Differences

This shows you the differences between two versions of the page.

--- wiki:using_hyperthreading_on_nodes [2015/07/05 10:09] – neyron
+++ wiki:using_hyperthreading_on_nodes [2015/07/05 10:29] – [On using hyper-threads as overbooking processors] neyron
@@ Line 26: / Line 26: @@
 ====== On using hyper-threads as overbooking processors ======
+The idea here, is to setup some admission rules in order to steer some jobs to use the 2 threads of the cores when some conditions are met (small jobs, jobs for some known computation types for which hyperthreading is relevant...).
 To be completed.
@@ Line 37: / Line 40: @@
 This is an issue if we want to consider disabling the 2nd thread of only some cores of a host, not all. Otherwise indeed, disabling all second threads or re-enabling all of them is straight-forward using a for loop and looking at /sys/devices/system/cpu/possible.
-We are discussing a way for handling that in OAR, by maybe changing the value (and type :( ) of the cpuset field in the resource table to for instance "0-8" instead of just 0 (in the case of the number of the 2nd thread of core 0 being 8), and adapting the job resource manager to handle that value accordingly to an HT job type.
+For that purpose, the cpuset field of OAR was changed to string with OAR 2.5.4+. This allows setting the value of the cpuset field in the resource table to,for instance, "0,8" instead of just 0 (core 0 has threads with logical cpu id 0 and 8). Then the job resource manager can handle that value very easily whenever an HT job type is in use by a job.
-Another option is to have the job resource manager look at a configuration file located on every nodes providing the information.
- --- //[[pierre.neyron@imag.fr|Pierre Neyron]] 2014/02/21 12:47//
+ --- //[[pierre.neyron@imag.fr|Pierre Neyron]] 2015/07/05 10:14//