John Hunter wrote: >>>>>>"Robert" == Robert Kern <[EMAIL PROTECTED]> writes: > > > Robert> H = U*D*V.T > > Robert> then I'm more than happy with that tradeoff. The small > Robert> syntactic conveniences MATLAB provides are dwarfed by the > Robert> intrinsic power of Python. > > Of course, U*D*V (transpose omitted for clarity) is the classic > problem for an interpreted language: the creation of temporaries. > weave allows you, via blitz, to do chained matrix/matrix operations > without multiple passes through the loop and w/o temporaries by > run-time compilation and linking of extension code. Perhap's the OP's > reference to JIT is referring to a just in time compilation mechanism > in matlab, similar to weave's. They've already discovered LAPACK and > FFTW; it wouldn't be surprising if they solved blitzm (blitz in > matlab), antialiasing, alpha transparency and multiple colormaps per > figure in upcoming releases.
I don't think so, although it is incredibly, ridiculously difficult to find information about what the JIT actually *does*. The best my Google-fu can do is this: http://www.mathworks.com/access/helpdesk_r13/help/techdoc/matlab_prog/ch7_per6.html#773530 It looks like they're primarily improving simple for loops and doing arithmetic with emitted machine code in order to speed up those operations that simply can't be "vectorized." They *might* be doing something smart with array expressions wrt temporaries, but they don't say. google("MATLAB JIT temporaries") gives nothing interesting. -- Robert Kern [EMAIL PROTECTED] "In the fields of hell where the grass grows high Are the graves of dreams allowed to die." -- Richard Harter -- http://mail.python.org/mailman/listinfo/python-list
