R520: 32 PS pipelines? Any truth to this?

Discussion in 'Pre-release GPU Speculation' started by Acert93, Jun 1, 2005.

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

    Veteran

    Joined:
    Jun 19, 2003
    Messages:
    1,001
    Likes Received:
    16
    Location:
    Montreal, Canada
    True, the architecture stayed largely the same. But then you can also argue that R200 was on 0.15u and much less complex architecturally than R420, yet it had a core clock of only 275MHz compared to 520MHz.
     
  2. Xmas

    Xmas Porous
    Veteran Subscriber

    Joined:
    Feb 6, 2002
    Messages:
    3,344
    Likes Received:
    176
    Location:
    On the path to wisdom
    (My bold)
    Well, that's the point. You have to design the signal paths so that the longest one (critical path) can be traversed in less than a clock cycle. E.g. if you target 500MHz, you have to make sure that every unit or pipeline stage takes less than 2ns from clock edge to stable output.

    Asynchronous circuits don't have that problem, because they're driven by their own switching speed, not an external clock. OTOH, they're more complex, and require more FIFOs and synchronization effort to compensate for the varying speed.
     
  3. Bjorn

    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    1,775
    Likes Received:
    1
    Location:
    Luleå, Sweden
    True, but then we're talking about a much older chip also. I don't doubt that we''ll see 300+ million transistor GPU's at 700+ MHz, the question is just when..
     
  4. Hellbinder

    Banned

    Joined:
    Feb 8, 2002
    Messages:
    1,444
    Likes Received:
    12
    Why not?

    not that i think the chip is clocked higher than 700mhz. Looking at ATi's track record it could be more along the lines of 625, or 680 or 715 heh...

    when you look at how far they pushed the clocks on the last generation it shows that they have been building a nack at this kind of thing. about 320m transistors in *excess* of 600mhz with max clocks above the 700 range.

    At any rate.. at 16 "pipelines" ;) it better clock that high or in certain cases its going to get whomped on. just like the 5800 V.s. R300
     
  5. ANova

    Veteran

    Joined:
    Apr 4, 2004
    Messages:
    2,226
    Likes Received:
    10
    Remember Orton emphasized how current architectures were not very efficient.

    Maybe the R520 will be able to keep up with the G70 at "just" 16 pipes and around a 600-650 MHz clock. :wink:
     
  6. Ailuros

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,511
    Likes Received:
    224
    Location:
    Chania
    If quad = quad between the two GPUs, then obviously yes. Exeption being geometry throughput since 8VS@600MHz = 1200MVertices/sec.
     
  7. KimB

    Legend

    Joined:
    May 28, 2002
    Messages:
    12,928
    Likes Received:
    230
    Location:
    Seattle, WA
    No more than nVidia.
     
  8. DegustatoR

    Veteran

    Joined:
    Mar 12, 2002
    Messages:
    3,242
    Likes Received:
    3,405
    Well, considering that RSX (which is more or less a 90nm G70) should be available pretty soon, it may become a situation where 16pp 600-700 MHz R520 will compete with 24pp 500-600 MHz G7x.

    ATI's late. I think they should cancel R520 and go with R580 already and not by the year's end.
     
  9. SugarCoat

    Veteran

    Joined:
    Jul 17, 2005
    Messages:
    2,091
    Likes Received:
    52
    Location:
    State of Illusionism
    Course then you have the dynamic clocks playing head games. I think the jump is 40+ so the G70, especially the out of the box overclock ones, are already operating close to or in excess of 500mhz.

    In the end on the R520, i think its going to come down to ALU performance across the board. Not just a raw number like "16" or "24".
     
  10. Ailuros

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,511
    Likes Received:
    224
    Location:
    Chania
    Yep.
     
  11. Razor1

    Veteran

    Joined:
    Jul 24, 2004
    Messages:
    4,232
    Likes Received:
    749
    Location:
    NY, NY
    I think they are having major problems still getting the clocks to where they want them....


    http://theinq.com/?article=24776

    why are they releasing this card when the r520 line should be out in a couple of months? Wasn't it stated that the midrange cards were going to be released before the high end?

    This could just be more Inq BS or a smoke screen but still. Or ATi could be getting rid of left over stock......
     
  12. Dave Baumann

    Dave Baumann Gamerscore Wh...
    Moderator Legend

    Joined:
    Jan 29, 2002
    Messages:
    14,090
    Likes Received:
    694
    Location:
    O Canada!
    Thats a "dead die" SKU - a way of making a little money of a bunch of otherwise dead cores; unlikely that this has any bearing on the next generation (although it may be an indicator that they are coming as they may want to dump these on the market before the next gen availability).
     
  13. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    12,058
    Likes Received:
    3,116
    Location:
    New York
    That makes sense.
     
  14. KimB

    Legend

    Joined:
    May 28, 2002
    Messages:
    12,928
    Likes Received:
    230
    Location:
    Seattle, WA
    Different price brackets?
     
  15. Arty

    Arty KEPLER
    Veteran

    Joined:
    Jun 16, 2005
    Messages:
    1,906
    Likes Received:
    55
    Is that a hint that RV515 is 8 pp too (just like G72) 8)
     
  16. Dave Baumann

    Dave Baumann Gamerscore Wh...
    Moderator Legend

    Joined:
    Jan 29, 2002
    Messages:
    14,090
    Likes Received:
    694
    Location:
    O Canada!
    Pipelines? 4, 8, 12 - take your pick, who knows these days... ;)
     
  17. ondaedg

    Regular

    Joined:
    Oct 5, 2003
    Messages:
    350
    Likes Received:
    1
    Is it safe to say that Nvidia's yields are good if they can release the GTX first while ATI may be releasing the midrange as "failed" high end parts first? Or am I reading into this wrong?
     
  18. Xmas

    Xmas Porous
    Veteran Subscriber

    Joined:
    Feb 6, 2002
    Messages:
    3,344
    Likes Received:
    176
    Location:
    On the path to wisdom
    They're supposedly going to sell 8-pipeline R430s. Not next-gen parts.
     
  19. KimB

    Legend

    Joined:
    May 28, 2002
    Messages:
    12,928
    Likes Received:
    230
    Location:
    Seattle, WA
    They're also, apparently, operating on different process technologies. Of course, nVidia's yields are sure to be fairly good since the 7800 GTX doesn't seem to be in horribly short supply.
     
  20. Ailuros

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,511
    Likes Received:
    224
    Location:
    Chania
    Isn't the "V" traditionally for low end at ATI?

    I recall the following numbers from internal presentations mentioned here in the past:

    R520= 16/1/1/1
    RV515= 4/1/1/1
    R580= 16/1/3/1
    RV530= 4/1/3/2

    Now if anything RV513 sounds like a budget single quad sollution to me, followed by RV530 based on R580 later on.

    I'd guess that R515 and R530 would be the according codenames for future mainstream parts and yes there 2 or 3 quads make obviously way more sense than only one.
     
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...