I don't know how long we should give it. Sage 4.0.alpha0 has been released thanks to the hard work of Michael Abshoff and many others over the past few weeks, however as there is so much new stuff in Sage 4.0 it seems that it may yet be a while before they get to final. MPIR 1.2 is not slated for release in Sage until 4.0.1, so that should be at least another week from now. Perhaps we can have a code freeze later than I thought, so long as we begin pretesting now so that we do not get surprised by the quantity of bugs that we need to fix at the last minute.
By the way, cuda1 is back up, though still without CUDA drivers. The switch to RHEL has been abandoned. Instead we are going to try and get CUDA working on Ubuntu apparently, but we need someone with more linux knowledge to figure out how to install them. This time around, one of us will have to do SkyNet testing whilst the other does fsffrance testing. The large buildfarm in Germany still isn't available to us, though we will probably get access eventually. That will also give us access to a large number of older machines. Bill. 2009/5/23 Jason Moxham <ja...@njkfrudils.plus.com>: > > On Saturday 23 May 2009 09:14:51 Bill Hart wrote: >> On 21 May, 07:40, Cactus <rieman...@googlemail.com> wrote: >> > On May 21, 4:36 am, Bill Hart <goodwillh...@googlemail.com> wrote: >> > >> > In theory it should be possible to use MSYS to build 32-bit versions >> > of MPIR on Windows. In practice, however, it seems quite likely that >> > this won't work because, as far as I am aware, nobody is maintaining >> > this build system. >> >> I don't know about maintaining support for 32 bit MSYS, however, my >> version of Vista is 32 bit and I am making sure the releases build and >> pass tests on that system with MSYS. I don't have access to other >> configurations to try though, so support is certainly limited. >> Obviously our main project goal is to support MSVC on Windows, which >> is a much, much more capable compiler for Windows. MSYS is improving >> over time though, I believe. For example I think the new version of it >> will basically be gcc 4.4.0 for Windows and should work on 64 bit >> Windows. I still expect it to be flaky compared to the native MSVC >> option though, at least in the early stages of development. >> >> Another option for people is Cygwin, which I have also been checking >> works on my system. However on Vista the performance of Cygwin is >> unacceptable. It may be ok for XP users though. However it should be >> noted that it really isn't the same thing as an MSYS build. It >> basically assumes you are building all your software in Cygwin, >> whereas if I understand correctly, one can have hybrid MSYS/MSVC >> software. >> >> Speaking of build testing, we should be close to build testing for the >> next version. We need to wait for Jason to see if he sorts out the >> divrem_2 issue, I will toughen up the FFT and we'll have to run lots >> of tests of all the new code, e.g. the Toom stuff and XGCD, which I >> *expect* to fail on multiple systems, as some code paths have not been >> tested at all yet. I think solid testing of this release will be much >> more important than past releases, due to the large amount of new >> code. >> >> Bill. > > Hi > > I've removed the badly predicted branch which was causing the slowdown , but > we dont get much of a speedup , as the current implementation is still bound > by latency. I'll see what I can do today/tommorrow. How long until code > freeze? > > Jason > > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en -~----------~----~----~----~------~----~------~--~---
