info-vax@ucbvax.ARPA (05/16/85)
From: hplabs!epicen!lee@BERKELEY (Lee Bell) We've recently installed VMS 4.0 and find that some of our signal processing programs run 50%, or more, slower than they did, in terms of CPU time. There's no I/O; it's straight convolution. It didn't seem to matter whether we ran the old images, or recompiled them, but we haven't installed the new FORTRAN compiler yet. Even MACRO programs seem to run substantially slower. Any ideas?
info-vax@ucbvax.ARPA (05/16/85)
From: msev%phobos@cit-hamlet.arpa >We've recently installed VMS 4.0 and find that some of our signal processing >programs run 50%, or more, slower than they did, in terms of CPU time. There's >no I/O; it's straight convolution. Well, you might check to see if your floating point accelerator is turned on. We once were worried about a factor of 2 slowness on a 750, and it turned out we hadn't even *ordered* an FPA. -Martin Ewing USPS: Radio Astronomy, Caltech MS 105-24, Pasadena, CA 91125 ATT: 818-356-4970 CITnet: mse @ phobos ARPAnet:mse%phobos @ CIT-HAMLET BITnet: mse@caltech 2 mtrs: AA6E 147.12 MHz (+0.6) Radio Astronomy DECnet: PHOBOS::MSE