Announcement

Collapse
No announcement yet.

Radeon Vega 20 Will Have XGMI - Linux Patches Posted For This High-Speed Interface

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Radeon Vega 20 Will Have XGMI - Linux Patches Posted For This High-Speed Interface

    Phoronix: Radeon Vega 20 Will Indeed Have XGMI - Linux Patches Posted

    It has been expected that Vega 20 would feature XGMI as a high-speed GPU interconnect alternative to PCI Express and that was firmed up today thanks to a new set of AMDGPU Linux driver patches...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Wait, is this the first step toward chiplets on AMD GPUs??

    Comment


    • #3
      Originally posted by fuzz View Post
      Wait, is this the first step toward chiplets on AMD GPUs??
      From what I've read, yes this is a first step toward chiplets but this does not necessarily mean that they are going in that direction.

      Comment


      • #4
        I guess XGMI has the same role as the Sideport in earlier Radeon generations, for now.
        Last edited by chithanh; 05 September 2018, 03:51 PM.

        Comment


        • #5
          Originally posted by chris200x9 View Post
          From what I've read, yes this is a first step toward chiplets but this does not necessarily mean that they are going in that direction.
          Incremental progress is the route toward success!

          Comment


          • #6
            Originally posted by chithanh View Post
            I guess XGMI has the same role as the Sideport in earlier Radeon generations, for now.
            Closer to home, its role is also similar to XGMI in dual-Epyc systems.
            Test signature

            Comment


            • #7
              This will dramatically improve HBCC. Higher bandwidth and lower latency then PCI-e means that reading from system memory no longer has a large performance penalty attached, and graphics cards will also be able to 'borrow' memory from other graphics cards if they run out.

              Comment


              • #8
                Anyone knows why they don't use the Infinity Fabric to interconnect the GPU dies?

                Comment


                • #9
                  If they can do it on the hardware side of things, yay, but if its anything like SLI or Crossfire which require driver and software workarounds which almost never get fully supported and when they do they result in performance/graphics render issues! , no thanks.

                  Comment


                  • #10
                    Originally posted by bridgman View Post

                    Closer to home, its role is also similar to XGMI in dual-Epyc systems.
                    but they do so
                    i the article
                    "XGMI is a peer-to-peer high-speed interconnect and is based on Infinity Fabric"

                    Comment

                    Working...
                    X