[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re[4]: kernel compilation time



Meir Litmanovich <meir@xpert.com> wrote:

>  > I don't understand the meaning of that "+1" (unless this is a way to
>  concur
>  > with another CPU-intensive process running simultaneously on the same box
>  -
>  > but then why not just SIGSTOP or renice it for awhile??).
>  
>  Oh, not. Just imagine your computer does nothing except for
>  kernel compilation (that's the situation, right ? ).
>  Then you really don't want to wait for I/O on this
>  machine. N+1 ensures that in time one gcc would stuck waiting
>  for I/O, other will effectively run. 

But why would one wait for I/O during compilation on an otherwise idle
machine? Only if one has an AT disk or something similar by I/O performance,
like NFS over very busy ethernet. Or, if the box is heavily swapping - but
then an extra compilation process wouldn't help :). But in normal situations
this shouldn't happen, IMO. At least, I've never seen any difference between
compile times of make -j N and make -j N+1 (or, for that matter, N+k for any k
> 0 - till the RAM is exhausted). Tried on several uni-processor and SMP
boxes.

Regards,

Evgeny


--
   ____________________________________________________________
  / Evgeny Stambulchik  <fnevgeny@plasma-gate.weizmann.ac.il>  \
 /  Plasma Laboratory, Weizmann Institute of Science, Israel \  \
 |  Phone : (972)8-934-3610  == | == FAX   : (972)8-934-3491 |  |
 |  URL   :    http://plasma-gate.weizmann.ac.il/~fnevgeny/  |  |
 |  Finger for PGP key >=====================================+  | 
 |______________________________________________________________| 

=================================================================
To unsubscribe, send mail to linux-il-request@linux.org.il with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail linux-il-request@linux.org.il