Page 1 of 2 12 LastLast
Results 1 to 25 of 30

Thread: Tweaking the 8800 shader..

  1. #1
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309

    Tweaking the 8800 shader..

    from what I understand, the default shader speed on 8800 cards is 1200 for gts and 1350 for gtx models

    using nbitor and nvflash, I tested the effect of increaing this to 1450 on a 8800 gtx

    system setup is as follows

    P5B Deluxe wifi
    E6600 running at 400 x 9 w/ 4/5 mem ratio

    xfx 8800 gtx running at default speeds and default voltage

    results in 3dm06

    sm2 = 4830
    sm3 = 4926

    score 11,221

    with shader at 1450

    sm2 = 4921
    sm3 = 5031

    score 11,401

    aquamark

    default 201,602 ran 3 times in a row with same result

    w/ shader at 1450 201,727...202,088...201,962 (3 times in a row)

    I don't believe that rivatune or atitool touches the shader clock

    So , please check it out and let us know your results

    from what I understand, 1500 is the limit (as per Nvidea), but who knows?


    Possibly the lower clocked gts can benefit more from this!!...
    Last edited by rjw; 11-28-2006 at 10:23 AM.


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  2. #2
    Xtreme X.I.P.
    Join Date
    Aug 2004
    Location
    Chile
    Posts
    4,151
    i believe you mixed the numbers with shaders at 1450 in sm3.

    looks interesting, can you push this further? can you try with OC on the vga?

  3. #3
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    Here is something to look:

    Click image for larger version. 

Name:	8800GTS clocks.JPG 
Views:	4098 
Size:	76.8 KB 
ID:	53371

    When you overclock your core, shader clock increrases too.

    I really think Nibitor does't work at the moment. (I will check it later tonight)
    You are as good as your samples are!

  4. #4
    Xtreme Member
    Join Date
    Jun 2006
    Location
    Hollanda
    Posts
    374
    sm3 = 7926 = 4926?? lol

  5. #5
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    Quote Originally Posted by metro.cl
    i believe you mixed the numbers with shaders at 1450 in sm3.

    looks interesting, can you push this further? can you try with OC on the vga?
    oops just a typo...but I correctd it

    I will be pushing this a lot further, but I was hoping that others would confirm the rivatune, ntune, atitool part for us

    i will have results later or tomorrow, ...awaiting the arrival of a qx6700 and replacement of my no boot evga 680i


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  6. #6
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    Nibitor does work for the shader

    I just tested it

    It may be an option if shader needs to be adjusted separately from gpu clock, no?


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  7. #7
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    Quote Originally Posted by lookmomnobrains
    sm3 = 7926 = 4926?? lol
    too much going on...yes please shoot me


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  8. #8
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    I just modified bios. I used 1296MHz shader speed (default 1188 + 96MHz). Have to test this more.

    Click image for larger version. 

Name:	shader kellotus.JPG 
Views:	3778 
Size:	109.7 KB 
ID:	53372
    You are as good as your samples are!

  9. #9
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    for anyone interested, here is a link to my post at another forum , where Nibitor and nvtweak is available

    http://www.mvktech.net/component/opt...d,40568/#40568


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  10. #10
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    I have to believe my own eyes

    1680x1050 3DMark06 (4AA/16AF, Driver quality high)

    Default shader clock 1188 (core 684MHz real)
    Click image for larger version. 

Name:	1680x1050 4aa_16af 24_7clocks.JPG 
Views:	3818 
Size:	168.5 KB 
ID:	53373

    Default shader clock 1296 (core 684MHz real)
    Click image for larger version. 

Name:	1680x1050 4aa_16af.JPG 
Views:	3793 
Size:	194.5 KB 
ID:	53374

    Noticeable difference with high resolution.
    You are as good as your samples are!

  11. #11
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    I also did another test ocing using riva, but with the base shader set to oem and then 1450 in the bios

    here are the results

    675/1000/1350

    aqua 206,856

    w/1450 bios 206,988

    3dm06 sm2= 5570 sm3= 5678

    w/1450 shader bios

    sm2= 5673 sm3 = 5765

    1 question arises...when ocing using riva, does it scale from the base settings...some think that it does oc the shader as well.

    so are we simply ocing on top of the base 1450

    This brings up an intereting question for the hard core benchmark crowd, and that is what are the gpu/shader limits and now that we can adjust separately, maybe there is more room to tweak

    perhaps setting the default core lower and shader higher or visa versa would allow using just riva to then tweak as a whole

    If this holds true, then we need to know the limits of the gpu and shader, so that one doesn't crash the system while the other still has headroom

    any thoughts?

    on another note....aqua is doing something strange

    I ran the test 3 times without exiting and the scores got progressively lower.

    I exited completely and ran it again and the score went up, but not as high as the first run

    So I used riva to return oc settings back to oem , then back to 675/1000

    Ran aqua again and the score was back up to the highest achieved.

    Any comments on why?


    Can some others please pick up the ball from here and post their findings


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  12. #12
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    Don't run aquamark. It doesn't tell what your GTX/GTS can do.

    3DMark06 with high resolutions is much better.

    Did you look that table I posted earlier. When you raise your core clock, shader clock always rise in certain point.
    You are as good as your samples are!

  13. #13
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    Quote Originally Posted by SF3D
    Don't run aquamark. It doesn't tell what your GTX/GTS can do.

    3DMark06 with high resolutions is much better.

