In article <mailman.986245212.31720.python-list at python.org>, "Ken says... > >From: "Grant Griffin" <not.this at seebelow.org> >> . . . >> The essential trade in using Python is that you give up some run-time >> speed to pick up development-time speed--and that's almost always a good >> trade. In any application where minimizing run-time speed isn't >> important, Python's your man. > >Actually python is a perfectly good choice for applications where run-time >speed -is- important, provided that the 2% of the code where significant >CPU cycles are actually burnt are written in C or C++. In many cases this >doesn't even require writing extensions since most time is spent inside >existing >libraries. For example many scientists are using python for serious number- >crunching (using numeric python and related libraries). Yes, but are they doing it in "real time"? Numeric Python is a wonderful thing (in fact, I've even used it myself--you know, to do DSP <wink>), but it's still useful mainly for "offline" processing--at least that's the only way I ever use it. >I use python to trade irrelevant run-time speed for development-time speed. >And that is always a good trade. :-) (You forgot "QED". <wink>) Problem is, there's rarely such a thing as irrelevant run-time speed in the real-time/embedded applications I often work on. In fact, something like Numeric only runs fast if you give it reasonably large pieces of data to chew on per operation. (Otherwise, you spend a lot of time in the Python interpreter.) And in the real-time/embedded world, you usually can't operate on large blocks of data because of memory and/or latency constraints. So I guess I remain un-tempted to use Numeric Python for real-time/embedded applications. But like Python itself, it's a wonderful thing for offline processing. Matlab is analagous to Numeric Python vis-a-vis run-time speed: in both cases, the idea is to use an interpreted language to control operations on blocks of data which themselves occur at compiled speed. But even Matlab adherents also don't attempt to embed Matlab itself: they use it offline. However, I think the Matlab folks are selling some sort of snake oil (in the bad sense <wink>) to automatically compile Matlab into C, which you are then supposed to compile into C. as-a-wise-man-once-said,-"I-pitty-da-fool"-ly y''rs, =g2 _____________________________________________________________________ Grant R. Griffin g2 at dspguru.com Publisher of dspGuru http://www.dspguru.com Iowegian International Corporation http://www.iowegian.com
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4