|
|
| Author |
Message |
| 25 new of 283 responses total. |
tao
|
|
response 250 of 283:
| 
Dec 4 18:08 UTC 1997 |
We'll give that idea all the attention it deserves, Dave.
|
tsty
|
|
response 251 of 283:
|
Dec 4 22:55 UTC 1997 |
<<response(s) quashed due to, ummm, errrr .... the phase of the mountain>>
|
mdw
|
|
response 252 of 283:
|
Dec 4 23:49 UTC 1997 |
Slicing pty's does not necessarily reduce system load. In fact,
the sad fact is that pty's are a relatively minor component of
overall system load. The other day, I found a user at another site
who had generated over 7000 finger queries over the course of 2.34 days.
This is one of the things that had contributed to a load average of 20,
by causing everything else to back up waiting for CPU time.
|
nephi
|
|
response 253 of 283:
|
Dec 5 03:11 UTC 1997 |
Sometimes you guys bewilder me. Grex is faster for me now that it has
ever been! I'm really suprised at how fast a computer can run when its
load average is at 30. 8^)
|
danr
|
|
response 254 of 283:
|
Dec 5 12:50 UTC 1997 |
re #251,252: Perhaps we should then disable fingers?
|
valerie
|
|
response 255 of 283:
|
Dec 5 14:50 UTC 1997 |
This response has been erased.
|
rcurl
|
|
response 256 of 283:
|
Dec 5 18:54 UTC 1997 |
Someone has 7,000 friends on Grex? (Or was he pounding Grex with fingers
for just a few friends - with a program, I presume?)
I have had the Y - "print to local computer" - command in pine end up
crashing my computer. It works OK from CAEN, but from Grex the print
starts to spool, and then just freezes, This is with connection via
the internet. I tried it with two different printer drivers, but the
same thing happens.
|
dpc
|
|
response 257 of 283:
|
Dec 5 23:39 UTC 1997 |
Well, what *can* be done (within reason) to keep the load average
below, say, 10? Now it's over 20 and the System is practically
unuseable.
|
kaplan
|
|
response 258 of 283:
|
Dec 5 23:57 UTC 1997 |
Re 256, Print to local computer uses escape codes to tell your term
program to control your printer. Check the version numbers and other
settings of pine on grex and caen. There must be some important
difference.
|
mcnally
|
|
response 259 of 283:
|
Dec 6 01:57 UTC 1997 |
Could also be a termcap difference, if Grex thinks your termcap
is different it could be sending different escape sequences..
|
aruba
|
|
response 260 of 283:
|
Dec 6 03:55 UTC 1997 |
Correct me if I'm wrong, but load average is a measure of the number or jobs
actually executing at one time, right? So if we got a faster computer, the
same load average would actually be faster, right?
|
valerie
|
|
response 261 of 283:
|
Dec 6 14:34 UTC 1997 |
This response has been erased.
|
dpc
|
|
response 262 of 283:
|
Dec 6 19:50 UTC 1997 |
How does the speed of the 670's CPU compare with the machine we're now
running on?
|
remmers
|
|
response 263 of 283:
|
Dec 6 21:54 UTC 1997 |
I don't know what the speed comparison is, but the 670 is a
multi-processor machine with two CPU's. So it will give Grex and
advantage comparable to what a second brain would give a person.
|
mcnally
|
|
response 264 of 283:
|
Dec 7 00:01 UTC 1997 |
I'm not sure a 2nd brain would be all that great an idea for a person.
They're not 100% efficient for computers, either, since there're
coordination and overhead issues involved, but with 2 processors the
670 should be more efficiently able to handle big numbers of users..
|
remmers
|
|
response 265 of 283:
|
Dec 7 02:27 UTC 1997 |
Well, I'd love a 2nd brain to help me multitask. But we digress.
Back to system problems...
|
rcurl
|
|
response 266 of 283:
|
Dec 7 19:20 UTC 1997 |
The problem described in 256, 258 and 258, went away - by itself - on
the next day's login. Weird.
|
janc
|
|
response 267 of 283:
|
Dec 8 00:10 UTC 1997 |
The each of the new computer's CPUs should be about 2-3 times faster than
Grex's current CPU. It is hard really to compare them, because the whole
architectures are significantly different - for example, the new machines have
more hardware contexts so the should be able to shuffle between large numbers
of processes faster. There are currently two of these faster CPU's in the
new system. But they will be interfering with each other to a significant
degree since both cannot be executing kernel code at the same time with our
current software. We can put four CPUs in, but rumor says the system can
actually be slower with four because they start tangling with each other too
much. We'll probably eventually get more CPUs if only to have some spares.
Of course, CPU speed isn't the only thing that determines the speed of hte
machine. Disk access speed is also an issue. It is quite possible that the
performance of the machine will be limited more by disk I/O than CPU speed.
We will be using the same disks we have now, so they won't be getting any
faster. However, we will be using a significantly newer disk controller,
which may well have some performance benefits.
So how much faster will the new machine be? Geez, I don't know. I'm not even
sure about half the information I just repeated. I've been using 4-5 times
faster as a guess, but it could be off in either direction. I've been using
it to compile software, but then I'm mostly the only user on it, and a dual
processor machine doesn't help a single user that much. Many of the
advantages of the 4/670 aren't just raw speed but ability to handle large
number of simultaneous users. So I can't really make even subjective
comparisons.
|
shyam
|
|
response 268 of 283:
|
Dec 8 12:06 UTC 1997 |
i am shyam,i love nature.
|
valerie
|
|
response 269 of 283:
|
Dec 9 02:04 UTC 1997 |
This response has been erased.
|
davel
|
|
response 270 of 283:
|
Dec 9 10:58 UTC 1997 |
Unfortunately, Grex is unnatural.
|
cmcgee
|
|
response 271 of 283:
|
Dec 11 05:23 UTC 1997 |
Erm, why cant I dial in on 3000 or 4931? All I got was ringing.
|
davel
|
|
response 272 of 283:
|
Dec 11 10:53 UTC 1997 |
I'm dialed in on 3000 now ...
|
scott
|
|
response 273 of 283:
|
Dec 11 21:03 UTC 1997 |
The Internet link is down, apparently somewhere upstream of Grex.
|
valerie
|
|
response 274 of 283:
|
Dec 12 02:28 UTC 1997 |
This response has been erased.
|
