Re: [Gcc-cfarm-users] postgres on gcc112

2016-03-10 Thread David Edelsohn 
On Thu, Mar 10, 2016 at 9:49 PM, Noah Misch  wrote:
> On Mon, Feb 08, 2016 at 11:40:55PM -0500, Noah Misch wrote:
>> On Mon, Feb 08, 2016 at 06:44:45PM -0500, David Edelsohn wrote:
>> > I have killed the postgres benchmark running on gcc112 and locked out
>> > the account.  Contact me if you want it restored.
>>
>> I would like that.
>
> Would you unlock the "nm" account on gcc112?

I have unlocked the account.

I will not hesitate to lock accounts that monopolize the system.

- David

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

Re: [Gcc-cfarm-users] postgres on gcc112

2016-03-10 Thread Noah Misch 
On Mon, Feb 08, 2016 at 11:40:55PM -0500, Noah Misch wrote:
> On Mon, Feb 08, 2016 at 06:44:45PM -0500, David Edelsohn wrote:
> > I have killed the postgres benchmark running on gcc112 and locked out
> > the account.  Contact me if you want it restored.
> 
> I would like that.

Would you unlock the "nm" account on gcc112?

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

Re: [Gcc-cfarm-users] postgres on gcc112

2016-02-12 Thread Markus Trippelsdorf 
On 2016.02.12 at 08:58 +0100, Bastian Bittorf wrote:
> * Stefan Ring  [10.02.2016 11:41]:
> > Unfortunately, process priority is absolutely worthless on machines
> > with hyperthreading. The other day I witnessed someone using all 64
> > virtual cores of gcc110 (the POWER7 machine), and it was godawfully
> > slow to work with.
> > 
> > My rule of thumb is to leave at least one physical core free in order
> > to not disturb other users. So if the machine is 8x8 core (8 physical,
> > 8x HT -- I don't know the real topologies of these POWER machines), I
> > strive to use at most 7 cores.
> 
> this is interesting. till now i always invoked a build with:
> 
> #!/bin/sh
> jobs=$( grep -sc ^'processor' /proc/cpuinfo )
> jobs=$(( $( cpu_count ) + 1 ))
> 
> make --jobs $jobs
> 
> so should i change that to -2 ?

You should change that to "/ 2" at least.

From https://gcc.gnu.org/wiki/CompileFarm
«For automatic jobs on N-core please launch no more than N/2 runnable
processes (total)«

-- 
Markus

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

Re: [Gcc-cfarm-users] postgres on gcc112

2016-02-12 Thread Bastian Bittorf 
* Markus Trippelsdorf  [12.02.2016 09:10]:
> You should change that to "/ 2" at least.
> 
> From https://gcc.gnu.org/wiki/CompileFarm
> «For automatic jobs on N-core please launch no more than N/2 runnable
> processes (total)«

thanks, done - sorry!

bye, bastian

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

Re: [Gcc-cfarm-users] postgres on gcc112

2016-02-10 Thread Stas Boukarev 
On Wed, Feb 10, 2016 at 1:34 PM, Stefan Ring  wrote:
> On Wed, Feb 10, 2016 at 4:44 AM, Noah Misch  wrote:
>> Killing them was the right decision.  Thank you.  Like all my cfarm batch
>> testing, the processes had setpriority(PRIO_MAX).  Perhaps I/O load remained
>> high enough to ruin things.
>
> Unfortunately, process priority is absolutely worthless on machines
> with hyperthreading. The other day I witnessed someone using all 64
> virtual cores of gcc110 (the POWER7 machine), and it was godawfully
> slow to work with.
Sorry about that. But it did allow me to trigger the bug I was looking for.

-- 
With best regards, Stas.

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

Re: [Gcc-cfarm-users] postgres on gcc112

2016-02-10 Thread Stefan Ring 
On Wed, Feb 10, 2016 at 4:44 AM, Noah Misch  wrote:
> Killing them was the right decision.  Thank you.  Like all my cfarm batch
> testing, the processes had setpriority(PRIO_MAX).  Perhaps I/O load remained
> high enough to ruin things.

Unfortunately, process priority is absolutely worthless on machines
with hyperthreading. The other day I witnessed someone using all 64
virtual cores of gcc110 (the POWER7 machine), and it was godawfully
slow to work with.

My rule of thumb is to leave at least one physical core free in order
to not disturb other users. So if the machine is 8x8 core (8 physical,
8x HT -- I don't know the real topologies of these POWER machines), I
strive to use at most 7 cores.

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

Re: [Gcc-cfarm-users] postgres on gcc112

2016-02-09 Thread Noah Misch 
On Tue, Feb 09, 2016 at 06:02:34AM -0500, David Edelsohn wrote:
> The program was consuming all resources on the system preventing
> others from working.  I received numerous messages of thanks from
> other users on the system after I killed the processes.

Killing them was the right decision.  Thank you.  Like all my cfarm batch
testing, the processes had setpriority(PRIO_MAX).  Perhaps I/O load remained
high enough to ruin things.

I have, since 2015-07, other resource-intensive batch testing ongoing within
cfarm machines.  While I took steps from the start to reduce harm to other
users, I am interested in reducing it further.  If any of you experience
unacceptable performance while a user=nm process is running, please contact me
with the symptoms you witness.

nm

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

Re: [Gcc-cfarm-users] postgres on gcc112

2016-02-08 Thread Noah Misch 
On Mon, Feb 08, 2016 at 06:44:45PM -0500, David Edelsohn wrote:
> I have killed the postgres benchmark running on gcc112 and locked out
> the account.  Contact me if you want it restored.

I would like that.  What problems did those processes cause?

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users

[Gcc-cfarm-users] postgres on gcc112

2016-02-08 Thread David Edelsohn 
I have killed the postgres benchmark running on gcc112 and locked out
the account.  Contact me if you want it restored.

- David

___
Gcc-cfarm-users mailing list
Gcc-cfarm-users@gna.org
https://mail.gna.org/listinfo/gcc-cfarm-users