From ATI to AMD back to ATI? A Journey in Futility @ [H]

Discussion in 'AMD Flavor' started by FrgMstr, May 27, 2016.

  1. drakken

    drakken [H]ard|Gawd

    Messages:
    1,196
    Joined:
    Aug 19, 2004
    the new architecture is henceforth socket which include an in house tracing etc... with multiple segments or parts. Sounds like it is going back to a north bridge and south bridge or the traces connect to multiple pieces in a new socket. I speak french but they hate when forigners learn the syntax so every six months they take a word that means one thing and assign it a new meaning that if you are not in France you have no idea which words changed unless you read the local papers online. Terminator is socket, mais is with, desormais is hench forth etc... et cela is etc... some of it is funny to translate when you get latin and french mixed in usage.
     
  2. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    details like that would mean that the chances are high, that there is more to it i guess
     
  3. Nenu

    Nenu [H]ardened

    Messages:
    19,090
    Joined:
    Apr 28, 2007
    There is a higher chance there is more to hide.
    Otherwise, why obfuscate and annoy fans.
     
  4. Anarchist4000

    Anarchist4000 [H]ard|Gawd

    Messages:
    1,659
    Joined:
    Jun 10, 2001
    What they're probably looking at is a MCM. With a motherboard transitioning towards interposer for simple analogy. GPUs like a big interconnect, so keeping them integrated or on an interposer saves a lot of pins on a CPU/APU design. A deal with AMD could theoretically cut R&D costs for Intel while providing more options than they currently maintain. Instead of putting a discrete board in the system they'd pick a GPU and stacked RAM to use with their CPU. Any type of processor could be integrated in that fashion given interposer space. Not to mention it avoids the licensing issue which was a significant expense as well. It would be a win-win for both companies. Intel could theoretically scrap their GPU for AMDs while still developing Phi for compute or using Altera based FPGAs.
     
  5. CSI_PC

    CSI_PC 2[H]4U

    Messages:
    2,200
    Joined:
    Apr 3, 2016
    And with the budget Intel has where exactly will that leave AMD in 4 years time when competing against Intel?
    AMD risk killing their CPUs (which ironically they are trying to resurrect into being competitive) and to a lesser extent APUs, if Intel has a broad use of the IP it should not be under-estimated just how much this could come back and bite AMD.

    If the deal is true I think AMD focused too much on Nvidia as their competitor to squeeze them from a broader tech adoption perspective without realising their biggest challenge is growth areas that Intel competes in.
    Their saving may be Intel having product strategy issues and morale with their best engineers (some leaving).

    You see signs of this as they keep focusing on Nvidia in marketing and various presentations, whereas Nvidia keeps focusing on Nvidia since Maxwell.
    Cheers
     
  6. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    is it possible that a sec filling is only necessary if a product is being launched?
    Edit: stupid question, please ignore
     
    Last edited: Jan 30, 2017
  7. Anarchist4000

    Anarchist4000 [H]ard|Gawd

    Messages:
    1,659
    Joined:
    Jun 10, 2001
    This would still take a few years to establish. AMD would get guaranteed licensing income or possibly some royalties from chips. I don't see it being any different than selling discrete AMD GPUs with Intel CPUs. Difference being the form factors have shrunken with technology. MCMs with elements from different vendors. Intel avoids their graphics patent deficit and gets more options. AMD sells more GPUs. Maybe Intel gets HSA working with their CPUs, but it may just be a matter of providing a low cost GPU for their offerings and not accelerating all workloads. Intel has Altera for FPGAs and future datacenter development which is probably where they're aiming. I'm sure they've done the math on the trade-off. Do GPU profits outweigh the potential for lost CPU/APU profits? Especially if Intel continued to develop graphics which already cover a majority of the market.
     
  8. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    You are not getting MCM parts. And specially not with interposers.

    And Intel isn't going to use an AMD GPU. If they for some imaginary reason had to abandon their own line. They would use Nvidia because they match the metrics Intel focuses on.

    Using AMD for GPU would also be a huge step back on video decoding and encoding (Quicksync).
     
  9. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    btw event if there is no deal or license agreement between AMD & Intel. Intel would still need something to protect themselves from patent infringement, right?
     
  10. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    They already are with NVidia.
     
  11. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    but as far as i know that agreement ends in march
     
  12. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    And they need to renew it, AMD or not. They use Nvidia IP in their IGP.
     
  13. Anarchist4000

    Anarchist4000 [H]ard|Gawd

    Messages:
    1,659
    Joined:
    Jun 10, 2001
    So they aren't looking at compact, power efficient systems a couple years down the line? The original rumor was adopting AMD tech, either through IP licensing or direct use. That would only make sense to replace the Nvidia deal which was the result of a legal settlement. As for video encoding/decoding there isn't that much of a difference. In theory the upcoming architectures and HSA would be very effective with video and jumping between scalar and parallel work anyways. By that time we're also looking at FPGAs performing the encode/decode for better codec support and they'd have their Altera IP to use.
     
  14. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    GCN isn't compact and efficient. Its the exact opposite. FPGAs performing the decode? Ye that sounds really cheap.

    You cant replace Nvidia IP with AMD IP unless you start all over.
     
  15. Anarchist4000

    Anarchist4000 [H]ard|Gawd

    Messages:
    1,659
    Joined:
    Jun 10, 2001
    FPGAs are the likely new broadcast standard so codecs can be updated. Not necessarily that much more expensive all things considered. They'll be in every TV and multimedia device soon enough.
     
  16. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    So what will drive the cost down so significantly to make this possible? And will it be an addon chip instead of integrated?
     
