🚀 go-pugleaf

hi,

i'm using gcc2.6.0 on a Pentium 90 running Solaris2.5.
i have a large application which is floating point
intensive.  prelimanry benchmarks show the float
performance to only be about half of what our 50mhz
sparc20s give, whereas the published specflops say
the pentiums should only be 5-10% slower.

i remember hearing a while back that originally gcc
didn't do a good job of optimization on the pentiums,
but that someone had fixed that. is the solution to
compile the latest version of gcc? or do i have to use
some specific compiler flags to improve performance?
i looked for a gcc faq but didn't see one, so i'd appreciate
it if someone could point me to a source of information.

thanks,
-ed