Announcement

Collapse
No announcement yet.

Intel GCC Patches + PRM Update Adds SERIALIZE Instruction, Confirm Atom+Core Hybrid CPUs

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

  • Intel GCC Patches + PRM Update Adds SERIALIZE Instruction, Confirm Atom+Core Hybrid CPUs

    Phoronix: Intel GCC Patches + PRM Update Adds SERIALIZE Instruction, Confirm Atom+Core Hybrid CPUs

    Intel has seemingly just updated their public programming reference manual as well as sending out some new patches to the GCC compiler for supporting new instructions on yet-to-be-released CPUs...

    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
    SERIALIZE.
    Is that like: "Hey. We can't verify our designs, so here is an instruction to reduce it to i586 speed so you can make sure you don't fall into our fuckups." ?
    And also, what is the difference between this and normal memory barriers?

    Comment


    • #3
      Originally posted by milkylainen View Post
      SERIALIZE.
      Is that like: "Hey. We can't verify our designs, so here is an instruction to reduce it to i586 speed so you can make sure you don't fall into our fuckups." ?
      And also, what is the difference between this and normal memory barriers?
      This looks like both an instruction and memory barrier at the same time. Might be needed for maximum security... and increased slowness

      Comment


      • #4
        Originally posted by ldesnogu View Post
        This looks like both an instruction and memory barrier at the same time. Might be needed for maximum security... and increased slowness
        Yeah, it's an explicit fence and memory barrier, like MIPS's "sync" instruction.

        Comment


        • #5
          Are the new instructions the operations needed to enable PCIE5/CXL asymmetric cache coherency, which is expected to be used in Aurora for the GPU shared memory?

          Comment


          • #6
            Confirm Atom+Core Hybrid CPUs
            big.LITTLE, asshat edition.

            Comment


            • #7
              Originally posted by SystemCrasher View Post
              big.LITTLE, asshat edition.
              What's wrong with it? Should bring more power efficiency so would be great for laptops/portables, could be swung the other way as bringing the market that usually uses atoms as getting some extra processing oomph when needed too.

              Dunno if that'd be something useful on laptops that AMD could do too with their 15W or whatever is lowest parts, paired with the higher wattage 45W parts? Since their zen architecture is modular, wouldn't that allow for a mix of power efficient cores and others with more processing grunt?

              Comment


              • #8
                What's wrong is x86 and surroundings + Intel's way of doing things (woeful overengineering, overinclination on way too fast and powerhungry parts + marketing BS). Sure thing, with enough thrust pigs fly just fine. And that thing looks exactly like that.

                Comment

                Working...
                X