RDROYA01@ULKYVX.BITNET.UUCP (01/16/87)
I couldn't resist this one. Someone recently flamed Alcyon for being
slow and producing bloated code. He compared it to a MAC compiler.
Well, I tried his example and instead of the 9K program size he quoted
and the 20+ second compile time (ramdisk) I got the following:
the program main(){}
program size 195 bytes
compile and link time on a hard disk (includes relmod) 16.6 seconds
The program size beats lightspeed by over 1K. I suggest that a number
of Alcyon users have yet to delve into ways of optimizing its setup,
such as customizing the startup file, spreading the source, temp files
and programs across three disks (a dramatic time saver), and linking
in stub routines to turn of floating point conversion, binary files,
direct tty input and the like. At least that's the only way I can
explain the times I've seen quoted. I also refer any interested
parties to the C code comparison ("680xx Computers: Where are They
Going") in the most recent DDJ (pp. 16-18) where Lightspeed was, on
average, 50% slower than Alcyon or Megamax, (I know, the MAC is slower
on its own.). BTW Alcyon tied Megamax on one test, bettered it on two
of them, and came in second to it on the Sieve.
Does megamax support structure assignment and passing, bitfields,
unsigned chars, or unsigned longs? (That's an honest question, not
sarcasm.)