  17. Anarchist4000

    Anarchist4000 [H]ard|Gawd

    Messages:
    1,659
    Joined:
    Jun 10, 2001
    For codecs they don't have to be huge, and they'll likely be integrated and discrete. AMD for instance had that magnum FPGA board for DTV, but we haven't seen what they're doing with it. FPGAs have a lot of applications, so no reason they won't be integrated on many chips to replace the existing en/decoders. They could very well be accelerating compression or encryption as a coprocessor, so makes sense lots of them will be around. They're simply more efficient once you get a large variety of codecs and formats needing support. Plus the ability to support new ones so long as they aren't overly complex algorithms.
     
  18. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    The magnum board was a prototype. Using FPGA for development is very common. Pascal, Polaris, Volta, Vega etc is also made this way tho in a different scale. Gsync uses FPGA as well, increasing its cost.
     
  19. CSI_PC

    CSI_PC 2[H]4U

    Messages:
    2,200
    Joined:
    Apr 3, 2016
    Yeah that is a good point, consider Intel will not effing put the eDRAM into their enthusiast CPUs as a L4 cache, or how they went out of their way to save 10c-$1 per CPU when it comes to cooling (context being focus on savings and not cost-absolute performance).
    Cheers
     
  20. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    No Intel-AMD licensing or otherwise deal in AMDs Q4 report and Q1 outlook.
     
  21. CSI_PC

    CSI_PC 2[H]4U

    Messages:
    2,200
    Joined:
    Apr 3, 2016
    Yeah you definitely would expect AMD to comment in these if it was looking like a solid deal.
    CHeers
     
  22. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    thought the same. so, it was a probably wrong
     
    Last edited: Feb 1, 2017
  23. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    Pretty strong words there brother. Still doing some digging on my end, but yes, I expected at least a mention in AMD's call.
     
  24. Meeho

    Meeho [H]ardness Supreme

    Messages:
    4,596
    Joined:
    Aug 16, 2010
  25. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    is the call over? if yes, did no analyst ask about it
     
  26. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    You dont ask about...fiction.
     
  27. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    touché
     
  28. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    If current dollar value is immaterial to each as far as can be seen, then the agreement could very well not be disclosed to stock holders.

    That said, this is what I just got back from one source that did have information on this, but was not the primary source. He said, "....the info is still valid, just not fully consummated..." Still waiting for information from the primary source, but it is not someone that I have easy access to. I will be patient even if you two are not. I see some really good custom titles coming your way soon. ;)
     
  29. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    Guess you better get a better dictionary.

    hoax
    hōks/
    noun
    1. a humorous or malicious deception.
     
  30. Gideon

    Gideon 2[H]4U

    Messages:
    2,437
    Joined:
    Apr 13, 2006
    Shintai thinks he knows how corporations do things, lots of things are last minute in the corporate world especially if one side does not want it public any sooner then it needs to be. Should know for sure in 1 month.
     
  31. Tyns

    Tyns Limp Gawd

    Messages:
    313
    Joined:
    Sep 28, 2010
    Intel's current licensing deal doesn't expire until March. If such a deal is going to be made with AMD, it doesn't surprise me that there hasn't been any mention of it yet as it's not currently needed
     
    FrgMstr likes this.
  32. CSI_PC

    CSI_PC 2[H]4U

    Messages:
    2,200
    Joined:
    Apr 3, 2016
    I think there is/was negotiations but still not agreed between the parties and no guarantee, AMD usually in part of their forward looking statement in those results/conference call would comment they have a cross-licensing deal agreed with a 3rd party (would not name the company yet) and with a rough estimate when coming into effect but that did not happen.
    This would make sense to do especially considering the more down comments they provided for their next quarter estimation.
    I think some in AMD jumped the gun with this leak-rumour.
    Cheers
     
    Last edited: Feb 1, 2017
  33. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    So you are suggesting here that I knowingly and purposely misinformed our readers for either my own amusement or malicious intent?
     
  34. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    Well, it would seem that you might want to edit your statement.
     
  35. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    Guess I am making stuff up that you are posting in thread about an article that I wrote and statements that I made. If you are going to be making statements of fact about our content, I would highly suggest you be ready to defend those statements with proof.
     
  36. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    I do not understand your question.
     
  37. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,781
    Joined:
    May 18, 1997
    No, I did not understand his question, "So what do think about cpc hardware's info then?"
     
  38. noko

    noko [H]ardness Supreme

    Messages:
    4,551
    Joined:
    Apr 14, 2010
    I wouldn't be surprised Intel just buys out Imagination Technologies, use that tech in upcoming Cannonlake. Believe Intel Atom already uses Imagination Technology anyways. Does not mean discussions are not going seriously with AMD even Nvidia and others. If Intel wants to get into more of Apple share, small or cell phone size computing and basically control that market Imagination Technologies seems to be more beneficial to them. Then again AMD maybe has some strong future tech they might want - who knows besides those dealing?
     
  39. Shintai

    Shintai [H]ardness Supreme

    Messages:
    5,691
    Joined:
    Jul 1, 2016
    Intel dumped Imagination Technologies twice, last time foe good.
    http://www.standard.co.uk/business/...fter-intel-dumps-cut-price-stake-9567763.html
    https://www.theguardian.com/busines...nologies-slips-as-intel-sells-remaining-stake

    Apple seems to be working on their own GPU design. They hired a lot of GPU specific people like Intel.
     
  40. BrianB

    BrianB [H]Lite

    Messages:
    89
    Joined:
    Jan 16, 2017
    that was my point.