PDA

View Full Version : Sieve client freeze



vjs
07-06-2004, 01:54 PM
I was wondering if anyone has ever had problems with the client freezing, but not locking up the computer?

One box (a dual) isn't overclocked or anything nor overheating etc but for some reason the CPU usage goes to 0% or 50% and I have to restart the client to get it running again. What's weird is sometimes only one client stops.

Sort of happens once every 3 days or so.

Also something strange on a another box that does 700+ kps according to the client...
At this speed it should do 60+ G per day but my calculations are showing around 400kps of actual computational speed (Basically around 35G per day progress).

Just curious if you have to restart the client or something after you access fact.txt or something.

Anyone Anyone....

Similar problems???

Mystwalker
07-06-2004, 03:23 PM
No problems whatsoever - unless I don't mark something (under Windows) and don't copy afterwards. Then, nothing happens...

Nuri
07-06-2004, 04:42 PM
Not the same problem but...


About a month ago, the sieving speed on two of my machines dropped significantly.

A P4-smtng. (notebook) from a steady speed of 184 kp/sec to a steady speed of 137 kp/sec.

An Athlon 2400 from a steady speed of 539 kp/sec to a steady speed of 435 kp/sec.


PS: I do not normally use the P4 for sieving, it's just a backup for P-1.

PS2: At what p levels do you encounter this problem? Are they above or below 2^49 (or both)?

vjs
07-06-2004, 05:55 PM
The problems are occuring on those ranges I've reserved below 2^49.

The three machines that are giving problems are one P3 800, a dual 500 celeron running two clients, a dual 866 P3 also running two clients.

I'd say the problem is with the dual boards getting confused etc, but the single P3 is doing it as well, I've also named the clients cmov.exe and cmov1.exe, but that didn't help.

Is there some wait period or something where the client does nothing CPU drops to zero and it does alot of disk writing etc then starts up again?

The Computer which is I think misrepreseting the rate is a Barton.

All of my machines are w2k running cmov.exe

But FYI on the 2^49 cross over I have two almost identical machines xp1800 (both of these are working fine) with 256mb, only differences really are the chipset on the Motherboards.

The KT333 board is getting ~290Kp/s at 593000-594000 <2^49
The KT266 board is getting ~430kp/s at 350000-351000 >2^49

Not going to do anymore above 2^49 until we actually get there.


No problems whatsoever - unless I don't mark something (under Windows) and don't copy afterwards. Then, nothing happens...

What do you mean by this exactly...

Mystwalker
07-07-2004, 08:03 AM
Originally posted by vjs
The KT333 board is getting ~290Kp/s at 593000-594000 <2^49
The KT266 board is getting ~430kp/s at 350000-351000 >2^49

I have this problem as well. The range at 550T slows down sieving speed to ~50%...


What do you mean by this exactly...

When you mark some part of the shell, but don't copy it afterwards (but leave the selection), proth_sieve won't proceed. Maybe it still takes computing power (I'm not sure about this anymore), but when you end the selection (e.g. by clicking the right mouse button), it will only output the next output line (typically 10M advancement) - regardless of the time passed so far! I've lost a few hours this way. :(

vjs
07-07-2004, 09:57 AM
Wow,

O.K. I'll make sure I don't do that.

mklasson
07-08-2004, 06:34 AM
When you mark some part of the shell, but don't copy it afterwards (but leave the selection), proth_sieve won't proceed. Maybe it still takes computing power (I'm not sure about this anymore), but when you end the selection (e.g. by clicking the right mouse button), it will only output the next output line
This goes for all win32 console programs. I guess the screen output call is getting put on hold by windows until you're done selecting, causing the program to wait in the next printf/cout/whatever statement.

Once you know about it though, it's actually very useful. I've used it numerous times to temporarily pause programs.

vjs
07-09-2004, 09:38 PM
I'm pretty sure this is what's happening,

They have been running no problem lately,

Just start the program and min to the tray works fine.