Message boards :
Number crunching :
MacOSX
Message board moderation
Author | Message |
---|---|
Send message Joined: 22 Jul 05 Posts: 31 Credit: 2,909 RAC: 0 |
Hi! I saw that LHC isn't availabl for MacOSX because of a compiler problem. Can't LHC be compiled with G95, http://g95.sourceforge.net/, for MacOSX? It seems to be doing quite well now. |
Send message Joined: 1 Sep 04 Posts: 506 Credit: 118,619 RAC: 0 |
The CERN team have been through a number of iterations with compilers. The underlying problem is with SixTrack's sensitivity to small numerical differences, where even a difference in bit 0 on certain maths function can alter the result substantially over the entire calculation. Early in the project this caused significant headaches where Intel and AMD processors differed in their results. Yes, the G95 compiler could be used, but without extensive testing there's no guarantee that the numerical results from the Apple platform would match those from the now trusted Intel/AMD application. CERn has changed the maths library to achieve this consistency. It has been suggested that they are now looking at using GNU compliers, and if this works they might consider an Apple version. However, this will take time, and given that the regular team on LHC@Home numbers about half a full time person spread between three individuals with other responsibilities, I wouldn't expect any news any time soon... Gaspode the UnDressed http://www.littlevale.co.uk |
Send message Joined: 22 Jul 05 Posts: 31 Credit: 2,909 RAC: 0 |
Developing and managing computational projects wtih too litle time to do it is something I've had my own share of too so I'm certainly not complaining her, just being curious. As long as one doesn't contribute to the coding one don't really have the right to complain much. Having this kind of machine dependency is something that is quite different from my own work, where any variation at all can't be tolerated. Given that there seems to be quite a lot of people who wants to run LHC it might be worth changing to a numerical library which does everything in machine independent way, at the cost of slowing things down a bit. But I realise that this might need a LOT of rewriting of nontrivial code. I like the project, maybe we'll see an ITER@home in a few years time too. |
Send message Joined: 23 Jul 05 Posts: 1 Credit: 0 RAC: 0 |
Pardon my ignorance, but where might I have found out that there was no Mac client software for this project before going through the trouble of signing up? |
Send message Joined: 2 Sep 04 Posts: 545 Credit: 148,912 RAC: 0 |
> Pardon my ignorance, but where might I have found out that there was no Mac > client software for this project before going through the trouble of signing > up? The "Apps" page. there is usually a link to it off of the accunt set-up pages. There are tables in the Wiki too ... if there is not a direct link, just take the normal part of the web address and add "apps.php" and that will get you home ... The other projects, for the most part have OS-X apps ... though the CPDN app always crashes on my PowerMac G5 ... so, I don't know what is going on there ... |
Send message Joined: 17 Sep 04 Posts: 103 Credit: 38,543 RAC: 0 |
> I like the project, maybe we'll see an ITER@home in a few years time too. > enlighten me. what is ITER@home? |
Send message Joined: 1 Sep 04 Posts: 506 Credit: 118,619 RAC: 0 |
> > I like the project, maybe we'll see an ITER@home in a few years time > too. > > > > enlighten me. what is ITER@home? > ITER is the name for the new experimental fusion reactor to be built in France over the next few years. It is supposed to be the first such reactor that will produce more energy than it consumes. ITER@Home is the speculative name for a DC project that could possibly be be put in place to support the build. I stress 'speculative', since as far as I know there is no such project, nor any plans for one at this time. See http://www.iter.org for more info. Gaspode the UnDressed http://www.littlevale.co.uk |
Send message Joined: 17 Sep 04 Posts: 103 Credit: 38,543 RAC: 0 |
I understood the speculative nature of the name when I asked. Thank you for the information. I will have to look at this site. It looks very interesting. I wish the people here in the US would accept alternative fuels for generating stations. Sheesh. Jim > ITER is the name for the new experimental fusion reactor to be built in France > over the next few years. It is supposed to be the first such reactor that will > produce more energy than it consumes. > > ITER@Home is the speculative name for a DC project that could possibly be be > put in place to support the build. I stress 'speculative', since as far as I > know there is no such project, nor any plans for one at this time. > > See http://www.iter.org for more info. > > |
Send message Joined: 22 Jul 05 Posts: 31 Credit: 2,909 RAC: 0 |
Yes I am not avare of any such DC project either. However using this approach to help fiding the correct starting points for machines like LHC and ITER will probably grow more and more common. Another future candidate for simulation tuning which is closer to LHC might be its planned sibling the International Linear Collider, which is still on the drawing board. ILC |
Send message Joined: 4 Oct 04 Posts: 8 Credit: 783 RAC: 0 |
That link doesn't work, try this one. |
Send message Joined: 1 Sep 04 Posts: 506 Credit: 118,619 RAC: 0 |
> That link doesn't work, try <A> HREF="http://www.interactions.org/linearcollider/"> this one[/url]. > Oops - my mistake. Try this (corrected) version: http://www.iter.org Gaspode the UnDressed http://www.littlevale.co.uk |
©2024 CERN