Page 66 of 87 FirstFirst ... 1656646566676876 ... LastLast
Results 651 to 660 of 869

Thread: 6000 series delivery date

  1. #651
    8-bit overflow
    Join Date
    Sep 2010
    Posts
    310
    Quote Originally Posted by bongo View Post
    Greater than 250mm is possible but would be approaching any lowend Cayman.
    Nah, if Cayman really sits somewhere between 380mm2 and 400mm2 like Charlie suggested, there'll still be a LOT of room for any lowend Cayman to find its cosy niche

    I'd not expect AMD to repeat the Cypress/Juniper scaling anyways. The performance gap between HD 5850 and HD 5770 was just too huge - and the HD 5830 idea didn't exactly prove successful.

    With Juniper effectively being Cypress/2 (even down to the memory bus), Cayman might very well turn out to be specced along the lines of ~ Barts*1.5 (probably using the same 256bit memory bus, though - and hence notably less than 1.5 times Barts in size).

  2. #652
    Quote Originally Posted by Grunt View Post
    Nah, if Cayman really sits somewhere between 380mm2 and 400mm2 like Charlie suggested, there'll still be a LOT of room for any lowend Cayman to find its cosy niche

    I'd not expect AMD to repeat the Cypress/Juniper scaling anyways. The performance gap between HD 5850 and HD 5770 was just too huge - and the HD 5830 idea didn't exactly prove successful.

    With Juniper effectively being Cypress/2 (even down to the memory bus), Cayman might very well turn out to be specced along the lines of ~ Barts*1.5 (probably using the same 256bit memory bus, though - and hence notably less than 1.5 times Barts in size).
    Hey i have been a leading advocate of Barts being better than 50% Cayman, but for me 250mm^2 seems to big for ATI(sorry AMD), but we will see.

  3. #653
    ATI is just building the GPU based on the market segment. They had nothing worthwhile around $200. A heavily crippled cayman XT made to fit that segment is a horrible idea as its power consumption alone requires a beefy cooler and with good yields, its unlikely a lot will end up that crippled like the 5830 is.

    Barts XT at 2/3 of a cayman is a excellent idea. ~5850 in performance, in a smaller die size of ~240mm2 means they can price it to match the competition very well. Barts Pro will be very good for mid-range, something like 20% faster than a 5770, 20% slower than Barts XT, great pricing.

  4. #654
    Quote Originally Posted by bongo View Post
    Hey i have been a leading advocate of Barts being better than 50% Cayman, but for me 250mm^2 seems to big for ATI(sorry AMD), but we will see.
    I'm going to have words with Lars-G?ran if this meme persists
    Open wide. Now choose Grandma's cookies or Grandma's Glock. They're both real tasty.

  5. #655
    One must consider that this new gen is still made using the same process node, there are limitations that must be factored, no matter how hard ATi try to improve the mArch efficiency, you simply can't expect a similar product & price structures like what was applied in Evergreen generation.

    So, a 260-280 mm^2 Bart is very much logical & probable to happen as long it gives the improved performance/watt & performance/die size compared to what was offered in the last gen. While the rearranged shader array saves diesize, the DX 11 improvement doesn't come for free, especially with 256 bit buswidth & higher TDP, i think the chip will be pad limited if the size is smaller than that proposed number.

    This new gen will bring better performance on the same price, or same performance on lower price, or better, lets say:

    Antilles: 699 US$

    Cayman XT: 429 US$, 35% faster than HD 5870 on average
    Cayman Pro: 339 US$, 25% faster than HD 5870 on average, matching GTX 480 512 SP performance wise
    Cayman LE: 269 US$, 15% faster than HD 5870 on average

    Bart XT: 199 US$, ~HD 5850 performance, good OC ability
    Bart Pro: 149 US$, GTX 460 1 GB performance

    Turk XT: 99 US$, Juniper XT performance
    Turk Pro: 79 US$, Juniper Pro performance
    Turk LE: 69 US$, less than Turk Pro for sure, hehehe

    Caicos: 49 US$ and lower, placeholder before Fusion takes reign in low end segment.

    Now, tell me how nVidia will bath in blood facing these adversaries until 28 nm arrives ?

  6. #656
    >intel 4004
    Join Date
    Jun 2009
    Posts
    11,281
    My source says it's Bart XT having the same performance level of GeForce GTX 460. Hm...
    Having too much optimism is not always a good thing.
    Let's keep our optimism/pessimism to a certain extent, shall we?

  7. #657
    Quote Originally Posted by 265586888 View Post
    My source says it's Bart XT having the same performance level of GeForce GTX 460. Hm...
    Having too much optimism is not always a good thing.
    Let's keep our optimism/pessimism to a certain extent, shall we?
    Did the source specify whether it was 460-768 or 460-1024?

  8. #658
    8-bit overflow
    Join Date
    Sep 2010
    Posts
    310
    Quote Originally Posted by 265586888 View Post
    My source says it's Bart XT having the same performance level of GeForce GTX 460. Hm...
    Hmm. Maybe you should make sure whether your source can actually tell a "6" from a "7" then?

  9. #659
    8-bit overflow
    Join Date
    Mar 2010
    Posts
    376
    Quote Originally Posted by iMacmatician View Post
    Did the source specify whether it was 460-768 or 460-1024?
    I don't have any sources, but from a marketing perspective, it would make sense for them to target the performance of an over-clocked gtx460-1MB but build it using less than 250mm2 of silicon and mass market memory.

    They should be able to do that with a part that has slightly over 1000 NI shaders running at 850mhz, with improved tessellation and 256 bit GDDR5 bus @ 1000/4000 Mhz.

  10. #660
    >intel 4004
    Join Date
    Jun 2009
    Posts
    11,281
    Quote Originally Posted by iMacmatician View Post
    Did the source specify whether it was 460-768 or 460-1024?
    If the source refers to a single model number, then you should expect it can tackle all variants.

    Quote Originally Posted by Grunt View Post
    Hmm. Maybe you should make sure whether your source can actually tell a "6" from a "7" then?
    It performs better than a GeForce GTX 460, the same when comparing with a Radeon HD 5830, so what's the problem?

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