Joined:
    Feb 19, 2002
    Messages:
    17,276
    Likes Received:
    1,788
    Location:
    Winfield, IN USA
    I thought it was another very revealing answer, and probably a very smart "political" decision on Mr.Sweeney's part.
     
  10. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    Hello members of Beyond3D, i am from 3Dgpu and on reading that, my impressions of what that geezer at Epic said was....

    "I ain't touching that with a 12 foot barge pool....erm....... :oops: Gotta go......(in a Snake from the Simpsons voice), byiiie" (and leaps over a fence half way to Timbuttoo.
     
  11. Anonymous

    Veteran

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

    its timbuktu not timbuttoo, its a state in Africa
     
  12. 3dilettante

    Legend Alpha

    Joined:
    Sep 15, 2003
    Messages:
    8,122
    Likes Received:
    2,873
    Location:
    Well within 3d
    Perhaps what he meant was that the answers are easily found to find once you try something outside of the corporate minefield.

    In a sense, I think he didn't want to get dragged into this mess when he's already navigating some mines already.
     
  13. Anonymous

    Veteran

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

    My geography is pretty sketchy, but isn't Timbuktu a city in Mali? All I really know is that denial isn't just a river in Egypt
     
  14. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    I read it as:

    Don't blindly trust benchmarks that are funded by the IHV's. Don't blindly trust game companies/benchmarks that have signed marketing deals with the IHV's, that includes Epic and Valve. Write your own DX9 shaders and draw your "own independent conclusions from that".
     
  15. CorwinB

    Regular

    Joined:
    May 27, 2003
    Messages:
    274
    Likes Received:
    0
    If TWIMTBP really only is a marketing agreement for Epic, then why has Sweeney not defended his engine and game in the light of the filtering butcheatoptimization (which was UT2K3-specific at first) ? After all, no matter what Nvidia's PR can say, replacing trilinear doesn't really fall as a "mathematically-legit optimization".

    I think TWIMTBP is something more in Epic's case...
     
  16. Reverend

    Banned

    Joined:
    Jan 31, 2002
    Messages:
    3,266
    Likes Received:
    24
    Oh and I left out a translation of mine wrt to this part of Tim's answer :

    There is absolutely no doubt in my mind which benchmark he's talking about. And no, it's not UT2003 or some game where the developer/publisher has a marketing arrangement with a IHV (like NVIDIA's TWIMTBP).

    I know for sure which benchmark Tim meant because, well, I correspond regularly with him.

    Without a doubt, Tim meant Futuremark's benchmark applications.
     
  17. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    It was a nice answer from Tim but Rev, for chrissake, how full of yourself can you be?

    (The frontpage emphasis on the whole of your long-winding, blabbering question that largely got unanswered; and here a smug reference to your correspondence with the man, when elsewhere you can't get precise answers out of him... And elevating yourself seems to be fairly typical of you. And it's pretty vexing for us readers. Are you just insecure? We do believe you are very competent in 3D graphics and a well-connected web journalist. No need to assure it to us all the time, whether it happens in a "light", "jocular" tone or not.)

    Pardon my French.
     
  18. LeStoffer

    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    1,253
    Likes Received:
    13
    Location:
    Land of the 25% VAT
    That since you already know the answer to the DX9 shader performance questions, there is no reason for him to stick out his neck stating the obvious. 8)

    BTW: I wouldn't be surprised if Tim Sweeney saw your mail and was himself asking:

    :wink:
     
  19. Sage

    Sage 13 short of a dozen
    Regular

    Joined:
    Aug 22, 2002
    Messages:
    935
    Likes Received:
    15
    Location:
    Southern Methodist University
    welcome, [K]yle, to the discussion :roll:

    as for my take on what he meant: "Don't trust anyone who recievs a large ammount from IHV's because some of them will be unfairly biased. If you really want to know how the cards perform, write your own shaders because you know whether or not you are being unfairly biased."
     
  20. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    I don't know if this is common knowledge or even very juicy.

    I talked to a developer who works on a TWIMTBP game and he told me that Nvidia devrel wrote a vertex shader for the game and part of the agreement is that the shader does not run on anything other than Nvidia hardware. This is not due to incompatibility, purely run-time vendor detection. I didn't really press the matter, so I have no idea how critical this shader is to performance.
     
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...