Re: 2.6.14-rt1

[Florian Schmidt <mista.tapas@gmx.net>]

On Tue, 1 Nov 2005 23:05:09 -0500
Carlos Antunes <cmantunes@gmail.com> wrote:

> Here's the thing:
> http://www.nowthor.com/OpenPBX/timing.c
> 
> Let me know what kind of results you get.

Hi,

running the code i simply get:

~$ ./timing 
Failed to create thread # 382
Failed to create thread # 383
Failed to create thread # 384
Failed to create thread # 385
Failed to create thread # 386
Failed to create thread # 387
..
Failed to create thread # 498
Failed to create thread # 499

and then

Segmentation fault

Probably caused by not handling that some threads didn't get created. I
reduced the number down to 300:

~$ ./timing 
Histogram
Delay(ms)       Count
 0              300000
 1              0
 2              0
 3              0
 4              0
 5              0
 6              0
 7              0
 8              0
 9              0
10              0
11              0
12              0
13              0
14              0
15              0
16              0
17              0
18              0
19              0
20              0
Num threads = 300
Total sleeps = 300000
Min error = 0.014 ms
Max error = 0.133 ms

What would you expect to see? BTW: cpu load stayed moderately small with
this setting here.

As a sidenote: Of course the scheduling works completely different with
hundreds of threads running SCHED_FIFO at the same priority than with
hundreds of threads running SCHED_OTHER. SCHED_OTHER threads can preempt
each other when the dynamic priority changes. SCHED_FIFO threads OTOH
just sit and wait until they get to run, not preempting other SCHED_FIFO
threads of the same priority. So basically each SCHED_FIFO thread waits
until all others have run.

Dunno, if that would make a difference in this case though..

Flo

-- 
Palimm Palimm!
http://tapas.affenbande.org
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/