JakeHillion/scx-upstream
1
0
Fork 0
mirror of https://github.com/sched-ext/scx.git synced 2024-11-24 20:00:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

scx_lavd: more explanation about sched_prio_to_latency_weight

Browse Source 
Signed-off-by: Changwoo Min <changwoo@igalia.com>
This commit is contained in:
Changwoo Min 2024-03-19 21:31:37 +09:00
parent 24bddad9b4
commit fa2282363b
1 changed files with 9 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
scheds/rust/scx_lavd/src/bpf/main.bpf.c
View File

@ -268,13 +268,18 @@ static const u64 sched_prio_to_slice_weight[NICE_WIDTH] = {
* MuQSS scheduler. We choose a different distribution for
* sched_prio_to_latency_weight on purpose instead of inversing
* sched_prio_to_slice_weight. That is because sched_prio_to_slice_weight is
* too steep to use for latency. We normalized the values so that the normal
* too steep to use for latency. Suppose the maximum time slice per schedule
* (LAVD_SLICE_MAX_NS) is 4 msec. We normalized the values so that the normal
* priority (nice 0) has a deadline of 10 msec. The virtual deadline ranges
* from 1.5 msec to 60.9 msec.
* from 1.5 msec to 60.9 msec. As the maximum time slice becomes shorter, the
* deadlines become tighter.Taking an example of 3 msec, the virtual deadline
* ranges from 1.15 msec (for nice -20) to 7.5 msec (for nice 0) and 45.65 msec
* (for nice 19).
*/
static const u64 sched_prio_to_latency_weight[NICE_WIDTH] = {
/* weight nice priority sched priority vdeadline (usec) */
/* ------ ------------- -------------- ---------------- */
/* weight nice priority sched priority vdeadline (usec) */
/* (max slice == 4 ms) */
/* ------ ------------- -------------- ------------------- */
383, /* -20 0 1532 */
419, /* -19 1 1676 */
461, /* -18 2 1844 */