• o1i
  • Cocoon
    Posts: 45 from 2003/2/25

    takemehomegrandma wrote:

    polluks wrote:
    MorphOS, what are you waiting for?

    Is it "silly SMP"?

    No. It is a completely new approach, which does not have much in common with silly-smp.

    Silly-smp was started on hosted and tried to keep most of the (user) code unchanged. It reached a state, where all started programs got distributed at task-startup to different cores. Runtime-scheduling between different cores was still missing. Worked "only" on hosted. Forbid() was quite a performance killer. But it proofed that SMP can be done.

    Current approach is different ("more general") and tries to get rid of many (all) cases where Forbid() needs to halt all other cores. Was mainly developed for x86_64 bit CPUs. Up to now there is no automatic distribution of tasks to the different cores. All code runs on core0, special coded code can run on all other cores without Forbid() speed penalties. *Is really not on a state where it is of any use for end-users.* Even application developers will have no benefit either.
  • »17.08.17 - 09:47