On Fri, 16 Mar 2007, Linus Torvalds wrote: > On Thu, 15 Mar 2007, Davide Libenzi wrote: > > > > Hacked up test case below ... > > This one seems to do benchmarking with 8MB buffers if I read it right > (didn't try). Yes, I just wanted to have the biggest time spent in inflate(). That why I use a big buffer. > The normal size for the performance-critical git objects are in the couple > of *hundred* bytes. Not kilobytes, and not megabytes. > > The most performance-critical objects for uncompression are commits and > trees. At least for the kernel, the average size of a tree object is 678 > bytes. And that's ignoring the fact that most of them are then deltified, > so about 80% of them are likely just a ~60-byte delta. Definitely. The nature of the data matters. Did you try to make a zlib with my patch and oprofile git on real data with that? - Davide - To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html