How to boost your Doom3 performance by 40% on ATI hardware

Discussion in 'PC Gaming' started by Humus, Aug 8, 2004.

  1. Florin

    Florin Merrily dodgy
    Veteran Subscriber

    Joined:
    Aug 27, 2003
    Messages:
    1,660
    Likes Received:
    251
    Location:
    The colonies
    No automatic replacement by the driver but a manual one by the user. I think the difference is clear to most people. What would you prefer people call the new code instead of 'replacement'?
     
  2. Sxotty

    Legend Veteran

    Joined:
    Dec 11, 2002
    Messages:
    5,091
    Likes Received:
    453
    Location:
    PA USA
    BTW that plasma lights thing sounds awesome I bet it would look freakin amazing, hopefully the nv50 will be able to handl it just think how cool a battle scene would look in multiplayer or something...
     
  3. LeGreg

    Newcomer

    Joined:
    Nov 1, 2003
    Messages:
    239
    Likes Received:
    3
    you'll also need to modify the actual code of the game, plus recreate the content, plus this would invalidate all beta testing of the levels. So it's a bit heavier than just a "mathematically" equivalent shader replacement.

    Knowing ATI they would probably have lobbied much earlier for code modification and a specific x800 path, wouldn't they ? Or could they not because of timing issues ?
     
  4. Ruined

    Regular

    Joined:
    Sep 20, 2003
    Messages:
    317
    Likes Received:
    4
    Not crapping at all, if you note my original post simply stated that from the screenshots and evidence of artifacting presented here, it seems reducing filtering quality would be a better IQ/performance tradeoff than this patch. Many others responded emotionally to this post, which I then debated. I have no problem with this thread going forward.

    If you are talking about my comments re: this shader replacement code being approved by ati, IMO either one of two things is happening:

    A) Humus did not get this approved by a supervisor and is taking a moderate risk by releasing shader replacement code which appears to cause artifacts for ATI cards & Doom3 to a public forum.

    B) Humus got the green light to unofficially/on his own time work on a tweak like this from ATI. If this was the case, he probably could not and would not admit if this was the case.

    Either way its fine though, I was just debating some thoughts of others and putting forth my own. You never know, maybe at some point this shader code will result in a tangible performance gain with little IQ loss.
     
  5. Cryect

    Regular

    Joined:
    Jul 4, 2004
    Messages:
    674
    Likes Received:
    0
    Hehe reduced radii should help it run faster if Doom 3 uses dynamic branching (either from PS3.0 or using Humus's dynamic branching or potentially just manually by checking if all vertices from a triangle are within the radius)

    Would be nice if we could allow projectiles to not have shadows from their light for rapid fire guns (kinda annoying not having my flamethrower emitting light)..
     
  6. jvd

    jvd
    Banned

    Joined:
    Feb 13, 2002
    Messages:
    12,724
    Likes Received:
    9
    Location:
    new jersey
    So what do you call this


    Thats thread craping and massive flaming if i've ever seen it before
     
  7. russo121

    Regular

    Joined:
    Aug 27, 2003
    Messages:
    283
    Likes Received:
    4
    1st one:

    2nd one:

    3rd one:

    Sorry this is off-topic but I want to say: ruined GET LOST! :evil:

    You are so worried about Humus job, don't you? For at least 3 times you showed concern..... I think Freud explains that..... As you are fuc... good why don't you get a job at Nvidia?

    Sorry to the other people trying to do something usefull in this thread.
     
  8. Ruined

    Regular

    Joined:
    Sep 20, 2003
    Messages:
    317
    Likes Received:
    4
    Um, if you read back, I was responding to a post by supertoker which stated:
    The above was my thoughts on his post, which is obvious as I quoted his post in my reply to it.
     
  9. Guden Oden

    Guden Oden Senior Member
    Legend

    Joined:
    Dec 20, 2003
    Messages:
    6,201
    Likes Received:
    91
    Dude, you need to go BACK to hardforum whence you CAME, or at least understand that this is not any kind of "patch", or "release" or "shader code replacement" or whatever other bullshit name you've called it by. It's a quick-and-dirty hack made by Humus in his spare time that was never advertised as being the end-all, be-all fix to whatever performance issues people percieve ATI cards have (as if 50fps at ultra quality 1600 res with 4xAA is not enough to play the game... :roll:). GEDDIT?

    No, many others pointed out to you you're completely wrong when you try to label this as an official patch/release that intentionally sacrifices IQ for performance.

    If you actually had stopped your trolling for a second and read the entire thread, you'd understood that this hack actually ADDS workload to the card, but the game still ends up running faster due to hardware differences between NV and ATi. The white spots have nothing whatsoever to do with the performance increase, but again, you'd known that if you'd bothered to read the thread.
     
  10. Ruined

    Regular

    Joined:
    Sep 20, 2003
    Messages:
    317
    Likes Received:
    4
    It is interesting to consider his position of employment at ATI and his recent demos/code releases, yes.
     
