J'accuse! PCPerspective points the finger of shame at ATI

Discussion in '3D Hardware, Software & Output Devices' started by Geo, Nov 2, 2006.

  1. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    213
    Location:
    Uffda-land
    http://www.pcper.com/index.php#NewsID-3049

    They seem to be saying that ATI is downgrading 4xaa to 2xaa in the driver in OGL apps, or possibly just D3 engine apps.

    Well, make up your mind, Ryan. . .are they "incriminating" or is "accidental bug" a possibility?

    Edit:. . . .err, at AMD? :)
     
  2. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    213
    Location:
    Uffda-land
    Come to think of it, a colleague of mine recently wrote a little OGL proggie that reports AA sample positions, and had me test it for him on my X1900XT. It found 6 positions when the CCC was set to 6x.

    But that was the 6.5 drivers, as I'm running FAH on my X1900 and 6.5 is better for that.

    Edit: Errrr, tho I just realized that PCPer is saying that it is only happening at 2560x1600. . .and my test was run at 1920x1200.
     
  3. Sobek

    Sobek Locally Operating
    Veteran

    Joined:
    Dec 17, 2004
    Messages:
    1,774
    Likes Received:
    18
    Location:
    QLD, Australia
    I'd almost certainly bet it's a software issue. He doesn't even mention what drivers he tested with..., what CCC/CP settings he was using, what the specs are on the test system he's using. Any number of things could just as well be responsible for that strange delta, but fingerpointing is a stupid thing to do so early (did he fingerpoint? I just assume nowadays that anything related to videocards involves finger pointing :roll: ). I say, test more drivers, reveal the EXACT testing conditions to make replication of the issue easier, test alternate cards too.

    Oh and geo... AMADIT?
     
    #3 Sobek, Nov 2, 2006
    Last edited by a moderator: Nov 2, 2006
  4. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    213
    Location:
    Uffda-land
    Well, he said "incriminating" and that's finger pointing in my book, yes. There is no incrimination without a crime. . .

    Personally, I've learned over the years that it usually doesn't pay to go apeshit on the first report of something like this. For every time you'd be right to do so, there will be several where you'll end up feeling (or should anyway, if you have any sense of fairness) sheepish later. If something really dirty is going on, there will be plenty of time to go apeshit a few days later. :wink:
     
  5. ChrisRay

    ChrisRay <span style="color: rgb(124, 197, 0)">R.I.P. 1983-
    Veteran

    Joined:
    Nov 25, 2002
    Messages:
    2,234
    Likes Received:
    26
    At such an extreme resolution there could perhaps be other reasons too. At super high resolutions Nvidia cuts back 8xS to 4xS ((or has in the past not sure if this still occurs)) due to framebuffer limitations. I'm not really sure if thats relevant to this though.

    Chris
     
  6. ZoinKs!

    Regular

    Joined:
    Nov 23, 2002
    Messages:
    782
    Likes Received:
    13
    Location:
    Waiting for Oblivion
    Seems to me that there used to be an issue with older, 128 meg Radeons at higher res. They'd drop back to either 2x or 4x max even if higher was requested. I don't recall the specifics but I'm sure it happened.
     
  7. karlotta

    karlotta pifft
    Veteran

    Joined:
    Jun 7, 2003
    Messages:
    1,292
    Likes Received:
    10
    Location:
    oregon
    And people say you cant see jaggies at highres.... One just looks brighter. Maybe "one" just wants a g80 to review....
     
  8. 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
    I think it's then just a case of this PCPer not being able to think very hard. Note the X dimension? I bet that because it is >2Ki it goes over some critical internal hardware limit.
     
  9. gybe

    Newcomer

    Joined:
    Feb 26, 2005
    Messages:
    27
    Likes Received:
    0
    Location:
    Montreal, Canada
    I have a 1900 XT, I put my monitor in 1800x1440 (the highest resolution supported by my monitor). Im using the latest Catalyst(6.10) and a simple OpenGL test application that shows the current subpixels pattern. I start the test application in a 512x512 pixels window wih 6x MSAA. I see the position of the 6 samples. Now if I resize the window to a near fullscreen state, my 6 subpixels disapear and I only see 4 subpixels. So I suppose pcper have a similar problem and they fallback from a 4x MSAA to a 2X MSAA at high resolution.
     
