diff options
Diffstat (limited to 'posts/jemalloc-hpa-reference')
| -rw-r--r-- | posts/jemalloc-hpa-reference/jemalloc-hpa-reference.md | 25 |
1 files changed, 12 insertions, 13 deletions
diff --git a/posts/jemalloc-hpa-reference/jemalloc-hpa-reference.md b/posts/jemalloc-hpa-reference/jemalloc-hpa-reference.md index 08d45f8..a210b34 100644 --- a/posts/jemalloc-hpa-reference/jemalloc-hpa-reference.md +++ b/posts/jemalloc-hpa-reference/jemalloc-hpa-reference.md @@ -9,9 +9,8 @@ application performance. ### Pageslab Pageslab is hugepage aligned and sized memory range. You can think of it as a -set of pages packed together into a hugepage. Pageslab is not necessary backed -up by a hugepage. - +set of pages packed together into a hugepage. Pageslab is not necessarily +backed up by a hugepage. ### Active Page @@ -81,9 +80,9 @@ Maximum value: `HUGEPAGE` bytes. Minimum number of active bytes in a pageslab necessary for pageslab to be placed into hugification queue. -Pageslab always produced in a non-huge state. Over time, when number of active -bytes became greater or equal than `hpa_hugification_threshold`, jemalloc puts -pageslab into hugification queue. +Pageslab is always produced in a non-huge state. Over time, when the number of +active bytes becames greater or equal than `hpa_hugification_threshold`, +jemalloc puts pageslab into the hugification queue. Unsigned integer. Default value: `0.95 * HUGEPAGE` bytes. Minimum value: `PAGE` bytes. Maximum value: `HUGEPAGE` bytes. @@ -121,8 +120,8 @@ Use `madvise(..., MADV_COLLAPSE)` to request OS back up pageslab by a hugepage alongside `madvise(..., MADV_HUGEPAGE)`. Increments `stats.arenas.<i>.hpa_shard.nhugify_failures` counter on failure. -Usual asynchronous hugification introduces delay of unknown length, between -request to OS has been made to hugify a pageslab and OS actually backs up +Usually asynchronous hugification introduces delay of unknown length, between +request to the OS being made to hugify a pageslab and the OS actually backs up pageslab by a hugepage. This option allows to eliminate this delay. Requires Linux 6.1 or higher. @@ -133,10 +132,10 @@ Boolean. Default value: `false`. Minimum time between two consecutive purge phases in milliseconds. Each `hpa_min_purge_interval_ms` jemalloc will check if purging criteria are -met and if they are, it will purge as much pageslabs as needed until purging +met and if they are, it will purge as many pageslabs as needed until purging criteria are no longer met. Minimal unit of purging is pageslab, meaning all -dirty pages will be returned back to the OS from chosen pageslab, even if less -pages required to be purged to reach purging target. If there are few +dirty pages will be returned back to the OS from chosen pageslab, even if fewer +pages are required to be purged to reach purging target. If there are few consecutive dirty pages, one syscall will be issued to purge them together in one go. @@ -180,8 +179,8 @@ when `hpa_peak_demand_window_ms` is set to `0`, then `npeak` equals to Option `hpa_dirty_mult` works in combination with `hpa_peak_demand_window_ms`. -Fixed-point fractional or `-1`. Default value is `0.25` (not a great default). -When set to `-1` disables purging completely. +Fixed-point fractional or `-1`. The default value is `0.25` (not a great +default). When set to `-1` disables purging completely. ### hpa_sec_nshards |