summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorTakashi Kokubun <takashikkbn@gmail.com>2024-01-24 10:26:57 -0800
committerGitHub <noreply@github.com>2024-01-24 10:26:57 -0800
commit303fef875240b19fc582ab80897f5e15df784eca (patch)
tree3ee5b656a763d0592a193d5564f5c10952bc1d47
parent1702528258c2b2a87dcf5673a80d999de35d0b71 (diff)
YJIT: Update yjit.md about mem size (#9687)
-rw-r--r--doc/yjit/yjit.md4
1 files changed, 2 insertions, 2 deletions
diff --git a/doc/yjit/yjit.md b/doc/yjit/yjit.md
index 02f8dd7948..e6446e3ed1 100644
--- a/doc/yjit/yjit.md
+++ b/doc/yjit/yjit.md
@@ -243,8 +243,8 @@ which often consumes more memory than JIT code. Generally, YJIT adds memory over
3-4x of `--yjit-exec-mem-size` in production as of Ruby 3.3. You should multiply that by the number
of worker processes to estimate the worst case memory overhead.
-We use `--yjit-exec-mem-size=64` for Shopify's Rails monolith, which is Ruby 3.3's default,
-but smaller values like 32 MiB or 48 MiB might make sense for your application.
+`--yjit-exec-mem-size=48` is the default since Ruby 3.3.1,
+but smaller values like 32 MiB might make sense for your application.
While doing so, you may want to monitor `RubyVM::YJIT.runtime_stats[:ratio_in_yjit]` as explained above.
### Enabling YJIT lazily