 |
 |
Hej Andreas! we have experience with both linux based processing boxes, and with macs (G4 and G5) running osx, and recently with panther. my comments regarding processing speed would be that the dual processor isn't utilised by matlab, nor is the 64bit architecture, but the macs are fast. another issue re speed, is that running the stats part of spm, it seems that disk access issues play a large role in speed, such that our G4 with dual 1.4GHz runs almost as fast as our G5 with dual 2.0GHz, when accessing data on a nfs raid disk. Regarding stability, i would issue a word of warning regarding panther and matlab. the patch to allow r13 to run is quite new, and we have had quite a few jobs crash (which wasn't the case on 10.2.1). at this stage we are in the midst of working out the conditions under which this happens, but panther and matlab definitely exhibit more issues than jaguar did. hopefully all this is just teething pains, and things will run smoothly soon. i guess the bottom line is that if you want a solution that will be probelem free now, i would warn you off the mac solution, despite the fact that we are trying to go that way ourselves. perhaps a dual platform solution is always the best? hope that was of some help kind regards tony On 20/11/2003, at 11:54 PM, Andreas Roepstorff wrote: > Dear SPM users, > We are currently upgrading our spm analysis system, and the new mac G5 > machines could be an attractive solution at least on one of our > setups. However, I have not been able to find any benchmarks on the > processing speeds. It seems that at present, Matlab is not able to > utilise a dual processor and/or the 64 bit architecture. Is that a > wrong impression? > Does anyone have experience with running SPM on a Mac G5 computer > using Panther compared with running it on a fast linux based machine? > Are there reasons, speedwise or in terms of stability, to prefer one > to the other? I hope someone on the list has some experiences to > share. > Best greetings > andreas > > > Andreas Roepstorff > Centre for Functionally Integrative Neuroscience. Aarhus University > Building 30, AKH, Nørrebrogade 44, DK-8000 Aarhus C, Denmark > [log in to unmask], +45 8949 3030 > http://www.pet.au.dk/~andreas/index.htm > ---------------------------- Tony Waites, PhD Brain Research Institute Ground Floor, Neurosciences Building Austin Health Banksia Street, Heidelberg West, Victoria, 3081, Australia. Telephone: +613 9496 4081 Facsimile: +613 9496 4071 Email: [log in to unmask] http://www.brain.org.au