Hey Ken/Chip

https://communities.intel.com/thread/126317

I can do this with P2s... In large number. I like Peter's board, and it will be possible to do demo's with it, but as a final design it would be a royal pain.

And of course I want 28nm.

I can easily imagine several ways to get it done. With your permission, I would like to open a discussion with Intel... not representing Parallax, but just trying to see if we could find a mutually acceptable direction.

Thanks,

Rich

Comments

  • Ken GraceyKen Gracey Posts: 6,530
    edited 2018-06-13 - 14:27:43
    Of course - talk with them. Thanks,

    Ken Gracey
  • Yes, we could really use their 14nm Tri-gate technology. 2GHz+ possible for our current design, and lots of on-chip RAM.
  • cgracey wrote: »
    Yes, we could really use their 14nm Tri-gate technology. 2GHz+ possible for our current design, and lots of on-chip RAM.
    But don't you have to redesign the instruction set for >1MB of hub RAM?

  • David Betz wrote: »
    cgracey wrote: »
    Yes, we could really use their 14nm Tri-gate technology. 2GHz+ possible for our current design, and lots of on-chip RAM.
    But don't you have to redesign the instruction set for >1MB of hub RAM?

    We would have to implement some paging scheme (per cog) in upper memory, or something.
  • Will do.

    Thanks,

    Rich
  • Will that make it a P2X16C4G64P-2G chip then? :)

    Tachyon Forth - compact, fast, forthwright and interactive
    useforthlogo-s.png
    P2 --- The LOT --- TAQOZ INTRO & LINKS --- P2 SHORTFORM DATASHEET --- TAQOZ RELOADED - 64kB binary with room to spare
    P1 --- Latest Tachyon with EASYFILE --- Tachyon Forth News Blog --- More
    paypal.png PayPal me
    Brisbane, Australia
    phone.png
  • cgracey wrote: »
    David Betz wrote: »
    cgracey wrote: »
    Yes, we could really use their 14nm Tri-gate technology. 2GHz+ possible for our current design, and lots of on-chip RAM.
    But don't you have to redesign the instruction set for >1MB of hub RAM?

    We would have to implement some paging scheme (per cog) in upper memory, or something.
    Ah, page tables, virtual memory, a translation look aside buffer. We can run Linux!
  • TorTor Posts: 1,983
    Ah, page tables, virtual memory, a translation look aside buffer. We can run Linux!
    .. or emulate an old minicomputer. I've already got Linux (everywhere) :)
  • David,
    We can run Linux!
    Noooo....

    Well, OK, put a RISC V core in there to run Linux. Let the COGs do what they are designed for.
  • We do not want to do what others do! Innovation comes from leaving the path used. Like, denuking NK by buying their nukes and stop local production.
    no reason to reason if you feel feelings: in love with the propeller

    How-2-TACHYON
  • Heater. wrote: »
    David,
    We can run Linux!
    Noooo....

    Well, OK, put a RISC V core in there to run Linux. Let the COGs do what they are designed for.
    That sounds like a good plan to me.

  • ErNa,
    Innovation comes from leaving the path used.
    Yes. That is why I like to follow what goes on in the RISC V world. And the Propeller world of course. Putting the two together sounds just great.

    Let's hope that straying from the well trodden path does not result in falling off a cliff !



  • Heater. wrote: »
    ErNa,
    Innovation comes from leaving the path used.
    Yes. That is why I like to follow what goes on in the RISC V world. And the Propeller world of course. Putting the two together sounds just great.

    Let's hope that straying from the well trodden path does not result in falling off a cliff !


    Unfortunately, I think it often does. Not a reason not to do it though. Otherwise we’d never get anything new and different.

  • rjo__rjo__ Posts: 2,076
    edited 2018-06-15 - 17:27:39
    For my purposes, the instruction set is perfect. But I also recognize the huge desire of others to fully integrate with other OSs.

    You never know what you are missing if you let an opportunity go by.

    The RealSense people are having a hell of a time integrating their VPU in Windows 10 with C++... I don't think it is all their fault:) I don't know what high level language their VPU uses (if at all).

    For argument's sake... where would the P2 have to be on the nanometer scale to make sure the Prop can handle
    USB 3.1?

    That seems like a pretty good target.
  • AleAle Posts: 2,356
    Paging ? How about some really nice segment registers... I hear they come in 4 flavors. (ducks)
Sign In or Register to comment.