  10. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    213
    Location:
    Uffda-land
    Well, if it's a resolution limitation thing only, then why the suggestion that it only happens in OGL? That kind of physical limitation would be API agnostic, wouldn't it?
     
  11. gybe

    Newcomer

    Joined:
    Feb 26, 2005
    Messages:
    27
    Likes Received:
    0
    Location:
    Montreal, Canada
    I suppose it's a bug in the OpenGL driver, maybe somebody can try with a FireGL to see if it also happens in the workstation driver.
     
  12. satein

    Regular

    Joined:
    Aug 17, 2005
    Messages:
    483
    Likes Received:
    21
    Location:
    Sheffield, UK.
    Hi, may the lappy with mobility FireGL V5200 do the job? It's X1600 equivalent. If yes, could you please instruct me how to do it? I will see if I may help or not. The laptop in question is Hp Nw8440.

    Edit: typo
     
  13. pelly

    Newcomer

    Joined:
    Sep 5, 2002
    Messages:
    159
    Likes Received:
    5
    Location:
    San Jose, CA
    Simon F
    Easy there Simon....

    Before you go making statements like that, please be sure to realize what you're saying. Ryan has been in this industry longer than most...He has a degree in Electrical Engineering, has been doing reviews for ages (AMDMB and now PC Per), and the guy has a tenacious appetite for understanding hardware. As the co-founder of PC Perspective and being no stranger to GPU reviews and technology myself, I can attest that this guy is one of the few editors that knows what he's doing, asks the right questions, and posts credible reviews.

    All I'm saying is that comments like that really aren't necessary unless you're out to flame. And if you're out to flame somebody like Ryan, there's bound to be people in his corner...

    Have a good one guys...let's hope ATI chimes in with a response sometime soon!
     
  14. jb

    jb
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    1,636
    Likes Received:
    7
    Actually Pelly,

    from what it sounds like it could be very well running out of memory as those are very high settings. Has he ran them at a lower res to see? If those trends still show, then yes he has found something. But if they don't hold true and we see a big change, they Simon was right an it was a reviewer error. Even the best of us make mistakes.
     
  15. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    213
    Location:
    Uffda-land
    I think most people don't realize that what is often referred to as the "Tunguska event" in Siberia in 1908, that levelled 2,150 square kilometers, was not actually an air-burst meteor, anti-matter, or anything of the kind that is usually speculated. . . it was actually the last time Simon got annoyed enough to engage in a flamewar.

    He's since promised not to ever again, so let's not strain that resolve, please. :smile:
     
  16. PeterAce

    Regular

    Joined:
    Sep 15, 2003
    Messages:
    489
    Likes Received:
    6
    Location:
    UK, Bedfordshire
    Seems likely.

    Hierarchical Z buffer on R520 was 2 Megapixels.

    3 Megapixel for R580 (which 2048x1356 is close to)

    2560x1600 is about 4 megapixels! This is well over the internal Hierarchical Z buffer size of R580, I presume the excess then spills onto regular Z-buffer? Could this cause the problem?

    .... Wavey Dave where art thou?
     
  17. Geeforcer

    Geeforcer Harmlessly Evil
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    2,297
    Likes Received:
    464
    First, I don't think ATI is doing anything here other then implementing the most basic AA fallback once the framebuffer hits a certain size that has been around since the days of Geforce 2.

    One issue that it does raise is whether the reviewers are aware of that - in other words, the benchmark results that are supposedly reporting 2560x1536 @4x/6x AA scores might be invalid. I confess I have not seen many tests run at that resolution with high AA levels, so I don't know how widespread of a problem this is.
     
  18. jb

    jb
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    1,636
    Likes Received:
    7
    One would hope if they were reviewing they should know about this stuff as it has happened many times before with other cards....
     
  19. Mark

    Mark aka Ratchet
    Regular

    Joined:
    Apr 12, 2002
    Messages:
    604
    Likes Received:
    33
    Location:
    Newfoundland, Canada
    If it was a framebuffer issue the same thing would happen on the X1950 Crossfire Ed card. Both it and the X1950 XTX share the same specs.
     
  20. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    10,430
    Likes Received:
    433
    Location:
    New York
    Good point, people seem to be ignoring this tidbit.
     
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...