neal@uikku.tut.fi (Norwitz Neal) (11/09/90)
I have a recursive algorithm that I would like to run as fast as possible. I am running a part of it on a SPARCstation SLC. But at this rate, it might finnish in a couple of years!! I was wondering if someone could suggest a couple of architectures that may suit this recursive algorithm better than others. The program takes no memory (about 12K). So all I need is speed. There are no floating point operations, except a couple of increments. Also, any information about speed ratings (in mips) from many different computers would be appreciated. Thanks a lot. Neal Norwitz neal@tut.fi Tampere University of Technology
tr@samadams.princeton.edu (Tom Reingold) (11/09/90)
In article <1990Nov8.212412.14317@funet.fi> neal@uikku.tut.fi (Norwitz Neal) writes: $ $ I have a recursive algorithm that I would like to run as fast as possible. I am running a part of it on a SPARCstation SLC. But at this rate, it might finnish in a couple of years!! I was wondering if someone could suggest a couple of architectures tha $ t may suit this recursive algorithm better than others. The program takes no memory (about 12K). So all I need is speed. There are no floating point operations, except a couple of increments. $ $ Also, any information about speed ratings (in mips) from many different computers would be appreciated. You pose a vague question, so I don't know what we (the net) can suggest. If it's really going to take years with the current algorithm, it sounds worthwhile changing it to an iterative one. Have you looked at lisp systems? Some of them handle recursion pretty efficiently. I am not talking about lisp running under UNIX. I am talking about machines with lisp semantics in the processor. -- Tom Reingold tr@samadams.princeton.edu OR ...!princeton!samadams!tr "Warning: Do not drive with Auto-Shade in place. Remove from windshield before starting ignition."
berg@cip-s04.informatik.rwth-aachen.de (AKA Solitair) (11/16/90)
Norwitz Neal writes: >I have a recursive algorithm that I would like to run as fast as possible. >I am running a part of it on a SPARCstation SLC. But at this rate, it might >finnish in a couple of years!! Why don't you have a look at the FORTH-chip (don't remember the manufacturer, the part number should be something like 4016 if I recall correctly), the processor has no general purpose registers, just a stack cache, the thing was designed to execute *native* FORTH (a medium to high level threaded stack based language) code. Last I heard about it, they we're designing a new one. This processor was practically designed for recursive algorithms. -- Sincerely, berg%cip-s01.informatik.rwth-aachen.de@unido.bitnet Stephen R. van den Berg. "I code it in 5 min, optimize it in 90 min, because it's so well optimized: it runs in only 5 min. Actually, most of the time I optimize programs."