Page 111 of 158 FirstFirst ... 1161101109110111112113121 ... LastLast
Results 1,101 to 1,110 of 1571

Thread: Kepler performance

  1. #1101
    Banned
    Join Date
    Jan 2012
    Posts
    185
    Quote Originally Posted by DarthShader View Post
    We'd have to have a GK104 at our disposal to do the tests and those are hard to come by and expensive. So a forum link to a relevant discussion would be best I guess.
    Ok! Btw I'm not gonna make any claims of GK104 performance on such "micro-benchmark" codes as I haven't had any time to play with it yet. The GF104 however I've had plenty of time to fiddle with

  2. #1102
    It looks like Nvidia is making a fairly big deal about the It that is coming. Which kind of suggests its 690 and not 670. Yet more vaporware, but surely to pretty clever marketing effect. AMD has once again been bulldozed by Nvidia's marketing team.

  3. #1103
    Senior Member
    Join Date
    Jun 2010
    Posts
    4,237
    Quote Originally Posted by Martin View Post
    Computerbase.de did some scaling testing for their launch article, and out of the four titles tested, Crysis 2 had the strongest memory clock-performance correlation with about a .5 scaling factor:

    http://www.computerbase.de/artikel/g...taktskalierung
    So with 10% more bandwidth it gains 4.2% more perf
    and with 10% less bandwidth it is 4.7% slower

    7970 with 33% less bandwidth is 12-13% slower.

  4. #1104
    Administrator Copper's Avatar
    Join Date
    Jun 2009
    Posts
    5,419
    Quote Originally Posted by Guild View Post
    It looks like Nvidia is making a fairly big deal about the It that is coming. Which kind of suggests its 690 and not 670. Yet more vaporware, but surely to pretty clever marketing effect. AMD has once again been bulldozed by Nvidia's marketing team.
    I'm not so sure that's clear. I don't know that it really matters that much at the halo level as supply is always tight there.
    Specialization is for insects. - Heinlein
    Laugh dammit, life is too damned short not to laugh
    ------------------------
    I don't care if the sub-pixels are so small I need TWO magnifying glasses to see them. I want a phone that has longer battery life.

    -Q

  5. #1105
    Senior Member
    Join Date
    Jan 2012
    Posts
    771
    Quote Originally Posted by Dahakon View Post
    On a more seriois note however, there is this claim that with more bandwidth GTX680 would seriously fly past HD7970.
    I find that hard to believe. The 680 does alright with the bandwidth that it has, and I'm not sure adding bandwidth necessarily means it'll do better. I'm kind of wondering if part of the reason that the 680 does so well is because of the limited bandwidth. More bandwidth means a larger controller, more transistors, and more data to control. Charlie is always talking about how bad Fermi's MC was and how bad Nvidia is at physical designs. Maybe smaller is better for them, in this case anyway.

    Or maybe I'm completely wrong. But it seems to me that the 680 is as good as it is for what isn't in it as much as what is.

  6. #1106
    Senior Member
    Join Date
    Nov 2011
    Posts
    2,632
    If increasing bandwidth by 50% increased performance by 50%, why isn't Sandy Bridge E with 4 memory channels significantly faster than regular Sandy Bridge with two memory channels?

  7. #1107
    Quote Originally Posted by Copper View Post
    I'm not so sure that's clear. I don't know that it really matters that much at the halo level as supply is always tight there.
    It gives Nvidia the chance to end AMD's 800+ day run of having fastest card though.

  8. #1108
    Senior Member
    Join Date
    Feb 2011
    Posts
    1,822
    Quote Originally Posted by sdlvx View Post
    If increasing bandwidth by 50% increased performance by 50%, why isn't Sandy Bridge E with 4 memory channels significantly faster than regular Sandy Bridge with two memory channels?
    Because you need several cores to saturate even two channels, and most programs aren't multithreaded and/or not memory BW limited.
    Speaking for myself.

  9. #1109
    Quote Originally Posted by James View Post
    It gives Nvidia the chance to end AMD's 800+ day run of having fastest card though.
    Really? Have you seen how bad the 6990 has been performing in alot of more recent reviews? I think most reviews have the 590 the faster card here recently. It really seems that AMD's dual chip cards make a splash with big benches on release and then are not very well supported a year or two out....IMO anyway.

  10. #1110
    Quote Originally Posted by BigDuke View Post
    Really? Have you seen how bad the 6990 has been performing in alot of more recent reviews? I think most reviews have the 590 the faster card here recently. It really seems that AMD's dual chip cards make a splash with big benches on release and then are not very well supported a year or two out....IMO anyway.
    That applies for all of AMD's cards. The recent selection of benchmarks and AAA titles have all been heavily Nvidia centric, Battlefield 3, Skyrim, Batman, Mass Effect 3, you name it, with very few exceptions that are neutral or AMD centric. All credit to Nv's TWIMTBP and driver team here.

    If the 7970 / 680 battle had happened in Q3 2011, 7970 would be winning.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
WordPress Appliance - Powered by TurnKey Linux