I am, slowly but surely, experimenting with OpenMP in my own git repository of MPIR. As I said a while back, I'd like to do that first, then bring in cuda etc at a later date.
I don't know what to suggest, really. On the one hand parallelisation work is underway, on the other hand it is nowhere near useable in its current form and probably won't be for a while whilst I'm the only person working on it. As I said before, if there are any other efforts going on, I'm more than happy to fall in line behind those, otherwise, I'll keep plugging away in my spare time. When I do get something working, there'll be an explosion of emails in your inbox. I think there are two courses of action: * Remove it. We don't have it in even alpha quality * Make it clear it is something we would like to achieve/are working but isn't at present useable. Something like "experimental areas: ... ... ..." Any thoughts? Antony On 05/24/2010 07:17 PM, Bill Hart wrote: > By the way, to mpir-devel only this time.... Is it still a plan to > work towards parallel support and support of GPU's in MPIR? > > Perhaps Antony still plans to work on this in mpir, in which case we > should leave as is. But if not, maybe it is biting off too much.... > > It's the one claim we've made for a very long time which we've not > been able to deliver on. Perhaps we should remove and put it back when > we have something? > > I dunno, just a suggestion. What do people think? > > I suspect there might end up being parallel factorial code in GMP. I > vaguely recall some stuff about Marco working on that, though I forgot > if he explicitly said parallel or not. > > Bill. > > On 24 May 2010 19:13, Bill Hart <goodwillh...@googlemail.com> wrote: >> OK, I made a little bit of a change to the "about" section of the MPIR >> website. It now has: >> >> "MPIR is an open source multiprecision integer (bignum) library forked >> from the <a href="http://gmplib.org/";>GMP (GNU Multi Precision)</a> >> project. It consists of much code from past GMP releases, in >> combination with much original contributed code." >> >> I doubt that addresses the concern you had, but it is, in my opinion, >> clearer and simpler than what we had. In particular I removed the bit >> about originally being a fork of GMP 4.2.1 (which was LGPL v2+), etc. >> I don't think that information is particularly useful or serving any >> definite purpose any more, though there wasn't anything particularly >> wrong with it. >> >> But again, I'm happy to change something if it isn't clear, or perhaps >> I should say that the new MPIR webmaster will (I've just been >> transferring credentials over for that). >> >> Bill. >> >> On 24 May, 18:43, Bill Hart <goodwillh...@googlemail.com> wrote: >>> Interesting observation. Do you think we should state the original >>> reasons for the fork? Bear in mind, some of these reasons are no >>> longer valid, e.g. GMP now has an open repository. >>> >>> We also aren't releasing new versions under LGPL 2.1+ ourselves any >>> more. After extensive consultation we decided to switch to v3+. The >>> only exception is MPIR 1.2.x which is LGPL v2.1+. I don't know of >>> anyone who has volunteered to take on the project of extending this in >>> any way. It's our intention to fix any bugs found in it only. I >>> suppose this might generate some new releases, but only very minor >>> updates. It certainly isn't a focus for development. >>> >>> I think we do make a pretty big fuss about MSVC support on our website >>> though. That will shortly change to MSVC 2010 support. We should also >>> be making a big fuss about 64 bit Windows assembly code, as that is >>> another feature of MPIR (note the Windows calling conventions differ >>> from the linux ones, so you have to modify the assembly code). >>> >>> Remember that we forked GMP years ago. I'm not sure it is a wise idea >>> to state on the MPIR website what our original reasons for a fork >>> were. Wouldn't that merely be reopening old wounds? >>> >>> I think the current ambitions of the MPIR project are all that really >>> matters and that is all stated pretty clearly. I'm sure we'd be happy >>> to reword if you think something should be added, though. >>> >>> Bill. >>> >>> On 24 May, 17:47, leif <not.rea...@online.de> wrote: >>> >>> >>> >>> >>> >>>> On 24 Mai, 15:59, Bill Hart <goodwillh...@googlemail.com> wrote: >>> >>>>> [...] >>>>> I *apologise unreservedly* for upsetting the GMP >>>>> developers. This flame war, whether well-intentioned or not, is much >>>>> to be regretted. It would have been better to simply state we forked >>>>> because we wished to support Windows development, have an open >>>>> repository and, at the time, continue with an LGPL v2.1+ licensed >>>>> library whilst GMP moved on with v3+, so that other v2+ projects had >>>>> time to assess v3+. That would have been more reasonable on our part. >>>>> We were naive in the extreme to not expect the hostile response we >>>>> got. This we regret deeply. >>>>> [...] >>> >>>> The MPIR web site (http://www.mpir.org/#about) does not state this, >>>> especially not the motivation for the fork. >>> >>>> -Leif >>> >>>> -- >>>> To post to this group, send an email to sage-de...@googlegroups.com >>>> To unsubscribe from this group, send an email to >>>> sage-devel+unsubscr...@googlegroups.com >>>> For more options, visit this group >>>> athttp://groups.google.com/group/sage-devel >>>> URL:http://www.sagemath.org >>> >>> -- >>> To post to this group, send an email to sage-de...@googlegroups.com >>> To unsubscribe from this group, send an email to >>> sage-devel+unsubscr...@googlegroups.com >>> For more options, visit this group >>> athttp://groups.google.com/group/sage-devel >>> URL:http://www.sagemath.org >> >> -- >> You received this message because you are subscribed to the Google Groups >> "mpir-devel" group. >> To post to this group, send email to mpir-de...@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. >> >> > -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-de...@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.
