tps@chem.ucsd.edu (Tom Stockfisch) (07/27/89)
The nice command of csh (on the 4D series machines, at least) does not
nice things very far. nice +19 is supposed to be the maximum niceness,
but the split in cpu time between a job so nice'ed and a job with
not niced is only 20/80. That is if you have one nice +19 job and
one not-nice'ed job, 20% of the cpu time still goes to the nice'ed job.
I would much prefer that the split be more like 2/98. Does anyone know
if there is a way to reconfigure things so that this is possible?
--
|| Tom Stockfisch, UCSD Chemistry tps@chem.ucsd.edu
HCART@VAX.OXFORD.AC.UK (08/27/90)
When I nice(1) a job on my 3130, the command seems to have no effect, and the job runs at the same priority as the interactive session whatever the increment used. However, if I issue a nice(1) command as super-user the job is pushed deep into the background as required. Why is this? And how can I change things so that users can nice their own jobs without being super-user? Hugh Cartwright. Physical Chemistry, Oxford University, UK. (HCART@vax.ox.ac.uk or hmc@physchem.ox.ac.uk)
corkum@csri.toronto.edu (Brent Thomas Corkum) (12/14/90)
A short question regarding the nice command. Does it work? What I notice is that the priority of a job submitted using nice, the priority based on what ps -el shows me, does not differ if I use nice. I tried all sorts of increment numbers (see man nice). I also couldn't use negative increment numbers if I was super user, I just get an error. I also take it that the larger the increment the lower the priority. What is the default increment for nice, and what are the default priorities without nice? Brent Corkum corkum@boulder.civ.toronto.edu