The NEXT LAST R600 Rumours & Speculation Thread

Discussion in 'Pre-release GPU Speculation' started by Geo, Mar 1, 2007.

Thread Status:
Not open for further replies.
  1. vertex_shader

    Banned

    Joined:
    Sep 8, 2006
    Messages:
    961
    Likes Received:
    14
    Location:
    Far far away
    I know, but default its not available,the picture with the results are resized, so its grain of salt as everything else :wink:
     
  2. AnarchX

    Veteran

    Joined:
    Apr 19, 2007
    Messages:
    1,559
    Likes Received:
    34
    Yes not in game, but via config-tool:
    [​IMG]
    :wink:
     
  3. Kombatant

    Regular

    Joined:
    May 29, 2003
    Messages:
    639
    Likes Received:
    19
    Location:
    Milton Keynes, UK
    Nice to be passed as an ex-Beyond3D mod btw... I must've worked and fought in the shade all this time :lol:
     
  4. Razor1

    Veteran

    Joined:
    Jul 24, 2004
    Messages:
    4,232
    Likes Received:
    749
    Location:
    NY, NY
    Hmm FEAR at those settings with a GTX gets well over 100 fps, probably around 115 FPS actaully, just checked out FEAR extraction point, I'm getting a good 105ish
     
  5. Brian118

    Newcomer

    Joined:
    Apr 28, 2007
    Messages:
    14
    Likes Received:
    0
    Location:
    Florida
    Aren't those numbers kinda low for FEAR? I thought this guy said that the HD 2900xt "totally smash" the GTX?
     
  6. BrynS

    Regular

    Joined:
    Jul 21, 2003
    Messages:
    407
    Likes Received:
    22
    Location:
    UK
    Slightly OT, but has there been any word/leaks on RD790 + Agena FX samples from Tunis and such samples being seeded to reviewers as a testing platform for R600 or at least RD790 for R6 Crossfire/(QuadFire?)/Stream Computing reviews?
     
  7. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    12,055
    Likes Received:
    3,109
    Location:
    New York
    That was mao5 - different chinese source I think.
     
  8. Brian118

    Newcomer

    Joined:
    Apr 28, 2007
    Messages:
    14
    Likes Received:
    0
    Location:
    Florida
    But they are coming from the same forum. chiphell.com
     
  9. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    12,055
    Likes Received:
    3,109
    Location:
    New York
    Oh right. Just realized mao is a "technical advisor" there. The smileys are funny as hell on that forum :lol:
     
  10. digitalwanderer

    digitalwanderer Dangerously Mirthful
    Legend

    Joined:
    Feb 19, 2002
    Messages:
    18,987
    Likes Received:
    3,529
    Location:
    Winfield, IN USA
    DailyWho? :|
















    ;)

    Me <3 DailyTech, Me <3 Tuan. :)
     
  11. nicolasb

    Regular

    Joined:
    Oct 21, 2006
    Messages:
    421
    Likes Received:
    4
    Okay, I have a question.

    Does anyone know yet for sure if R600 is going to support the same digital-audio-movie-soundtrack-passthrough-via-HDMI feature that we're supposed to get on RV6xx? Initial rumours were that the R600 chip didn't have this feature at all. Later rumours suggested that it did. However, still-later rumours suggest that the board design for R600 cards includes only HDCP-enabled DVI outputs, not HDMI (with HDMI compatibility only via an external dongle). If that's correct then (I think) it means that R600 cards won't be able to output audio over HDMI at all. :(

    Can anyone confirm or deny any of that?
     
  12. Geeforcer

    Geeforcer Harmlessly Evil
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    2,320
    Likes Received:
    525
    Well, not exactly. DT got 79 fps, but they had 16AF as oppoused to 8.
     
  13. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    12,055
    Likes Received:
    3,109
    Location:
    New York
    That's from the XTX article. They got 84fps in their original XT article. And yeah they had 16xAF enabled.
     
  14. Arty

    Arty KEPLER
    Veteran

    Joined:
    Jun 16, 2005
    Messages:
    1,906
    Likes Received:
    55
    Correct me if I'm wrong (which I usually am in most cases)

    2600
    120 SPUs (24 Vec5)
    8 TMUs
    4 ROPs

    2400
    40 SPUs (8 Vec5)
    4 TMUs
    4 ROPs
     
    #3674 Arty, Apr 29, 2007
    Last edited by a moderator: Apr 29, 2007
  15. Kaotik

    Kaotik Drunk Member
    Legend

    Joined:
    Apr 16, 2003
    Messages:
    10,244
    Likes Received:
    4,462
    Location:
    Finland
    I seriously hope 2600 will have 8 ROPs, how can they make the same mistake again? :???:
     
  16. Arty

    Arty KEPLER
    Veteran

    Joined:
    Jun 16, 2005
    Messages:
    1,906
    Likes Received:
    55
    There is something that we have no clue about (yet) which seems to be consuming significant space in all the parts. Looking at the size of RV630, in a simple world should be enough for 'half R600'. However if they can manage to squeeze only 120 SPUs (& 4 ROPs) then I wonder what size would end up as 'half R600' on 65nm.

    P.S. I think UVD as the likely culprit was already discussed some where around here.
     
  17. Kaotik

    Kaotik Drunk Member
    Legend

    Joined:
    Apr 16, 2003
    Messages:
    10,244
    Likes Received:
    4,462
    Location:
    Finland
    Well UVD + Audio departments takes the same space from all 3 chips, does RV610 follow the same "it should have more" pattern?
     
  18. mao5

    Regular

    Joined:
    Apr 14, 2004
    Messages:
    276
    Likes Received:
    5
    I'm coming
     
  19. Jawed

    Legend

    Joined:
    Oct 2, 2004
    Messages:
    11,708
    Likes Received:
    2,132
    Location:
    London
    Instruction Packing

    I've been musing over the organisation of R600's ALU pipeline.

    The diagram shows a fat yellow unit sitting beside 4 skinnier yellow units. It's worth noting that those skinny units are disjoint, they're not a single ALU.

    The implication here is that R600 is capable of issuing five separate scalar instructions per clock, per pixel (or vertex or primitive).

    So, I propose that the fatboy slug is a combined special function and MAD ALU and its weedy cousins are boring old MAD/MUL/ADD ALUs.

    So, it's time to roll out my super-nasty code fragment, for one last time (?) :

    [​IMG]

    Note, that's the revised version with a DP3 at the end, not a RCP as I did originally.​


    [​IMG]


    R600 doesn't get much chance to stretch its legs on this code, because the instruction dependencies are tighter than a slug's arse. But you should be able to see a bit of instruction-component jiggery-pokery going on in cyles 3 and 4.​

    Note I've clocked the pipeline at 800MHz on the basis that in theory there'll be such a thing (whose peak capability is 512GFLOPs)...​

    For comparison, R580:​

    [​IMG]

    And Xenos:​

    [​IMG]

    Even though the FLOPs utilisation of R600 is only marginally higher than R580 (and it's the same as Xenos), you can see that the per-component utilisation is much higher. Summarising utilisation on the basis of components:​
    • R600 = 80%​
    • R580 = 52.5%​
    • Xenos = 70%​
    Jawed​
     
    Geeforcer likes this.
  20. Geeforcer

    Geeforcer Harmlessly Evil
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    2,320
    Likes Received:
    525
    Wow, the I didn't know R600 numbers were THAT good!

    :wink:
     
Loading...
Thread Status:
Not open for further replies.

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...