Bad, but not too bad

I had been quite smart yesterday and outsmarted me by leaving my computer at work. I had to sleep earlyish. And that didn’t work either. Felt quite horrid when I woke today. Things got better after I showered with steam, and I upgraded my condition to 73.342%. Yes, I like numbers that sound insanely accurate and not really tell you anything.

The day went off quite well (considering it’s cold and wet) with the highlight being at a nice talk by the chair of the ME department at M.I.T, Dr. Abeyaratne. The cutest three year old was sitting next to me with her dad. I was totally amazed at how well she behaved, considering continuum mechanics isn’t horribly interesting when you’re three. Hmm, or is it.

Back at the lab, rebuilt legolas with RH9, mostly. The nice thing about legolas, apart from the multiple processors, is it’s got multiple hard drives too. That way the transition from RH7x was a lot less lossy, at least people’s data wise. The same can’t be said about the programming environment (and hence productivity), sadly. Asked Hashem to try his code, since it normally takes weekends to finish simple runs.

Bad move. gcc/g77 is pretty sad when it comes to performance as compared to icc/ifc. The code which was happily whizzing by on sauron (intel compiler compiled) was crawling, as in literally. There was no need to time any processes, it downright sucked. But then again, g77 found a few blatant bugs that ifc had happily ignored while compiling.

The sad, and in a “this is going to be a free only software machine” way, good thing is that intel’s current gen compilers don’t like glibc 2.3x. Which sucks, because I like glibc 2.3x. And this is why I was struggling so much to get objects from icc and gcc to coexist and live happily on my laptop as well.

But then again, free software just feels good. Slower or not. I wish intel would drop their line and merge fancy schmancy extreme x86 optimizations into the gcc line. That’d be good karma.