Comment real computation for real scientists... (Score 1) 152
I'm a real scientist and a real problem -- I use a variety of computation platforms for various purposes and so I'm constantly working with several different systems (vendors not mentioned here.)
I've recently adopted Mac OSX and the TiBook platform for the convience of using my unix/X11/GL based tools with standard desktop applications such as Word, Excel, and Powerpoint. Analogous desktop applications are available in the Linux/Unix world to fill this gap but I've found compatiblity problems with collaborators prevent this from being a robust alternative. So my desk usually has a Wintel box with cygwin/Xfree, a Wintel laptop similarly configured, a multiprocessor Linux workstation (for more brawny computing but not needing beowolf type performance), an SGI Octane/2 (for specific hardware/software interface to data acquisition eqipment which still has a truely impressive graphics engine regardless of the problems with CPU performance) and yes a Mac. Now using OSX I've put all but 1 of these platforms in one box (the SGI apps are hardware specific). From my viewpoint, the power of OSX is not that "it's pretty" or that "it has killer performance" but that its an all - in - one package. My work involves image processing, analysis, modeling, and relatively large (but not grand challenge level) calculations. Mac OSX is reasonably easy to port things to (at least as far as the X11 / OpenGL interface) and I'm quite content with it's performance and all the room I've regained on my desk.
In synopsis - OSX is a great platform to me in that I can develop my software, perform my calculations, and analyze my data in a fashion that is familar and transferable to other platforms without sacrificing the availability of the now-standard (at least in my organizational world) of Micro$oft office and other necessary desktop applications. Its a nice bonus that the architechture of the Mac enables such things as
PVM and other parallel computation tools but for the most part the economics do not work in favor of trading in my Linux beowolf cluster for a similarly configured rack of Xservers running 10.2.
I've recently adopted Mac OSX and the TiBook platform for the convience of using my unix/X11/GL based tools with standard desktop applications such as Word, Excel, and Powerpoint. Analogous desktop applications are available in the Linux/Unix world to fill this gap but I've found compatiblity problems with collaborators prevent this from being a robust alternative. So my desk usually has a Wintel box with cygwin/Xfree, a Wintel laptop similarly configured, a multiprocessor Linux workstation (for more brawny computing but not needing beowolf type performance), an SGI Octane/2 (for specific hardware/software interface to data acquisition eqipment which still has a truely impressive graphics engine regardless of the problems with CPU performance) and yes a Mac. Now using OSX I've put all but 1 of these platforms in one box (the SGI apps are hardware specific). From my viewpoint, the power of OSX is not that "it's pretty" or that "it has killer performance" but that its an all - in - one package. My work involves image processing, analysis, modeling, and relatively large (but not grand challenge level) calculations. Mac OSX is reasonably easy to port things to (at least as far as the X11 / OpenGL interface) and I'm quite content with it's performance and all the room I've regained on my desk.
In synopsis - OSX is a great platform to me in that I can develop my software, perform my calculations, and analyze my data in a fashion that is familar and transferable to other platforms without sacrificing the availability of the now-standard (at least in my organizational world) of Micro$oft office and other necessary desktop applications. Its a nice bonus that the architechture of the Mac enables such things as
PVM and other parallel computation tools but for the most part the economics do not work in favor of trading in my Linux beowolf cluster for a similarly configured rack of Xservers running 10.2.
