19:08:38stassatsbut i hit a fixed bug in sb-fasteval
19:21:11stassatsit would be more useful to time make-host-2 for scaling, although all the serial bits are important as well
19:22:32josheI thought timing a parallel build under clisp would make the differences more obvious, but now I'm bisecting to find the last commit that broke clisp hosts
19:23:01stassatsSBCL_MAKE_JOBS=-j160 => 120% cpu 5:42.97 total + SBCL_MAKE_PARALLEL=160 => 532% cpu 2:26.69 total
19:36:41stassats(that's make-host-2.sh without make-genesis)
19:40:50stassatsso it does seem to scale to 12 cores, but since that's the only part of the build that scales well, the total times get little improvement