David Haller wrote: > Hello, > > On Thu, 24 Dec 2015, Alan Grimes wrote: >> Hey, thanks for putting out gcc 5.3... >> >> Unfortunately, it fails to bootstrap on my machine. I am getting >> differences between the stage 2 and stage 3 compilers and it's dying... =( > What compiler and C(XX)FLAGS are you using? It builds nicely with > gcc-4.9 (and rather agressive flags) here. With how many processes > (MAKEOPTS) are you compiling?.
6, one per core, because it doesn't go to 100% utilization, I can continue to use the machine normally for desktop. > Also, compiling gcc might expose flaky RAM (removing and reseating the > RAM DIMMs might help there[0]). Have you e.g. encoded stuff with > mencoder or ffmpeg lately[1]? If you're compiling with any MAKEOPTS > '-j1' it could be the same phenomenon as I had in [1], so try with > -j1. If that works ... Consider using prime95/mprime[2] for a test. Been through that, I'm fairly confident with the integrity of the machine right now. > -dnh > > [0] stuff expands and shrinks depending on temps, so connections can > get flaky even if the box isn't moved ;) > > [1] I've had mencoder segfault on me reproducibly on one box after > 20-30 mins encoding with one process, and <= 10mins with two > processes. Memtest86 found nothing in one round. New RAM (replaced > on guarantee :) cured the problem for good. > > [2] http://www.mersenne.org/download/#download -- IQ is a measure of how stupid you feel. Powers are not rights.
