jonas@lkbpyr.UUCP (Jonas Heyman) (06/13/91)
Hello, I'm coming from a Pyramid environment, there we had both the System V and BSD tools for system performance checking such as sar,vmstat,iostat and more. We also had two programs called top and mon ported to our platform, are these two programs available for Aviion 4000 ? Any information on this is most welcome, if you have the programs could you please send them to me. Thanks in advance, Jonas. (If there is interest I will send a summary to this group). -- /jonas@lkbpyr.lkb.se/
tom@ssd.csd.harris.com (Tom Horsley) (06/14/91)
One of my biggest objections to DG/UX is the lack of tools like this
(personally, I would like to see 'top' more than most of the others).
Under DG/UX however, there are some potential tools waiting to happen. Check
out the man pages for the system services that start off with the string
'dg_'. In particular the dg_sys_info service provides access to a lot of the
information you need to print performance info for the system as a whole and
dg_process_info provides info about the running processes.
If anyone has ported any of the public domain performance tools so they use
these services, I would like to hear about it (probably a lot of other
people would as well).
--
======================================================================
domain: tahorsley@csd.harris.com USMail: Tom Horsley
uucp: ...!uunet!hcx1!tahorsley 511 Kingbird Circle
Delray Beach, FL 33444
+==== Censorship is the only form of Obscenity ======================+
| (Wait, I forgot government tobacco subsidies...) |
+====================================================================+newton@gumby.cs.caltech.edu (Mike Newton) (06/15/91)
tom@ssd.csd.harris.com (Tom Horsley) writes: >One of my biggest objections to DG/UX is the lack of tools like this >(personally, I would like to see 'top' more than most of the others). >... >If anyone has ported any of the public domain performance tools so they use >these services, I would like to hear about it (probably a lot of other >people would as well). Well.... They are _very_ crude, but I have some tools that were useful on several occasions (eg: "why do char.s get droped at 9600 baud", "what the ____ is X doing", ...). They are are all small, and many of them just play w/ the IO devices. k-params -- print out your environment (the 88k one, not *environ[]!) dg_info -- an EXTREMELY crude system profiler. start it, start your job/run/tty/... and then wait for the time out (I said it was crude). Get more useless statistics than you can imagine. Change the time appropriately, or better yet, make it a command line arg, or even better, fork off the thing to run... kbd-read -- so just why does my keyboard go to IBM mode 2... (reads raw chars) kbd-reset -- if you use Xmodmap to switch caps/cntrl lock, or you run 9600 baud on older AViiON's, you can have your keyboard lock. Put this in the main (root level) X popupwindow (w/ all modifiers). Then you dont have to reboot... mousie.c / move-mouse.c / mouse-advise.c -- like kbd-read and advise advise -- the infamous program (Someone took a copy of my 15 minute hack and used it as an example in a conference paper on 'why unix programs are bad'). Great for stuck ptys and such rot. An old version is on csvax.cs.caltech.edu in pub/M88K. The name is something like dg-tools or dg-io (.t.Z). If anything's missing, send me a note and i'll put newer versions on. Again: they are crude hacks. Most were written in 5-15 minutes to try and find out what was happening / fix something. One was written by taking sed to the right /usr/include files... . However, I've ended up using them often (and have absolutely no time to improve them). - mike ps: also note that DG/Motorola have 'm88kscheduler'and 'pipesim' avail. ps: i've also written part of a better system spier. however i've never been able to figure out how to read IO space (negative addresses) so it was never finished. (Neither reading kmem nor mem will work -- they do not include 0xf..... type addresses.) If anyone knows how, tell me and sooner or later (more likely 'later') i'll finish the new tools
gary@proa.sv.dg.com (Gary Bridgewater) (06/16/91)
In article <TOM.91Jun14065821@rcx1.ssd.csd.harris.com> tom@ssd.csd.harris.com (Tom Horsley) writes: >If anyone has ported any of the public domain performance tools so they use >these services, I would like to hear about it (probably a lot of other >people would as well). Many people in DG have or had Suns and have seen top. Most of them would like it on DG/UX. The problem with a straight port is that top is written very much around having access to /vmunix and the various BSDish structs so there is no hierarchy (e.g. get_process_stuff() { get_memory_usage(); get_cpu_usage..}). Also, the information that is processed is orthagonal to what is available from the DG/UX dg_ calls. Consequently, you have to re-write most of it. At that point you realize you might as well make it X-based (well, MOTIF based anyway) so you would end up writing a completely new utility. As long as you are doing all that you might as well make some of the info graphic (bar charts or speedometers, etc.) and on and on. This is getting into some serious development time so it is typically beyond the "spare time" level. That means it is up to the real tools folks (which I am in no way associated with) so an STR with a formal request becomes necessary to get the marketing folks to approve such a thing. Internal STRs don't cut as much ice with marketing as those from paying customers. {insert your conclusion here} If you disagree with my analysis then print the sources out and read through them. This is in no way to disparage top - it is very useful in its milieu but it is useless, as such, under DG/UX other than as a model. -- Gary Bridgewater, Data General Corporation, Sunnyvale California gary@sv.dg.com or {amdahl,aeras,amdcad}!dgcad!gary "I am a pizza. I am a pizza. ..."