    Did you look that table I posted earlier. When you raise your core clock, shader clock always rise in certain point.
    yes that is interesting

    I can't seem to get the hardware monito in riva to work with my setup

    Do I need to change a setting somewhere? I am using 2.0 candidate 16.1 with forcevga device set at 92


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  14. #14
    Xtreme Addict
    Join Date
    May 2005
    Location
    Sugar Land, TX
    Posts
    1,418
    Yes. I can't overclock my CPU right now and this is what I get in AM3..

    8800GTX @ 575/900 - 1482xx
    8800GTX @ 660/1062 - 1482xx

  15. #15
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    Unwinder released Rivatuner 16.2 with full 8800*** support.

    http://downloads.guru3d.com
    You are as good as your samples are!

  16. #16
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    Quote Originally Posted by SF3D
    Unwinder released Rivatuner 16.2 with full 8800*** support.

    http://downloads.guru3d.com

    thanks...I was just there, but didn't spot it...


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  17. #17
    Xtreme Mentor
    Join Date
    Nov 2005
    Location
    Devon
    Posts
    3,437
    Good job guys!
    Wish I have 8800
    RiG1: Ryzen 7 1700 @4.0GHz 1.39V, Asus X370 Prime, G.Skill RipJaws 2x8GB 3200MHz CL14 Samsung B-die, TuL Vega 56 Stock, Samsung SS805 100GB SLC SDD (OS Drive) + 512GB Evo 850 SSD (2nd OS Drive) + 3TB Seagate + 1TB Seagate, BeQuiet PowerZone 1000W

    RiG2: HTPC AMD A10-7850K APU, 2x8GB Kingstone HyperX 2400C12, AsRock FM2A88M Extreme4+, 128GB SSD + 640GB Samsung 7200, LG Blu-ray Recorder, Thermaltake BACH, Hiper 4M880 880W PSU

    SmartPhone Samsung Galaxy S7 EDGE
    XBONE paired with 55'' Samsung LED 3D TV

  18. #18
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    This doesn't increase performance in games. F.E.A.R and Oblivion run worse with increased shader speed.

    Well, there is a good reason why they are not clocked any higher by Nvidia.

    This can improve 06 score when trying to break somebodys WR

    I flashed original bios back.
    You are as good as your samples are!

  19. #19
    Xtreme 3DTeam Member
    Join Date
    Oct 2006
    Posts
    309
    try lowering it....you may be able to bump your gpu clock up a bit and keep the shader under control

    it's all fairly new, but the facts will come out shortly, I am sure

    Just some food for thought


    E6600 week 28, Asus P5B vanilla, 4.64Ghz,P5B vanilla w/ Vmods, vapochill and other stuff, Super Pi 1M = 11.0000s

    QX6700 on EVGA 680i w/ twin 8800 GTX's SLI

    Second Place hwbot 3DMark06 Sli 20999 (12/2/06 till ...???)

  20. #20
    Xtreme Mentor
    Join Date
    Jul 2004
    Location
    Ontario
    Posts
    2,780
    I wonder why faster is worse in games? Odd...
    Silverstone Temjin TJ-09BW w/ Silverstone DA750
    Asus P8P67
    2600K w/ Thermalright Venomous X Black w/ Sanyo Denki San Ace 109R1212H1011
    8GB G.Skill DDR-1600 7-8-7-24
    Gigabyte GTX 460 1G
    Modded Creative X-Fi Fatal1ty w/ Klipsch Promedia 2.1
    1 X 120GB OCZ Vertex
    1 X 300GB WD Velociraptor HLFS
    1 X Hitachi 7K1000 1TB
    Pioneer DVR-216L DVD-RW
    Windows 7 Ultimate 64


    Quote Originally Posted by alexio View Post
    From the hip and aim at the kitchen if she doesn't approve your purchases. She'll know better next time.

  21. #21
    Xtreme Addict
    Join Date
    Jan 2005
    Location
    Bay Area, CA
    Posts
    1,331
    Quote Originally Posted by cantankerous
    I wonder why faster is worse in games? Odd...
    An increase in error rate would slow things down.

  22. #22
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    There is a lot of lag in games. In oblivion ~5-10 fps lower.

    I think, that when streaming processor (shader) speed is so much higher than default, it does not get enough input data from lumenex engine (core).

    Look this page:
    http://www.xbitlabs.com/articles/vid.../gb8800_5.html

    I don't know much about these, so someone with more knowledge could clear things up.
    You are as good as your samples are!

  23. #23
    k|ngp|n/Sham my brothers
    Join Date
    Aug 2002
    Location
    Athens---Hellas
    Posts
    5,693
    Try to see the XTAL's (Crystal oscilator on the PCB) frequency and then add THIS frequency as much times as you wish on the default shader frequency: Default shader frequency + XTAL's frequency x 1 (or x 2 ,x 3, x 4, etc.) .....
    INTEL PWA FOR EVER

    Dr. Who my arss...

    .........



  24. #24
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    hipro5

    Why? What is the point in that?
    You are as good as your samples are!

  25. #25
    k|ngp|n/Sham my brothers
    Join Date
    Aug 2002
    Location
    Athens---Hellas
    Posts
    5,693
    ALL frequencies (core, rams, etc) have a reference frequency and they are "based" on that...
    INTEL PWA FOR EVER

    Dr. Who my arss...

    .........



Page 1 of 2 12 LastLast

Bookmarks

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •