jevans@.ucalgary.ca (David Jevans) (11/21/88)
Several articles have compared the Transputer to 68000s
(SUN 3s) and VAX 780s. I had got similar results when
testing some T800s with a beta C compiler (the developer
of which shall remain nameless). Needless to say I was
less than impressed. About 2 months ago I got ahold of
another demo board and a Logical Systems C compiler
(from CSA). My ray tracer, which is very cpu and
floating point intensive, ran 25 times faster than
with the other vendors C compiler. The execution
time was equal to that of a SUN 4!
This just goes to show you that while hardware may be
the same, the compiler can make all the difference.
I would reccommend potential buyers to test as many
different compilers (C, OCCAM, whatever) before
buying. You could be wasting a lot of compute
power.
David Jevans, U of Calgary Computer Science, Calgary AB T2N 1N4 Canada
uucp: ...{ubc-cs,utai,alberta}!calgary!jevans
David Jevans, U of Calgary Computer Science, Calgary AB T2N 1N4 Canada
uucp: ...{ubc-cs,utai,alberta}!calgary!jevansnick@lfcs.ed.ac.uk (Nick Rothwell) (11/22/88)
In article <228@cs-spool.calgary.UUCP> jevans@.ucalgary.ca (David Jevans) writes: >Several articles have compared the Transputer to 68000s >(SUN 3s) and VAX 780s. I had got similar results when >testing some T800s with a beta C compiler (the developer >of which shall remain nameless). Needless to say I was >less than impressed. About 2 months ago I got ahold of >another demo board and a Logical Systems C compiler >(from CSA). My ray tracer, which is very cpu and >floating point intensive, ran 25 times faster than >with the other vendors C compiler. The execution >time was equal to that of a SUN 4! > >This just goes to show you that while hardware may be >the same, the compiler can make all the difference. I view this as a non-sequitur. You don't say whether the second board was identical to the first - same clock speed and so on? I very much doubt that the compiler has much to do with it, but it's probably the run-time support (the floating point package, or whatever); perhaps your first version was running with floating point emulation rather than using the hardware? Perhaps your second version was configured to use the on-chip RAM for stack space? Perhaps a lot of things... :-) Disclaimer: I don't know much about transputer hardware systems, but I have programmed transputers and know something about compilers. >David Jevans, U of Calgary Computer Science, Calgary AB T2N 1N4 Canada >uucp: ...{ubc-cs,utai,alberta}!calgary!jevans Nick Rothwell, Laboratory for Foundations of Computer Science, Edinburgh. nick@lfcs.ed.ac.uk <Atlantic Ocean>!mcvax!ukc!lfcs!nick ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ...while the builders of the cages sleep with bullets, bars and stone, they do not see your road to freedom that you build with flesh and bone.
braner@batcomputer.tn.cornell.edu (Moshe Braner) (11/25/88)
[response to doubts expressed about posted benchmark] I've seen compiler differences of 100-fold before, using the same program on the same hardware. In this case (a ray-tracing program) I would guess that the key compiler feature is the trigonometry library, since the T800 FPU only does +-*/ plus sqrt (sort of). I have seen rather naive trig libs using simple algorithms written in C, and, in contrast, good algorithms hand-optimized in assembler. The Logical Systems math library is of the latter sort, and is the fastest there is on the T800 as far as I know. So the quoted 25-fold advantage (over a beta version of an unnamed compiler) seems believable to me. BTW for graphical rendition purposes single precision is frequently adequate and a compiler that allows breaking the standard C rule of doing everything in double wins if it can do single precision faster. A library that does it even faster in half-precision can also be useful. - Moshe Braner "compiled BASIC can be faster than interpreted C" "Spaghetti code is possible even in Pascal"