GeForce FX Tech Q&A

Discussion in 'Beyond3D News' started by Dave Baumann, Jan 31, 2003.

  1. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    Does nobody else see the double standard in the halves of the answer?

    Start with 'of course there's a frame rate impact' and end 'with no drop in frame rate!'
     
  2. Tagrineth

    Tagrineth SNAKES... ON A PLANE
    Veteran

    Joined:
    Feb 14, 2002
    Messages:
    2,512
    Likes Received:
    9
    Location:
    Sunny (boring) Florida
  3. nggalai

    Regular

    Joined:
    Feb 6, 2002
    Messages:
    402
    Likes Received:
    0
    Location:
    /home/rb/Switzerland
    Hehehe. Yeah, my thought exactly. Just look at the last answer.

    *still laughing*

    Considering how thoughtfully we guys phrased the questions to be sent in, this is kind of ridiculous . . .

    ta,
    -Sascha.rb
     
  4. Tahir2

    Veteran

    Joined:
    Feb 7, 2002
    Messages:
    2,978
    Likes Received:
    86
    Location:
    Earth
    Hmm it all seemed so unreal

    Are you sure you talked to a real person and not a robot? I mean that Q&A wouldn't pass the turing test!!! :shock: :cry:
     
  5. Joe DeFuria

    Legend

    Joined:
    Feb 6, 2002
    Messages:
    5,994
    Likes Received:
    70
    That is actually the problem! B3D Knows too much...

    Here's the strategy to use for the next "interview" with any IHV.

    When asking a question, include comments about "how you think it works" (much like was done in the floating point performance question.) However, PURPOSELY understate the performance / feature in question. Give the interviwee a reason to "correct" your understanding. Even better, undersell the COMPETITION'S performance as a reference point.

    For example, how I would word the floating point performance question:

    Question: We know that NV30 supports the ability to produce two FP16 (64-bit) shader instructions in the time it takes to do one FP32 (128-bit instructions), however the number of cycles these take is unclear. We've heard that the FP16 instructions require 8 clocks per instruction to complete, hence FP32 instructions would take 16 clock cycles. In Comparison, we know that the R300 only requires 4 cycles per fp24 instructions. Comments?

    Answer: Your understanding is incorrect. The GeForceFX can actually process a fp32 instruction in 2 cycles!

    Of course, the downside to that approach is that the interviewer looks like he doesn't know wtf he's talking about....but the end result is you might have better luck actually getting answers. ;)
     
  6. Tahir2

    Veteran

    Joined:
    Feb 7, 2002
    Messages:
    2,978
    Likes Received:
    86
    Location:
    Earth
    I think the best method is to talk to someone you personally know at NVIDIA but considering this is NVIDIA's big launch and all that this kind of PR-crap was a 'must' for NVIDIA. I think it shows NVIDIA is hurting from this launch. It was a bit of a mess IMHO and I guess B3D is lucky NVIDIA even looked at these questions at this early stage.
     
  7. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    Joe: I've got an answer to your question.

    "The GFFX is able to usher you in a new empire of cinematic effects and its revolutionary architecture is able to do things you can't even imagine. To achieve those goals, we've proactively optimized our shading architecture and are able to deliver industry leading performance. We believe we are significantly ahead of our competitors, including the R300, in achieving that goal."

    Satisfied? :p

    It's really sad there's no way to truly make PR answer questions. But oh well...


    Uttar
     
  8. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    Did they just add gamma-correct shaders to their spiel after it was revealed that the R300 does gamma-correct FSAA, and it really means nothing?
     
  9. Brimstone

    Brimstone B3D Shockwave Rider
    Veteran

    Joined:
    Feb 8, 2002
    Messages:
    1,835
    Likes Received:
    11
    LOL... :lol:
     
  10. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    they can't stop the bleeding can they...
     
  11. nggalai

    Regular

    Joined:
    Feb 6, 2002
    Messages:
    402
    Likes Received:
    0
    Location:
    /home/rb/Switzerland
    Hi Joe,

    good suggestion, but I wonder whether it works. We (i.e. 3dcenter) got an unofficial interview slot with NV, sent our contact a couple of questions phrased exactly as you proposed--and haven't heard anything from him apart from "sorry, been busy, answering soon," for the past two months or so.

    I wonder whether their interview A.I. has tilted after answering Wavey's questions. ;)

    ta,
    -Sascha.rb
     
  12. lwells

    Newcomer

    Joined:
    Nov 13, 2002
    Messages:
    62
    Likes Received:
    0
    Well, congrats to B3D (Wavey and readers) for the very probing questions -- shame really that the questions were too good for the peron(s) answering.

    My personal favourite was...

    LW.
     
  13. g__day

    Regular

    Joined:
    Jun 22, 2002
    Messages:
    580
    Likes Received:
    1
    Location:
    Sydney Australia
    I was really interested to see if they'd answer my questions about cleaning the cooling system. Seems like they totally ducked the first question and conviently forget to even answer the second question; which was simply how do you clean this sucker when it does get clogged.

    Very crappy reply from NVidia given that my original thread here on heat management and cleaning had a ridiculous tidal wave of views and replies.
     
  14. fresh

    Newcomer

    Joined:
    Mar 5, 2002
    Messages:
    141
    Likes Received:
    0
    No shit, everything's programmable in the pixel shader. Question is, is anybody gonna do it? And the FSAA won't be programmable cause the hw is *supposed* to do it when combining the multisample buffers.

    No gamma corrected FSAA AND no 30bit frame buffer. Two more nails FX's coffin. Not to mention the thing isnt available yet.

