#7606: Stride scheduling for Haskell threads with priorities ---------------------------------+------------------------------------------ Reporter: ezyang | Owner: ezyang Type: feature request | Status: new Priority: normal | Milestone: 7.8.1 Component: Runtime System | Version: 7.7 Keywords: | Os: Unknown/Multiple Architecture: Unknown/Multiple | Failure: None/Unknown Difficulty: Unknown | Testcase: Blockedby: | Blocking: Related: | ---------------------------------+------------------------------------------ Comment(by ezyang): When I attempt to replicate the conditions of #3838 (by removing the thread bump), I do not see any appreciable increase in runtime. This occurs even if I explicitly yield while inside the atomicModifyIORef (using unsafePerformIO), or I increase the number of threads being run. My conclusion is that a small pileup does occur, but since there's no longer a global list we keep traversing it's not expensive. But I can't rule out (1) the test-case having enough incidental other changes that have made it stop tickling the bug, or (2) some other RTS change masking over this particular problem. A conundrum! -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/7606#comment:37 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler