Messages in this thread | | | From | "Doug Smythies" <> | Subject | RE: [PATCH] sched: Folding nohz load accounting more accurate | Date | Wed, 13 Jun 2012 20:13:29 -0700 |
| |
>> On Wed, 2012-06-13 at 08:33 -0700, Doug Smythies wrote: > On 2012.06.13 14:58 -0700, Peter Zijlstra wrote:
>> >> All that being said, what I typically do with a new code test is: >> >> . select a known, previous bad operating point. For example 2 >> processes, actual load average 0.30 (0.15 for each process) currently >> reporting ~1.5.
> OK, I'll try and apply this. Waiting 63 hours for feedback on patches is > something I'm not patient enough for.
> Would this be:
> ./waiter 2 900 230608 10000
Actually it would be:
./waiter 2 900 345912 9444
At least on my computer, with the CPUs locked into powersave mode (lowest clock rate). It might be different on your computer as the exact numbers are computer dependent. I will change the script generating program to add comment lines as to the expected execution scenario, as I have troubles also looking up command lines.
> I haven't even bothered reading the waiter proglet yet, but I did notice > the 'help' provided when started without arguments doesn't seem to > actually match what load_180 does.
Right, sorry. I think you will find the newer version (from the "wang" experiment write up) is O.K.
| |