ssmith@merengue.mpr.ca (Shaun Smith) (06/18/91)
In article <1991Jun16.081435.20572@proa.sv.dg.com> gary@proa.sv.dg.com (Gary Bridgewater) writes: At that point you realize you might as well make it X-based (well, MOTIF based anyway) so you would end up writing a completely new utility. As long as you are doing all that you might as well make some of the info graphic (bar charts or speedometers, etc.) and on and on. Funny you should mention X! We've recently got an AViiON and have had nothing but frustration getting help from our DG representative. We have concerns over X performance and the best they can do is say "tune your application" instead of helping us tune the installation. As a result we get our applications running faster on our SUNS and DECS, but are left with slightly faster but still much slower on the AViiON X applictions! Aaaargggg! Ok. I'm calm. This is my first time in comp.sys.m88k so please excuse me if these issues have been raised before. We want to run a particular X application we have developed on an AViiON. X applications seem to run slowly. In fact, everything seems to be slower on the AViiON than on my SPARCstation 1+. We expected better. Today I ran the X application "xengine" on 3 platforms. I displayed the image on the console and on a remote display (not at the same time). Here's what I found: \Display | SUN4* AViiON** DEC*** \ | CPU \ | -------------------------------------------------------- | SUN4 | 257 68 656 | AViiON | 225 70 630 | DEC | 225 68 715 *: greyscale SPARCstation 1+, SunOS 4.1, 16MB, MIT X11R4 **: colour Data General AViiON, DGUX 4.2, single CPU 16MB, DG supplied X11R4 ***: DECstation 5000 with colour graphics accelerator, Ultrix 4.2, 24MB, MIT X11R4 What I deduce from the above data is that for xengine, the AViiON's X server is nearly 3 times slower than on my Sparc and nearly 10 times slower than on the DECstation described above. The DECstation is fast I admit, but the difference in speed between the Sun and DG is very large. I won't speculate right now, but I'd sure like to get some feedback on what is going on. As I said above, our local DG reps have been very unhelpful. Any comments would be appreciated. Shaun (ssmith@mprgate.mpr.ca) -- Shaun M. Smith | ssmith@mprgate.mpr.ca MPR Teltech Ltd. | mprgate.mpr.ca!ssmith@uunet.uu.net 8999 Nelson Way, Burnaby, BC | ssmith%mprgate.mpr.ca@relay.ubc.ca Canada, V5A 4B5, (604) 293-5345 |
tom@ssd.csd.harris.com (Tom Horsley) (06/18/91)
I recently uggraded my AViiON from DG/UX 4.31 to 4.32 and the X server
definitely *seems* much faster (this is purely subjective, I ran no
benchmarks). If you are not running 4.32, you might want to upgrade
and try your X tests again.
P.S. I have a monochrome AViiON, so even if I had run benchmarks they
probably wouldn't compare with the color display.
--
======================================================================
domain: tahorsley@csd.harris.com USMail: Tom Horsley
uucp: ...!uunet!hcx1!tahorsley 511 Kingbird Circle
Delray Beach, FL 33444
+==== Censorship is the only form of Obscenity ======================+
| (Wait, I forgot government tobacco subsidies...) |
+====================================================================+tom@ssd.csd.harris.com (Tom Horsley) (06/18/91)
gary> If you disagree with my analysis then print the sources out and read
gary> through them. This is in no way to disparage top - it is very useful
gary> in its milieu but it is useless, as such, under DG/UX other than as a
gary> model.
Oh, I don't disagree, that's why I was hoping someone else had done it
already :-).
If it was a trivial port, I would have just done it myself...
--
======================================================================
domain: tahorsley@csd.harris.com USMail: Tom Horsley
uucp: ...!uunet!hcx1!tahorsley 511 Kingbird Circle
Delray Beach, FL 33444
+==== Censorship is the only form of Obscenity ======================+
| (Wait, I forgot government tobacco subsidies...) |
+====================================================================+