Message boards :
Number crunching :
Mac Intel anyone?
Message board moderation
Author | Message |
---|---|
Send message Joined: 18 Jan 06 Posts: 6 Credit: 7,941 RAC: 0 |
I've searched the boards, and while I found a suiting thread it didn't have a conclusive reply. I recently got a Macbook and I thought its two cores would come handy for crunching while I wrote my papers and whatnot. However, communication is constantly deferred (24 hours) with the message "Message from server: platform '1686-apple-darwin' not found". I assume this is because the sixtrack thingy is not ready for the Mactels yet. Two questions though: a) Is there any official word from LHC@Home on the matter? Einstein@Home works correctly. b) If I reverted to an old PowerPC version of BOINC, would it work? c) Would the fact that the PPC version is not native hurt my number-crunching capacity? Thanks a bunch in advance. I can't wait for my Mac to contribute to LHC. |
Send message Joined: 1 Sep 04 Posts: 506 Credit: 118,619 RAC: 0 |
There is no version for any Mac available. The issue has been extensively explored in the message boards, but put simply to get the best from the Windows and Linux x86 systems (the most common) a compiler hass been chosen isn't available for the Mac environment. There is very little development or admin support available for LHC at present, so it's very unlikely that any Mac version (even for Intel Macs) will become available. Gaspode the UnDressed http://www.littlevale.co.uk |
Send message Joined: 2 Sep 04 Posts: 378 Credit: 10,765 RAC: 0 |
Since you're running the MAC OS on Intel hardware, you may be able to run Boinc using something like WINE. http://darwine.opendarwin.org/ Note: I don't have a mac, but I did play with Boinc and Ran Boinc under WINE using Linux (including a few LHC work units in the early days) on my intel boxes. b) If I reverted to an old PowerPC version of BOINC, would it work? Nope. LHC doesn't have any PowerPC clients running sixtrack. The floating point processors have different precisions, so when you add up the differences over a million rotations, you get a significantly different result from the AMD and Intel results, which means the Power PC boxes wouldn't verify. c) Would the fact that the PPC version is not native hurt my number-crunching capacity? I played with Linux runing the Win32 boinc using WINE. It was a little bit slower than running under windows, and there were issues with the boinc app. Try it out. Since the results verified when I played with Linux, I think the results should work with DarWINE as well. Since you have an intel box, you'd use the Intel version of Darwine. Maybe I'll get myself a mac mini one day. Oh, avoid QEMU for crunching. It's a processor emulator, so you'll get a 85% performance hit if you were to run a cruncing app. I'm not the LHC Alex. Just a number cruncher like everyone else here. |
Send message Joined: 21 May 06 Posts: 73 Credit: 8,710 RAC: 0 |
I've searched the boards, and while I found a suiting thread it didn't have a conclusive reply. Have you considered running "PARALLELS?" You may wish to ask your question here: http://forum.parallels.com/forum53.html |
Send message Joined: 1 Sep 04 Posts: 137 Credit: 1,691,526 RAC: 22 |
c) Would the fact that the PPC version is not native hurt my number-crunching capacity? I suppose this is barely worth discussing since there isn't a PPC version of sixtrack... but running a PPC binary on intel macs using rosetta is very different than running an x86 windows binary under x86 linux using wine. Wine just provides emulation for system calls but the actual crunching code runs directly on the CPU. Rosetta on the other hand must translate machine code on the fly so it would be more like using QEMU. I have heard that rosetta is pretty good about it but I'm pretty sure you will see a big performance hit. - A member of The Knights Who Say NI! My BOINC stats site |
©2024 CERN