RivaTuner v2.20 on the way, major changes are coming with new version

Discussion in 'RivaTuner Advanced Discussion forum' started by Unwinder, Sep 25, 2008.

  1. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    Overclocking or Overdrive overclocking? These are compeltely different things. Also, define your "broken" please. No tab? Tab doesn't function? No hardware detected?
     
  2. mR Yellow

    mR Yellow Ancient Guru

    Messages:
    1,935
    Likes Received:
    0
    GPU:
    Sapphire R9 Fury
    Sorry for not explaining properly. The Overclocking tab is gone. Fan Control tab is visible and work 100%.

    I can OC using Overdrive and ATT but not with RT.
     
  3. harvik780

    harvik780 Member Guru

    Messages:
    195
    Likes Received:
    0
    GPU:
    2x 8800GTS 640M SLI,2x8600GT,8800GT
    I have both my systems having using multi-GPU steup so i am eagerly waiting for this release.
     
  4. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    Investigated the problem:

    ATI broken error reporting system in PowerPlay clock control interfaces in 8.10 hotfix 2. The issue is still not fixed in 8.11 and error reporting is also broken in this version. PowerPlay interfaces themselves work fine, but the driver reports failure codes even on successful PowerPlay clock requests or clock programming calls. It means that all overclocking tools using these inerfaces and controlling the driver's error codes (RivaTuner, AMDGPUClockTool, ATITool and special vendor specific bundle utilities) will stop working starting from 8.10 hotfix 2.
    Fun thing, it still works in ATITrayTools due to lack of error control in the code. That's actually rare example of situation when incomplete code is actually a good thing. :)
    ATI are notified about the problem and I hope that new hotfix will be available soon. It is also possible to create a temporary workaround in RivaTuner (e.g. disable error reporting control if 8.10 hotfix 2 or newer version is detected), however I'm not sure that it worth the efforts. That's clearly a bug in ATI driver and it will be surely fixed in future versions of Catalyst.
     

  5. mR Yellow

    mR Yellow Ancient Guru

    Messages:
    1,935
    Likes Received:
    0
    GPU:
    Sapphire R9 Fury
    Alex, your a genius! Now that makes perfect sense. :)
     
  6. LBJM

    LBJM Guest

    Messages:
    104
    Likes Received:
    8
    GPU:
    latest green team

    any chance you could make it a toggle in the super user tab? because I like using RT launcher for custom OCs with fan speeds. I don't like having CCC installed and RT worked great with the regular 8.10s and I miss it already.

    also you never know if ati will break it again.
     
  7. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    Done. ATIPowerPlayErrorReportingWorkaround entry in the PowerUser tab will allow you to bypass a problem until it is fixed by ATI.
     
  8. mR Yellow

    mR Yellow Ancient Guru

    Messages:
    1,935
    Likes Received:
    0
    GPU:
    Sapphire R9 Fury
    Thanx
     
  9. mikeyakame

    mikeyakame Master Guru

    Messages:
    207
    Likes Received:
    0
    GPU:
    GTX280 729/1566/2700
    Unwinder. You wouldn't happen to have worked out what either the bit mask or representations of the PFB_TIMING0 -> 2 gpu register values are?

    I've worked out a few of the timing values, but there are many possible timings which could be and I'm not sure how exactly to prove which hex byte is which timing. My problem with narrowing them down is the drivers modify the memory timings it appears, so the bios values and timing set only from what I can see are a guideline on how to set the ACT, PRE, READ, WRITE windows and turnarounds.

    PFB_EMRS sets the burst latency and other extended mode registers.
    PFB_MRS sets the CAS latency and a couple of other timings, Write Recovery or WTR are covered by its bitmask.

    I've spent quite some time looking into it, and just not sure what to do to confirm timing values any farther.

    I was also unsure how timing set is detected, and probably the PFB_CFG register holds the timing set select bitmask in the low word. lowest bit is timing set 0, highest bit is timing set 7.

    ie
    NV_PFB_CFG 0x00000800h

    0x0008 = 00001000, selects timing set 4 since 5th bit from lowest bit is masked
    0x0010 = 00010000, selects timing set 3 since 4th bit from lowest bit is masked

    Am I on the right track here?

    0b00010000 represents timing set 3
    0b00001000 represents timings set 4

    and such. just looks like that to me.
     
    Last edited: Nov 16, 2008
  10. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    Cannot help you with that, sorry. Timings registers layout changes really often (almost with any family) and decoding it is rather boring stuff for me, so I've not peeked in these regs for a while.
     

  11. mikeyakame

    mikeyakame Master Guru

    Messages:
    207
    Likes Received:
    0
    GPU:
    GTX280 729/1566/2700
    I understand and respect that. I could decode them since I only need for my current card. Is there any particular way to isolate and prove correctness for which timing has been modified? If I can work that out, the rest is easy

    I've got all the hynix datasheets for the GDDR3 used, so I've figured out roughly what may be what. I'm just not certain because of the driver appearing to select its own tighter values and varies between sometimes one driver to the next.
     
    Last edited: Nov 16, 2008
  12. d1fferent

    d1fferent Maha Guru

    Messages:
    1,425
    Likes Received:
    0
    GPU:
    N/A
    Unwinder

    Any beta avaliable?
     
  13. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    No. I don't provide betas just for fun, sorry.
     
  14. Ray Adams

    Ray Adams Guest

    Messages:
    4,914
    Likes Received:
    0
    GPU:
    ATI HD6950
    What was the problem exactly?
     
  15. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    PowerPlay ExtEscapes (0xC080xx) no longer return correct error codes and always return zero on both succeded and failed calls. Normally it is supposed to be equal to 0xC00000 for success and not equal to 0xC00000 for failure.
     

  16. Ray Adams

    Ray Adams Guest

    Messages:
    4,914
    Likes Received:
    0
    GPU:
    ATI HD6950
    Strange, but I do have error checking.
    Code:
    function GetPMCallResult(returnValue: Integer): Integer;
    begin
        if (returnValue <> $c00000) then
            begin
                Result := returnValue;
                exit
            end;
        begin
            Result := 0;
            exit
        end
    end;
    
     
  17. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    Weird. So does it work in ATT or not? Probably mR Yellow just confused regular overclocking module with Overdrive 5 overclocking module?
     
  18. Ray Adams

    Ray Adams Guest

    Messages:
    4,914
    Likes Received:
    0
    GPU:
    ATI HD6950
    Overclocking works. Didn't test OverDrive. :3eyes:
    Currently I don't have access to ICQ to contact you.
    Lets talk today evening. Okay? :)
     
  19. Unwinder

    Unwinder Ancient Guru Staff Member

    Messages:
    17,083
    Likes Received:
    6,567
    Well, 180.xx WHQL is now online, I guess that I'll need a few hours to ensure that 2.20 is compatible with them then we'll publicly release it at last.
     
  20. ZeW0 ToleraNce

    ZeW0 ToleraNce Master Guru

    Messages:
    360
    Likes Received:
    0
    GPU:
    MSI R5770
    THANK YOU UNWINDER!! U DA MAN!
    I've been waiting to see this post since forever :D
     

Share This Page