Related link: http://www.intel.com/cd/ids/developer/asmo-na/eng/255716.htm?iid=HMPAGE+HL_06011…
Intel has unveiled beta versions of developer tools for Intel-based Macs. The Development Support for Intel-Based Mac page anounces a set of software development products, including highly-optimized C++ and Fortran (really?!) compilers, plus a math library and a primitives library that is said to make digital signal processing happy. Tools are beta and require you to apply to the testing program (disclosure: I’m on PPC and will be for a while, so I haven’t downloaded them)
As a Mac-only person, I first started paying attention to Intel and their message at their keynote JBoss World 2005. The message of “use our chips, we’ll help you” is certainly a change from the indifference the Mac developer community is used to from the PowerPC crowd (pity too - we probably could have put all those PPC registers to good use, but who knows how optimized gcc was for PPC?).
Anyways, they supposedly integrate with XCode, though there’s an obvious question of how useful they’ll be for Cocoa developers working in Objective-C, which isn’t mentioned on Intel’s page. Also, I wonder if its possible to build a universal binary and pick up the Intel optimizations for the i386 architecture. I don’t think anyone’s ready to build apps that don’t run on PPC. Yet. Give it a year or two.
Got an iMac or MacBook? Tried these? Do they beat gcc?
1000 engineers
Didn't the Intel chief at the keynote say that they've had 1000 engineers at Intel working with Apple on the new Macs? That's pretty impressive, I would think, and shows committment. I just hope that they stick with it. A problem between Apple and IBM, percevied or real, was that IBM didn't put in enough effort into the partnership.
Apple's working on gcc for both arches
And will continue to do so, as long as they ship products with both arches, I guess. I'd hope to see some nice improvements in gcc generated code's performance judging by http://gcc.gnu.org/gcc-4.1/changes.html . Apple's also interested in wielding gcc together with llvm, which could be some really good news, in particular for gcj(x), giving it rapid access to a sweet jitter.
cheers,
dalibor topic