1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA1 |
3 |
|
4 |
I have been running this kernel for about a week on an SB1000-SMP (1x900, |
5 |
1x750) (sparc64) system and on a U60-SMP(2x450) system. I have just begun |
6 |
a test on U2(2x400) system. So far, on SB1000 and U60, this kernel seems |
7 |
as stable as anything we have seen. On U2, it is too early to tell, but |
8 |
since NO 2.6.xx kernel has been stable on U1/U2 systems, I do not hold out |
9 |
much hope there. |
10 |
|
11 |
The reason for any sort of note mentioning this is the kernel's execution |
12 |
characteristics on SB1000. Here there seems to be a good |
13 |
news/you-might-want-a-local-mod news situation. |
14 |
1. Good news: For me, at any rate, up until now programs (occasionally |
15 |
gcc, but also the build of dev-lisp/sbcl-0.9.9 always, sometimes xterm) |
16 |
sometimes failed with unpredictable BUS errors, SegFaults, and similar. |
17 |
(Frequency was about once a day or so). I have not seen this at all for a |
18 |
week, now, sbcl-0.9.9 builds fine, etc. |
19 |
|
20 |
2. Not so good news: With this version of the kernel, kenvctrld is very |
21 |
intrusive. Well, so what and what is it, anyway? kenvctrld is a kernel |
22 |
module which periodically (literally) takes the system's temperature at a |
23 |
couple spots, and if necessary kicks up the fan speed to keep the CPUs |
24 |
from melting (and slows fans back down once temperature is lowered, or |
25 |
eventually shuts the system down if things get worse). To be fair to this |
26 |
kernel, I do not think kenvctrld actually managed to run with previous |
27 |
kernels, but with this series, it has been restructured and is most |
28 |
certainly does run. |
29 |
|
30 |
It runs every five seconds. It takes 2 seconds with one CPU running at |
31 |
100% in the system to actually read the sensors. So, with a 2-CPU system, |
32 |
40% of CPU0 is dedicated to figuring out how hot the system is. |
33 |
|
34 |
Well, that is pretty extreme. But wait! There's more! If your system |
35 |
happens to be doing some actual work (like compiling something), and if |
36 |
you happen to be doing interactive work using X besides, then during this |
37 |
2-out-of-every-five second interval, your display gets little or no |
38 |
service. (E.g., you have things like a 2-second type-ahead on the |
39 |
keyboard, you move your mouse and nothing happens, things like xosview |
40 |
stop. You get the idea). |
41 |
|
42 |
There are two ways to make this a little better. First, you can build |
43 |
kenvctrld as a module (CONFIG_BBC_I2C=m) and not load it. Then kenvctrld |
44 |
does not run and the fans always run at top speed. If your system is in a |
45 |
different room from you, this might be tolerable. If it isn't, with this |
46 |
option you will go deaf. |
47 |
|
48 |
Second, you can play with the polling interval. In the source module |
49 |
drivers/sbus/char/bbc_envctrld.c, there is this definition for the polling |
50 |
interval: |
51 |
#define POLL_INTERVAL (5 * 1000) |
52 |
With this definition, you lose about 40% of one CPU (CPU0). I have found |
53 |
that I can tolerate |
54 |
#define POLL_INTERVAL (5 * 1000 * 18) |
55 |
which changes the interval to 90 seconds. This seems too long, but if I |
56 |
am correct about earlier kernels, it is better than the "never run" we had |
57 |
earlier. (Or, it could be that earlier kernels did take that 2 seconds |
58 |
out of every five without killing interactive performance while doing so.) |
59 |
|
60 |
Hope this is useful, |
61 |
Regards, |
62 |
Ferris |
63 |
- -- |
64 |
Ferris McCormick (P44646, MI) <fmccor@g.o> |
65 |
Developer, Gentoo Linux (Devrel, Sparc) |
66 |
-----BEGIN PGP SIGNATURE----- |
67 |
Version: GnuPG v1.4.2 (GNU/Linux) |
68 |
|
69 |
iD8DBQFD69xTQa6M3+I///cRAlo8AKCxwsDRvF/O8Hy8oJWIDvDMXfHUdgCgtBRo |
70 |
5Nir5e0eVpzjcXesW1K/JS8= |
71 |
=Bmb2 |
72 |
-----END PGP SIGNATURE----- |
73 |
-- |
74 |
gentoo-sparc@g.o mailing list |