newwave@desire.wright.edu (07/09/90)
Is it true that the VAX 8650 outperforms the VAX 8200 6 times. Why is it that a VAX 8200 outperforms the VAX 8650 10 times on a UNIBUS write cycle? What can be done to improve the 8650 performance. (Just a few questions for those DEC pros out there.) Thanking you in advance. newwave@desire.wright.edu K. Desai
don@zl2tnm.gp.govt.nz (Don Stokes) (07/11/90)
newwave@desire.wright.edu writes: > Is it true that the VAX 8650 outperforms the VAX 8200 6 times. > > Why is it that a VAX 8200 outperforms the VAX 8650 10 times on a UNIBUS > write cycle? Because it isn't UNIBUS. It's VAXBI. I've seen 8200s go to 100% cpu utilization for long periods while doing backups -- the bus simply isn't the bottleneck..... (this was in older versions of VMS BACKUP, before I/O efficiency was improved....). I've never seen UNIBUS connected peripherals do that to a machine. Not even 11/750s. > > What can be done to improve the 8650 performance. You could attach a CI and an HSC. It'll cost $$$ though. If the 86xx series includes a MASSBUS (does it? never met one), get MASSBUS devices ... until fairly recently (ie latest CI controllers), MASSBUS connected RM05s were easily the fastest things on spindles if driven properly. Seek times weren't as good as RA disks, but sheer throughput........ Keep the disk defragmented, use a large cluster size and watch it fly. (I only ever used then on PDP-11/70s -- 24k block I/O to RM05s meant *incredible* speed if you knew what you were doing.) Bear in mind that RM05s are an old drive; you might be able to pick them up very cheap second hand. They're big, power hungry removable pack jobs. Be prepared to have to move them back against their controller cabinet periodically -- they walk....... Don Stokes, ZL2TNM / / Home: don@zl2tnm.gp.govt.nz Systems Programmer /GP/ Government Printing Office Work: don@gp.govt.nz __________________/ /__Wellington, New Zealand_____or:_PSI%(5301)47000028::DON