> To those who have reported bit-vector trouble over the past couple of
> weeks: it would be helpful if you could check out CVS as of now and
> see if the performance problems are resolved. I have implemented what
> I believe are a much faster SXHASH and a much faster EQUAL on
> simple-bit-vectors, but I haven't exercised them apart from in simple
> If your bit-vectors are actually non-SIMPLE-ARRAYs, then you won't get
> a performance improvement currently... this is fixable, but is
> significantly fiddlier, so I've left it alone for now.
My current hotspot is SIMPLE-BIT-VECTOR EQUAL. My previous benchmark
time for 0.7.13 was ~ 1.5 seconds, and 0.7.13.30 runs in ~ 0.9
seconds. In my particular program, this makes the measurements of
bit-vector/vector/list performances I've been doing more
comprehensible, since I'd turned to bit-vectors as sets and was
surprised I wasn't doing better than other approaches wrt EQUAL.
Thanks a lot for this improvement; I know performance tuning often
isn't that interesting.