jls2@spdcc.COM (Jeff Stoner) (05/02/91)
Hi all. We have several DECstation 3100s (the RISC boxes) and a VAXStation 3100 (the VAX CPU), all running at 4.0 or 4.1 of Ultrix. I am received complaints that the VAX 3100 is "intolerably slow". I understand that the VAX CPU will be inherently slower than the RISC ones, but is there anything I can do to speed up the VAX? I don't have Ultrix source code, but anything else would be fair game. I am running 4.0 Ultrix "out of the box" (no mods) with DECwindows and a secondary nameserver on the system. There is generally one user on the main keyboard and screen, running maybe 4 DECterms and a few X-things (like xload and xclock). Suggestions? Thanks in advance... -- ====== Jeff L. Stoner === Boulder, Colo. ============ /\ = /\ ========== | Work: jls2 @ ellery.esi.com /\ / \ / /\ /\ --*-- Home: jls2 @ bearhug.boulder.co.us /\/ \/ / / \/ /\ | News: jls2 @ ursa-major.spdcc.com /\/ \ \ /\ / \ /\
alan@shodha.enet.dec.com ( Alan's Home for Wayward Notes File.) (05/02/91)
In article <7453@spdcc.SPDCC.COM>, jls2@spdcc.COM (Jeff Stoner) writes: > > > I am received complaints that the VAX 3100 is "intolerably slow". > I understand that the VAX CPU will be inherently slower than the RISC ones, > but is there anything I can do to speed up the VAX? I don't have > Ultrix source code, but anything else would be fair game. The problem is that the phrase "intolerably slow" contains almost no information about the system performance. It could be any number of things: o The system is out of memory and something as simple as uniconifing a window causes it to page wildly. o There is a serious system problem that is causeing the system to spend large amounts of time in kernel mode. For example; the error logger is busy. o The one user is doing multiple things in multiple windows and is running the system out of; I/O capacity and/or CPU capacity. o The user has a low tolerence for response time and is doing a reasonable amount of work. The problem therefore is to find out WHY the system seems slow, track down the cause to and fix it. Start by using cpustat, vmstat and iostat to look the time spent in various CPU states, the amount of available memory, the amount of pageing, the I/O rates on the disk drives and the load average. If you have Internet access consider picking up a copy of the program Monitor from gatekeeper.dec.com. > > -- > ====== Jeff L. Stoner === Boulder, Colo. ============ /\ = /\ ========== | -- Alan Rollow alan@nabeth.cxn.dec.com
grr@cbmvax.commodore.com (George Robbins) (05/03/91)
In article <7453@spdcc.SPDCC.COM> jls2@spdcc.COM (Jeff Stoner) writes: > > > Hi all. We have several DECstation 3100s (the RISC boxes) and a > VAXStation 3100 (the VAX CPU), all running at 4.0 or 4.1 of Ultrix. > > I am received complaints that the VAX 3100 is "intolerably slow". > I understand that the VAX CPU will be inherently slower than the RISC ones, > but is there anything I can do to speed up the VAX? I don't have > Ultrix source code, but anything else would be fair game. Nobody who has used a DECstation 3100 or Sparc I/II is going to find the VAXstation performance acceptable. Give it to some VMS freaks (where it's pretty sad too 8-) or find some non-critical application for it. -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing: domain: grr@cbmvax.commodore.com Commodore, Engineering Department phone: 215-431-9349 (only by moonlite)
dayhoff@ddnvx1.afwl.af.mil (dayhoff@ddnvx1.afwl.af.mil -- harv) (05/11/91)
In article <3058@shodha.enet.dec.com>, alan@shodha.enet.dec.com ( Alan's Home for Wayward Notes File.) writes: > In article <7453@spdcc.SPDCC.COM>, jls2@spdcc.COM (Jeff Stoner) writes: >> >> >> I am received complaints that the VAX 3100 is "intolerably slow". >> I understand that the VAX CPU will be inherently slower than the RISC ones, >> but is there anything I can do to speed up the VAX? I don't have >> Ultrix source code, but anything else would be fair game. > > The problem is that the phrase "intolerably slow" contains > almost no information about the system performance. It > could be any number of things: > [possibilities deleted] > I must agree, my stock VS3100, 8 Meg, color, local disks, does quite well w/ 4 DECterms and calendar. I was taking a bit of a performance hit due to short swap space and slow RZ23's.... Fixed by adding high speed Wren. I still get impatient at times, but such is life. I do program development and admin. If I made my living at number crunching or supper spiffy graphics the VS would be too slow. > If you have Internet access consider picking up a copy of > the program Monitor from gatekeeper.dec.com. ...Seems as though Monitor takes an unfair ammount of disk space. >> >> -- >> ====== Jeff L. Stoner === Boulder, Colo. ============ /\ = /\ ========== | > -- > Alan Rollow alan@nabeth.cxn.dec.com harv dayhoff@ddnvx1.afwl.af.mil