On Fri, Dec 13, 2002 at 05:31:57PM +0000, Stephen White wrote: > My company does a fair amount of work in C++, preprocessed C++ source > files can be quite large and bits of our internal network are not > terribly fast .. so I thought I'd look at implementing the LZO > compression @todo in the source code. > > I've attached my current work for anyone interested to play with and > comment on. It's only been lightly tested so far, and is not ready for > prime time for reasons outlined below.
This looks very nice. However, I noticed that the assembler code was ifdef'd out even on i386. It would be nice if this were used because it should cause a significant performance boost. As the code is already in the file, this seems like low hanging fruit. In addition, the real LZO distribution has some pure i386 assembly versions which are probably even faster than the few lines of optimizations in minilzo. The "fast" (i.e. crashes when fed corrupted data) assembly decompressor runs at between 200MB/sec and 300MB/sec my machine. _______________________________________________ distcc mailing list [EMAIL PROTECTED] http://lists.samba.org/cgi-bin/mailman/listinfo/distcc
