aboutsummaryrefslogtreecommitdiffstats
path: root/erts/emulator/internal_doc/CarrierMigration.md
diff options
context:
space:
mode:
authorJohn Högberg <[email protected]>2018-07-03 09:33:02 +0200
committerGitHub <[email protected]>2018-07-03 09:33:02 +0200
commit6cfa8af60a9c6fe9043e4992434d40d4a34019e3 (patch)
tree822993e0a4fb987014238369db045bf53eaafba3 /erts/emulator/internal_doc/CarrierMigration.md
parentbc22321f10c1ad71fb7e28275a8e6ed26a34d4ab (diff)
parent29ce7ab0e4403fa688d1e5e5e2170e88d7e8f299 (diff)
downloadotp-6cfa8af60a9c6fe9043e4992434d40d4a34019e3.tar.gz
otp-6cfa8af60a9c6fe9043e4992434d40d4a34019e3.tar.bz2
otp-6cfa8af60a9c6fe9043e4992434d40d4a34019e3.zip
Merge pull request #1854 from jhogberg/john/erts/cross-type-carrier-migration/OTP-15063
Allow carrier migration between different allocator types
Diffstat (limited to 'erts/emulator/internal_doc/CarrierMigration.md')
-rw-r--r--erts/emulator/internal_doc/CarrierMigration.md15
1 files changed, 1 insertions, 14 deletions
diff --git a/erts/emulator/internal_doc/CarrierMigration.md b/erts/emulator/internal_doc/CarrierMigration.md
index 3a796d11b7..6c79bd731c 100644
--- a/erts/emulator/internal_doc/CarrierMigration.md
+++ b/erts/emulator/internal_doc/CarrierMigration.md
@@ -34,8 +34,7 @@ Solution
--------
In order to prevent scenarios like this we've implemented support for
-migration of multi-block carriers between allocator instances of the
-same type.
+migration of multi-block carriers between allocator instances.
### Management of Free Blocks ###
@@ -130,10 +129,6 @@ threads may have references to it via the pool.
### Migration ###
-There exists one pool for each allocator type enabling migration of
-carriers between scheduler specific allocator instances of the same
-allocator type.
-
Each allocator instance keeps track of the current utilization of its
multi-block carriers. When the total utilization falls below the "abandon
carrier utilization limit" it starts to inspect the utilization of the
@@ -287,11 +282,3 @@ reduced using the `aoffcbf` strategy. A trade off between memory
consumption and performance is however inevitable, and it is up to
the user to decide what is most important.
-Further work
-------------
-
-It would be quite easy to extend this to allow migration of multi-block
-carriers between all allocator types. More or less the only obstacle
-is maintenance of the statistics information.
-
-