earle@smeagol.JPL.NASA.GOV (Greg Earle) (04/09/87)
I ran Jonathan Hue's airbrush program on three machines, all lightly loaded (as in, Just Me On Them :-) : Sun-3/280HM, 8 Mbytes, 25 MHz 68020 Sun-3/180S, 4 Mbytes, 16.67 MHz 68020, *diskless* client of 3/280HM Sun-3/160C, 8 Mbytes, 16.67 MHz 68020, local disk Since the `user' times are relevant (easiest way to see this is to note that the `user' times on the diskless and diskful machines are identical, yet the `real' times are .5 to 1.0 seconds slower on the diskless), I thought I'd post (all using `cc -O', SunOS 3.3 BTW) : Sun-3/280HM 14.4 Sun-3/160C 23.3 Sun-3/180S 23.4 Since the 3/280 runs at 1.5 times the clock speed, and the added 64k write-back cache helps performance a bit, the ~1.6 x performance increase seems right in line with what should be expected. Load was about 0.20 (no, I'm not going to go to single user mode just to test a benchmark!). The figure for the 3/280HM when *not* using -O was around 17.6, I believe; it appears Sun have hired some reasonably adept compiler optimizers ... Be interesting to see the figures for a MIPS-500, a machine I am particularly fond of (alas! but for a big bitmapped terminal, perhaps it could compete with Suns for our spacecraft test systems of the future - John, are you listening?) but do not own any of ... -- Greg Earle UUCP: sdcrdcf!smeagol!earle; attmail!earle JPL ARPA: elroy!smeagol!earle@csvax.caltech.edu AT&T: +1 818 354 4034 earle@jplpub1.jpl.nasa.gov (For the daring) I think my career is ruined!