R520 Infomania

Discussion in 'Pre-release GPU Speculation' started by russo121, Mar 19, 2005.

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

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,442
    Likes Received:
    181
    Location:
    Chania
    It has, but since I do have a direct link for the forums I don't check the frontpage on a daily basis (good enough for a "sorry excuse"?).
     
  2. hovz

    Regular

    Joined:
    May 10, 2004
    Messages:
    920
    Likes Received:
    0
    r420 is for all intents and purposes 2 r300s on 1 board.
     
  3. Unknown Soldier

    Veteran

    Joined:
    Jul 28, 2002
    Messages:
    2,451
    Likes Received:
    171
    So what you trying to say? The R520 is for all intents and purposes 2 r420s on 1 board?

    :lol:

    US
     
  4. vb

    vb
    Regular

    Joined:
    Jun 5, 2003
    Messages:
    367
    Likes Received:
    2
    Joining this conversation rather late
    (BTW what on earth made resurect this thread?)

    All pipes being equal, 600 MHz x16 pipes is, IMHO, faster than 400 MHz x 24 pipes.

    you get same filrate, 50% faster HyperZ, 50% faster memory controller, 50% faster PCI-E controller, higher latency that can be hidden by increasing amount of cache, and with transistors to spare this should be easy.

    the 24 pipes rumor is in conjunction with decoupled 16 ROP. If true, both variants should have 16 ROP; one running at 400MHz behind a crossbar and one hardwired and running 600MHz. Which one should be faster in FP blending?
     
  5. Ailuros

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,442
    Likes Received:
    181
    Location:
    Chania
    And? If R520 is truly after all an SM3.0 chip, then you could also say that it's a "2*R300+SM3.0" or "R420+SM3.0".

    For all intends and purposes a simple revamp isn't good enough for SM3.0 capabilities and if R520 has after all any similarities to Xenon/R500 then I can't figure out anything else.
     
  6. Ailuros

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,442
    Likes Received:
    181
    Location:
    Chania
    I don't exactly understand the crossbar/hardwired line, since AFAIK all recent high end GPUs have a 4*64 memory controller.

    As for the question: IMHLO the one that handles it's bandwidth more efficiently?
     
  7. jvd

    jvd
    Banned

    Joined:
    Feb 13, 2002
    Messages:
    12,724
    Likes Received:
    9
    Location:
    new jersey
    why do u think there will be a speed diffrence ?
     
  8. MuFu

    MuFu Chief Spastic Baboon
    Veteran

    Joined:
    Jun 12, 2002
    Messages:
    2,258
    Likes Received:
    51
    Location:
    Location, Location with Kirstie Allsopp
    They should launch it at 5:20 on 5/20, lol.
     
  9. neliz

    neliz GIGABYTE Man
    Veteran

    Joined:
    Mar 30, 2005
    Messages:
    4,904
    Likes Received:
    23
    Location:
    In the know
    In a presentation near the Fudo bridge? with the host of the show being Fuad Abazovic..
     
  10. vb

    vb
    Regular

    Joined:
    Jun 5, 2003
    Messages:
    367
    Likes Received:
    2
    I was reffering to the Fragment Crossbar evidenced here, not to the Memory Crossbar and it (question) was rather rethorical. Obviously 16ROPs@600MHz would work better than 16ROPs@400MHz. FP blending is one of the most important advantages of the NV40, and one that it is clearly targeted by R520.

    Second language, sorry

    And ATI cannot have a NV30 event even if they try. NV40 vs R480 is nowhere close to NV28 vs R300.
     
  11. _xxx_

    Banned

    Joined:
    Aug 3, 2004
    Messages:
    5,008
    Likes Received:
    86
    Location:
    Stuttgart, Germany
    Errr...what? :?:
     
  12. Ailuros

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,442
    Likes Received:
    181
    Location:
    Chania
    As with many aspects of different architectures I guess here also applies that ROP=!ROP.

    Are 16 ROPs on R480@540 "working better" than 16 ROPs on NV40@400MHz?
     
  13. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    10,578
    Likes Received:
    622
    Location:
    New York
    I don't quite get this analogy but I don't think, even with all the changes going into R520, ATI will make such (relatively) poor design decisions as Nvidia did with the FX.
     
  14. Ailuros

    Ailuros Epsilon plus three
    Legend Subscriber

    Joined:
    Feb 7, 2002
    Messages:
    9,442
    Likes Received:
    181
    Location:
    Chania
    I think the line in question (built on INQ's rumour) is more related to NV30's low-k 130nm related manufacturing hurdles than the shader core and it's aspects. There were more than one questionable design decisions surrounding NV30 back then.

    Albeit I don't think any IHV is or ever will be immun to poor design decisions, I'd also think that it's unlikely for ATI to have stepped in more than one pitfalss; if there's actually one and there aren't any factors we're not encounting.
     
  15. Xmas

    Xmas Porous
    Veteran Subscriber

    Joined:
    Feb 6, 2002
    Messages:
    3,314
    Likes Received:
    140
    Location:
    On the path to wisdom
    - they have much different Z-culling methods anyway
    - mem/bus interface controller isn't running at core clock, but at mem clock

    ROPs in NV4x (and in R420 as well AFAIK) run at mem clock. So NV40 has them running at 550MHz.
     
  16. russo121

    Regular

    Joined:
    Aug 27, 2003
    Messages:
    283
    Likes Received:
    4
    Yeah.... R520 will be like nv30.... but... it's G70 who needs 225W of power.... Dustbuster 2 :D

    http://www.theinquirer.net/?article=22917

    "...It's simple maths, 75 W from PCIe motherboard connector, and two times 75 from a six pin power connector...."
     
  17. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    10,578
    Likes Received:
    622
    Location:
    New York
    Hopefully that applies only to the highest-end card. If not then they're going to have a hard time convincing people to stick two molexes into sub $400 cards. Or maybe they're lobbying congress for a G70 tax rebate that will offset the extra electricity charges :)
     
  18. vb

    vb
    Regular

    Joined:
    Jun 5, 2003
    Messages:
    367
    Likes Received:
    2
    I was talking about this

    I asume a hypothetical 24 pipes R520 will have 16 ROPs, but I don't have a R520 chip diagram to illustrate

    yes, but if R520 turns out a slow, hot POS that needs a dustbuster ATI can choose just not to release it. R480 fares rather well vs NV40 and they have their SLI coming, which sound a lot better than AFR or SFR. They are fine for at least a year, when R600 is expected.

    See above

    Didn't know that :oops:
     
  19. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    10,578
    Likes Received:
    622
    Location:
    New York
    I don't follow. If R480 is on par with NV40 and ATI decides not to release R520 (which has a snowball's chance in hell of happening) - how are they 'fine' when Nvidia is rumoured to be preparing a refresh of NV40?
     
  20. russo121

    Regular

    Joined:
    Aug 27, 2003
    Messages:
    283
    Likes Received:
    4
    :D I'll ask my electricity company if I can have a connection directly to their generator :lol: :lol: :lol:
     
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...