Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!snorkelwacker.mit.edu!ai-lab!zurich.ai.mit.edu!jaffer From: jaffer@zurich.ai.mit.edu (Aubrey Jaffer) Newsgroups: comp.lang.scheme Subject: Re: Benchmarking Scheme Message-ID: Date: 4 May 91 03:29:55 GMT References: <1991May2.155838.20830@bronze.ucs.indiana.edu> Sender: news@ai.mit.edu Distribution: comp.lang.scheme Organization: M.I.T. Artificial Intelligence Lab. Lines: 27 In-reply-to: huntley@copper.ucs.indiana.edu's message of 2 May 91 15:58:38 GMT huntley@copper.ucs.indiana.edu (Haydn Huntley) asks: >Did you use the same compiler and same options on both your 68000 and >68020? No and Yes. The 68000 is a 1984 vintage unix and the 68020 is an HP-UX. I used no compiler options on both. >Do they both run at the same clock rate? Your C benchmarks >show that kaleph is 8 times faster than your 68000! My 68000 is 10Mhz. The 68020 is 25Mhz. >I'd say that your interpreter seems to be incredibly fast! Is it an >really an interpreter or actually a compiler? Scm has an interpreter written entirely in C (it uses no macros written in scheme). In addition, it memoizes the locations of all variable references by modifiying the code. This is described in code.doc in the distribution. >Does it support generic arithmetic? (fixnums, bignums, and flonums) Not yet. I am adding this now. However, 30 bit fixnum arithmetic speed will not be affected. >Does it do run-time checks to see if the heap is nearly full? Yes. The times include GC. >Does it perform any type checks at run-time? It checks for number of arguments, type of arguments, structure references, numeric overflow, and unboundness of varibles.