Re: mipsel/cobalt: GCC producing extremely slow code

Thank you both for the responses. Interestingly enough on the clock
question, I just managed a gcc3 -O2 build that peaked at 33 MFLOPS and
now is running 900 KFLOPS after a reboot. Per Jonathan's suggestion, I
dumped that binary to a file and then did the same for the next build
showing better results, they are indeed the same, so I think that
rules GCC out for sure.

I think Martin is on to something with the clock question. Would a
dead or shoddily attached RTC battery potentially cause issues with
timing like this? The battery was previously dead, and then when I
attempted to replace it while swapping out the hard drive to prepare
for a new 9.2 install the socket came out with it and I had to
resolder it. The connection seemed good and it's keeping
time/maintaining its configuration while powered off, but now I'm not
so sure...