    It's stillborn as far as I'm concerned. Bring on the NV35!
     
  15. Ostsol

    Veteran

    Joined:
    Nov 19, 2002
    Messages:
    1,765
    Likes Received:
    0
    Location:
    Edmonton, Alberta, Canada
    From [H]ardOCP regarding NVidia's FSAA:

     
  16. Anonymous

    Veteran

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

    Maybe a sarcasm?
     
  17. WaltC

    Veteran

    Joined:
    Jul 22, 2002
    Messages:
    2,710
    Likes Received:
    8
    Location:
    BelleVue Sanatorium, Billary, NY. Patient privile
    It's a shame for a site to give a bully pulpit to these PR types so that rather than answer the questions they can have the site publicize their meaningless PR drivel instead. I know there doesn't seem to be much one can do when a subject refuses to answer the question and provides propaganda instead, but...

    How do you think it would work if the person was told something like this:

    "If we aren't very clear on the answer you provide to a particular question, or we don't feel it's actually been answered, we'll ask it again and tell you that we aren't clear on the connection between the answer and the question. If you can't make that connection clear then we'll just omit both the question and the answer from the published interview."

    Too many of these guys seem to consider that Internet sites are their personal PR playgrounds to use and abuse at their discretion--and I've never felt that some sites quite understood their importance to these companies. It's very tiring to read Clinton-esque, equivocating double talk when you've asked a simple question that deserves a simple answer.
     
  18. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    i think the interview was awesome. You learn just as much from the unwillingness to answer. i.e. Maybe I wasn't aware of the R300 supporting multi gpu funcitonality , but then I read his F-up'ed reply and alarm bells start going off and it makes me wonder why he is being evasive. I mean how hard is it to same the word one? This might cause someone to look into just out of curiosity.
     
  19. Anonymous

    Veteran

    Joined:
    May 12, 1978
    Messages:
    3,263
    Likes Received:
    0
    [Q]Q: How do you feel the pixel and vertex shader power and the pixel precision of the FX will come into play when the new wave of Direct3D 9 media hits the store shelves?

    A: The CineFX engine is really the “soul of the machineâ€￾ for GeForce FX. It encompasses the VS2.0+ vertex engine, the PS2.0+ pixel shading engine and the studio-quality 128-bit color. The long shader programs and full 128-bit color enable a whole new level of image quality. These features are what make the characters Dawn and Ogre from the GeForce FX demos look so good. We’re going to see that level of quality and that level of detail in future games based on DX9. You’ll also see games take advantage of GeForce FX’s native 64-bit color mode that offers high-precision math, but delivers blazing performance too. [/Q]
    :: From nVidia's Geoff Ballew in an Interview at EliteBastards
    It's implying that 128bit-mode(FP32) is slower than 64bit-mode(FP16).


    [Q]The R300 can run Doom in three different modes: ARB (minimum extensions, no
    specular highlights, no vertex programs), R200 (full featured, almost always
    single pass interaction rendering), ARB2 (floating point fragment shaders,
    minor quality improvements, always single pass).

    The NV30 can run DOOM in five different modes: ARB, NV10 (full featured, five
    rendering passes, no vertex programs), NV20 (full featured, two or three
    rendering passes), NV30 ( full featured, single pass), and ARB2.

    The R200 path has a slight speed advantage over the ARB2 path on the R300, but
    only by a small margin, so it defaults to using the ARB2 path for the quality
    improvements. The NV30 runs the ARB2 path MUCH slower than the NV30 path.
    Half the speed at the moment. This is unfortunate, because when you do an
    exact, apples-to-apples comparison using exactly the same API, the R300 looks
    twice as fast, but when you use the vendor-specific paths, the NV30 wins.

    The reason for this is that ATI does everything at high precision all the
    time, while Nvidia internally supports three different precisions with
    different performances. To make it even more complicated, the exact
    precision that ATI uses is in between the floating point precisions offered by
    Nvidia, so when Nvidia runs fragment programs, they are at a higher precision
    than ATI's, which is some justification for the slower speed. Nvidia assures
    me that there is a lot of room for improving the fragment program performance
    with improved driver compiler technology.[/Q]
    :: From the John Carmack Interview
    When using FP32, FX is at only the half speed of 9700p.


    [Q]We know that NV30 supports the ability to produce two FP16 (64-bit) shader instructions in the time it takes to do one FP32 (128-bit instructions), however the number of cycles these take is unclear. At one launch presentation an NVIDIA representative mentioned that 32bit integer instructions can be done at two per clock, which was twice as fast as an single FP16, which suggests that FP16 instructions operate in one clock cycle, hence FP32 instructions would take two clock cycles. Is this the case?

    We have not disclosed these details of our architecture.[/Q]
    :: From GFFX Tech Q&A about HOS & Shaders by Beyond3D
    nVIDIA says no comment on this.


    Sum) Seems FX FP32 needs 2 clock cycles to me. Am i wrong? I'm no 3d pro. Anyone would correct me plz?
     
  20. Dave H

    Regular

    Joined:
    Jan 21, 2003
    Messages:
    564
    Likes Received:
    0
    That's the current speculation, yes. I should also point out that sustained shader throughput will probably vary depending on what combinations of instructions are being run at the same time, and of course you're always subject to stalls resulting from data dependencies.

    But, as long as Nvidia won't release the details, it will be impossible to tell until someone gets their hands on a card and can code some purely synthetic shader benchmarks aimed at figuring it all out. Conversely, it seems inevitable that once GFfx is truly released, Nvidia will have to release the details so developers will know how to optimize their shaders.
     
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...