AMD: R8xx Speculation

Discussion in 'Architecture and Products' started by Shtal, Jul 19, 2008.

?

How soon will Nvidia respond with GT300 to upcoming ATI-RV870 lineup GPUs

Poll closed Oct 14, 2009.
  1. Within 1 or 2 weeks

    1 vote(s)
    0.6%
  2. Within a month

    5 vote(s)
    3.2%
  3. Within couple months

    28 vote(s)
    18.1%
  4. Very late this year

    52 vote(s)
    33.5%
  5. Not until next year

    69 vote(s)
    44.5%
  1. Malo

    Malo YakTribe Gaming
    Veteran Subscriber

    Joined:
    Feb 9, 2002
    Messages:
    4,469
    Location:
    Pennsylvania
    The 5770/5750 available now on Newegg. Just ordered the Gigabyte version.
     
  2. codedivine

    Regular

    Joined:
    Jan 22, 2009
    Messages:
    271
    Removed.
     
    #4262 codedivine, Oct 13, 2009
    Last edited by a moderator: Oct 13, 2009
  3. EduardoS

    Newcomer

    Joined:
    Nov 8, 2008
    Messages:
    131
    Since DP is a DX11 feature, it should have.

    @ALL
    Anyone have idea why it doesn't always outperform HD4850? All numbers are bigger... Where is the bottleneck of this new card?
     
  4. rpg.314

    Veteran

    Joined:
    Jul 21, 2008
    Messages:
    4,298
    Location:
    /
    AFAIK, DP is an optional DX11 feature.
     
  5. DegustatoR

    Veteran

    Joined:
    Mar 12, 2002
    Messages:
    1,311
    Location:
    msk.ru/spb.ru
    128-bit bus?
     
  6. itsmydamnation

    Regular

    Joined:
    Apr 29, 2007
    Messages:
    934
    Location:
    Australia
    the ram is a lot higher clocked then a 4850 (GDDR3) which is very bandwidth limited. given the rather poor performace of the 5770, it should be + side of a 4870 not - the 5850 is by far and away the best buy of the evergreen family.

    the great overclocking/overvolting seals it from my point of view :twisted:

    PS why do reviewers always do so little on the overclocking side when people have 5870's at 1300mhz core, i want to see scaling with clocks!!!!!
     
  7. wishiknew

    Regular

    Joined:
    May 19, 2004
    Messages:
    329
    I think why the 5770 is so close to the 4850 is because maybe there is lless bandwidth between the L2, L1 and hub. Maybe the L1 is smaller?
     
  8. Blackraven

    Newcomer

    Joined:
    May 23, 2007
    Messages:
    68
    I was just browsing Wikipedia (yeah its reliability is questionnable) but anyways:

    I saw that ATI will be launching their R8xx flagship video cards for the Christmas season this year (mainly the higher-end forms of the 5870).

    It should really be interesting (and I hope Nvidia will step it up considering that they *might* face delays with their upcoming product)

    But yeah, it looks like ATI will be aiming for their flagship video cards for 2009. They're definitely aiming for 3 Teraflops and beyond :D

    :)
     
    #4268 Blackraven, Oct 13, 2009
    Last edited by a moderator: Oct 13, 2009
  9. Kaotik

    Kaotik Drunk Member
    Legend

    Joined:
    Apr 16, 2003
    Messages:
    6,562
    Considering they're at 2.7x TeraFLOPS with HD5870 and X2 version coming up, we'll be looking at over 5 TeraFLOPS theoretical peaks
     
  10. Blackraven

    Newcomer

    Joined:
    May 23, 2007
    Messages:
    68
    Yup the 5870 high-end models (Eyefinity and X2). Designed to make sure that you can run Company of Heroes faster and better than your R7xx (or any older ATI video card)....

    :D
     
  11. no-X

    Veteran

    Joined:
    May 28, 2005
    Messages:
    2,140
    Any idea, why a pair of HD5770 in CrossFire (= 1600 SPs, 80 TMUs, 64 ROPs @ 800MHz) is slightly faster than a single HD5870 (= 1600 SPs, 80 TMUs, 64 ROPs @ 850MHz)? :???:
     
  12. Broken Hope

    Regular

    Joined:
    Jul 13, 2004
    Messages:
    483
    Location:
    England
    Hmm interesting from the new OpenCL beta drivers.

    [edit]

    Also

     
  13. Mintmaster

    Veteran

    Joined:
    Mar 31, 2002
    Messages:
    3,897
    Surely that's rarely the case, but when it is, the obvious answer is triangle setup and other geometry related factors. AFR can potentially double the framerate, and you can see that HD5770 is often faster than half the speed of HD5870.
     
  14. ShaidarHaran

    ShaidarHaran hardware monkey
    Veteran

    Joined:
    Mar 31, 2007
    Messages:
    3,941
    Also it would be interesting to note the minimum framerate in the scenarios wherein CFX HD 5770 >= 5870.
     
  15. Kaotik

    Kaotik Drunk Member
    Legend

    Joined:
    Apr 16, 2003
    Messages:
    6,562
    Little bird which flew in from my window suggested that Hemlock would be called HD5900 series (so skipping X2 monickers this time?)
     
  16. Iron Tiger

    Regular

    Joined:
    Dec 27, 2004
    Messages:
    848
    Location:
    SoCal
    Has it ever been confirmed that Hemlock was going to be the X2?
     
  17. Tim

    Tim
    Regular

    Joined:
    Mar 28, 2003
    Messages:
    874
    Location:
    Denmark
  18. Iron Tiger

    Regular

    Joined:
    Dec 27, 2004
    Messages:
    848
    Location:
    SoCal
  19. elsence

    Newcomer

    Joined:
    Aug 31, 2009
    Messages:
    80
    I saw the reviews, 5770 scores around -10% than what i expected.
    In some games like Far Cry 2 or like Company of Heroes the 5770 is faster than a 4890.
    But overall it seems like -10% of the 4870 1GB perf. (1920X1200 4AA 16AF)

    I saw some results with the memory overclocked to 1,4-1,5GHz and the performance was not what i expected, so either the overclocked GDDR5 brings a lot of errors at this speed or the memory controller isn't so bandwidth limited as many sites are saying (i mean in the sense that if you increase 20-25% the memory bandwidth (5,5-6Gbps) still the 850MHz 5770 isn't faster than the 750MHz 4870 overall, but around same speed) or the drivers need work or we missing something or a combination of the above.

    I would prefer $139-$149 price instead of $159 (What is ATI's RV770/RV790 stock situation?) (Also a little NV DX11 competition would be good...)

    Probably with newer games (H1 2010) the performance difference with 4870 1GB will be smaller (GPU speed increases in a higher tempo than what memory bandwidth (mem speed*mem con width) increases and the GPU speed increase after DX9 era has much faster SP increasement tempo than ROP increasement tempo)

    Now that we know that juniper is 1040 million transistors and 170mm2 we can say that probably Cypress (2150 million transistors and 334mm2) has a lot of redundancy transistors. (24SIMDS?)
     
    #4279 elsence, Oct 13, 2009
    Last edited by a moderator: Oct 13, 2009
  20. Kaotik

    Kaotik Drunk Member
    Legend

    Joined:
    Apr 16, 2003
    Messages:
    6,562
    Which consists of 2 chips, sized the same as chip next to cypress, while Juniper, Cedar and Redwood have smaller chips next to them.
     

Share This Page

  • About Beyond3D

    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...