On Fri, Mar 21, 2014 at 4:56 AM, Stefan Zager <szager@xxxxxxxxxxxx> wrote: >> Considering all that, Duy's solution of opening separate file descriptors per thread seems to be the best pattern for future multi-threaded work. > > Does that mean you would endorse the (N threads) * (M pack files) > approach to threading checkout and status? That seems kind of > crazy-town to me. Not to mention that pack windows are not shared, so > this approach to multi-threading can have the side-effect of blowing > out memory consumption. Maybe we could protect and share the delta cache. Pack windows are mmap'd so we should not need to worry about their memory consumption. > We have already had to dial back settings for > pack.threads and core.deltaBaseCacheLimit, because threaded index-pack > was causing OOM errors on 32-bit platforms. Hm.. I don't think index-pack uses sha1_file.c heavily. Local pack access is only needed for verifying identical objects (and that should never happen often). Something is fishy with these OOM errors. > Cygwin (and MSVC) should be able to share a "mostly" compliant pread > implementation. I don't have any insight into NonstopKernel; does is > really not have a thread-safe pread implementation? If so, then I > suppose we have to #ifdef NO_PREAD, just as we do now. > > I realize that these are deep changes. However, the performance of > msysgit on the chromium repositories is pretty awful, enough so to > motivate this work. -- Duy -- 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