  11. jpaana

    Newcomer

    Joined:
    Jul 31, 2002
    Messages:
    154
    Likes Received:
    2
    Location:
    Tampere, Finland
    Replacing the POW with POW_SAT (or the previous DP3 with DP3_SAT which might be a more logical place considering what the code is doing, but anyway) seems to fix the white pixel problem. To really work it would need some code from ID to pass the wanted specular exponent in constant as mentioned before. The white pixels were a result of de-normalization in the normal map, which is made worse by having the map compressed (thus looking worse on medium and low quality settings than high), and pixels getting over 1.0 bright highlights in corner cases. The texture lookup does saturation (clamping texture coordinates) as well so these should be equal in function, assuming the specular is just a power function as it seems to be. If any, the math version does the operation more accurately and thus might show aliasing in the normal map more visibly.

    Interestingly the shader also uses cubemap for normalizing the light vector for diffuse lighting but math for normalizing half vector for specular lighting. The reason probably is that the code seems to be geared towards cards that have lower math performance (read: FX cards) and trades texture lookups for math ops. For specular the lookup is simply too inaccurate and causes blobby looking highlights so it has to be done with math. In retrospect this is kind of sad and having a separate nv3x path that does texture lookups and a proper math using ARB2 path would have been better for both current ATI (and NV40) cards and even better for future cards where math power is growing much faster than texture bandwidth.
     
  12. Ruined

    Regular

    Joined:
    Sep 20, 2003
    Messages:
    317
    Likes Received:
    4
    Ah, so then ATI cards have no performance issues, but this "hack" is needed to change public perception even though it reduces IQ? Not really sure what your point is in this paragraph.

    Not at all official, no. I was ruminating that it might be an unofficial release due to Humus' position in the company. And it has been shown that it causes artifacting and IQ issues.

    I did read the thread, and I even read how the artifacting could be reduced. But this is still a degradation in IQ. It is certainly an interesting technique, maybe it could be more refined in the future with less IQ loss.
     
  13. digitalwanderer

    digitalwanderer Dangerously Mirthful
    Legend

    Joined:
    Feb 19, 2002
    Messages:
    17,656
    Likes Received:
    2,145
    Location:
    Winfield, IN USA
    Ruined, if you want to 'discuss' this any further with me please do so at EB...I'm allowed to tell you my feelings unhindered over there.
     
  14. Neeyik

    Neeyik Homo ergaster
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    1,231
    Likes Received:
    45
    Location:
    Cumbria, UK
    Can people stick to the topic at hand, please? I'm sorely tempted to do some serious pruning in here...
     
  15. AlphaWolf

    AlphaWolf Specious Misanthrope
    Legend

    Joined:
    May 28, 2003
    Messages:
    9,050
    Likes Received:
    1,129
    Location:
    Treading Water
    Please do. This thread now has 5 pages of useless Ruined posts and responses.
     
  16. Dave Baumann

    Dave Baumann Gamerscore Wh...
    Moderator Legend

    Joined:
    Jan 29, 2002
    Messages:
    14,081
    Likes Received:
    651
    Location:
    O Canada!
    And this is what I find curious about the situation, since that appeared to be what was happening until a little while ago, as quoted earlier in the thread. Perhaps we'll see an "ARB3" path! ;)
     
  17. clicknext

    Newcomer

    Joined:
    Aug 8, 2004
    Messages:
    6
    Likes Received:
    0
    In an attempt to post something on topic, I'm going to post my rather interesting results.

    System:
    Athlon XP 2.2Ghz
    512MB RAM 166Mhz
    Asus A7N8X-E Deluxe 2.0
    ATI All In Wonder 9600 Pro (500/350Mhz)

    All tests are done at 800x600 because higher resolutions significantly lower my performance. All tests are first run, after restarting Doom3. Timedemo1.


    ---

    Normal shader, no AF, medium:
    38 FPS

    Normal shader, 8XAF, medium:
    34 FPS

    Normal shader, no AF, high:
    29 FPS

    ---

    Humus/Demirug shader, no AF, medium:
    37 FPS

    Humus/Demirug shader, 8XAF, medium:
    38 FPS

    Humus/Demirug shader, 16XAF, medium:
    40 FPS

    Humus/Demirug shader, 8XAF, high:
    39 FPS

    Humus/Demirug shader, 16XAF, high:
    40 FPS

    ---

    The results sure aren't in my ability to explain. Anyone know why this would happen? I'm quite happy with these results on my 9600Pro. ^_^
     
  18. tEd

    tEd Casual Member
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    2,098
    Likes Received:
    64
    Location:
    switzerland
    well i guess we'll see at least a gslang path in the near future.
     
  19. Star_Hunter

    Newcomer

    Joined:
    Jan 21, 2003
    Messages:
    77
    Likes Received:
    0
    Location:
    Michigan
    If thats the only difference I don't see why this path can't be added relatively quickly.
     
  20. bugsbunny

    Newcomer

    Joined:
    Aug 8, 2004
    Messages:
    1
    Likes Received:
    0
    Hi I used the tweak but I got the white spots and lines.
    I changed the 16 to a 4
    PARAM specExp = { 4, 0, 0, 0 };
    And it solved all problems as far as I can tell.
     
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...