[#86787] [Ruby trunk Feature#14723] [WIP] sleepy GC — ko1@...

Issue #14723 has been updated by ko1 (Koichi Sasada).

13 messages 2018/05/01
[#86790] Re: [Ruby trunk Feature#14723] [WIP] sleepy GC — Eric Wong <normalperson@...> 2018/05/01

[email protected] wrote:

[#87095] [Ruby trunk Feature#14767] [PATCH] gc.c: use monotonic counters for objspace_malloc_increase — ko1@...

Issue #14767 has been updated by ko1 (Koichi Sasada).

9 messages 2018/05/17
[#87096] Re: [Ruby trunk Feature#14767] [PATCH] gc.c: use monotonic counters for objspace_malloc_increase — Eric Wong <normalperson@...> 2018/05/17

[email protected] wrote:

[ruby-core:86844] Re: [Ruby trunk Feature#14718] Use jemalloc by default?

From: Eric Wong <normalperson@...>
Date: 2018-05-02 21:00:50 UTC
List: ruby-core #86844
[email protected] wrote:
> An alternative is to tune glibc by reducing the number of arenas. Call this on startup:
> 
>     #include "malloc.h"
>     mallopt(M_ARENA_MAX, 2)

Probably acceptable.  We need to verify it doesn't crash on
systems where Ruby is built against various glibc versions but
somebody tests with jemalloc via LD_PRELOAD.

We will also respect getenv("MALLOC_ARENA_MAX") if set; as I
prefer MALLOC_ARENA_MAX=1 for my low-priority stuff.

Btw, do you have more information on which version(s) of glibc
and compile options (or distro package) used?

I would like the Ruby community to work more closely with glibc
developers in the future.  Thanks.

Unsubscribe: <mailto:[email protected]?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>

In This Thread