Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138484103
Would be interesting to benchmark the off-heap mode for memory intensive
jobs.
I would tend to agree with your performance assessment. Since Flink always
did do
Github user StephanEwen commented on a diff in the pull request:
https://github.com/apache/flink/pull/1093#discussion_r38902942
--- Diff:
flink-core/src/main/java/org/apache/flink/core/memory/MemorySegmentFactory.java
---
@@ -0,0 +1,135 @@
+/*
+ * Licensed to the Apache
Github user mxm commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138480430
That's an impressive pull request, @StephanEwen! From a first glance, it
looks very well-thought-out. I would be glad to work on some of the remaining
issues like the
Github user mxm commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138506847
Let's merge this without changing the current state of affairs, i.e.
keeping on-heap memory management as the default. Then we should file a JIRA to
keep track of the
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138527113
Okay, let me merge this and see if I can address Henry's comment as part of
this...
---
If your project is set up for it, you can reply to this email and have your
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138676203
Thanks! Let us keep up the good work :-)
@hsaputra I addressed your comment in the commit. Actually made the field
volatile, we should be safe that way.
Github user asfgit closed the pull request at:
https://github.com/apache/flink/pull/1093
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is
Github user hsaputra commented on a diff in the pull request:
https://github.com/apache/flink/pull/1093#discussion_r38966751
--- Diff:
flink-core/src/main/java/org/apache/flink/core/memory/MemorySegmentFactory.java
---
@@ -0,0 +1,135 @@
+/*
+ * Licensed to the Apache
Github user hsaputra commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138676857
@StephanEwen ah yes I saw the update =)
Sorry I missed it in the commit check, thanks again for the hard work!
---
If your project is set up for it, you can reply
Github user hsaputra commented on a diff in the pull request:
https://github.com/apache/flink/pull/1093#discussion_r38966473
--- Diff:
flink-core/src/main/java/org/apache/flink/core/memory/MemorySegmentFactory.java
---
@@ -0,0 +1,135 @@
+/*
+ * Licensed to the Apache
Github user hsaputra commented on a diff in the pull request:
https://github.com/apache/flink/pull/1093#discussion_r38951685
--- Diff:
flink-core/src/main/java/org/apache/flink/core/memory/MemorySegmentFactory.java
---
@@ -0,0 +1,135 @@
+/*
+ * Licensed to the Apache
Github user hsaputra commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138636557
I agree to keep on-heap memory management as default. Will need to test on
YARN deployment to make sure the kinks are resolved when working with it.
---
If your
Github user uce commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138237365
Impressive! :-) Thanks for the write up. I think I've learned something
about the JVM by reading your PR description. ;)
With some minor refactorings in the network
Github user tillrohrmann commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138242989
Really impressive work @StephanEwen. Will be hard to review due to its
sheer extent.
---
If your project is set up for it, you can reply to this email and have
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138269258
I think it looks larger than it is. Many small changes dues to changes
signatures of the MemoryManager and the MemorySegment instantiation.
Also, the vast
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-138270482
Since this does not break or alter any existing behavior, I would like to
merge it rather soon (as soon as concerns have been cleared).
We an polish the
Github user hsaputra commented on a diff in the pull request:
https://github.com/apache/flink/pull/1093#discussion_r38890190
--- Diff:
flink-core/src/main/java/org/apache/flink/core/memory/MemorySegmentFactory.java
---
@@ -0,0 +1,135 @@
+/*
+ * Licensed to the Apache
GitHub user StephanEwen opened a pull request:
https://github.com/apache/flink/pull/1093
[FLINK-1320] [core] Add an off-heap variant of the managed memory
This pull request extends the Flink managed memory to work transparently
on-heap and off-heap.
In Flink's core were
Github user hsaputra commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-137857146
Isn't there another PR sent for this issue by Max? Or was it for something
else.
---
If your project is set up for it, you can reply to this email and have your
reply
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1093#issuecomment-137872521
I took the code from #290 (by @mxm and me) as the basis, but heavily
reworked it. Some code is similar most notably, the collapsed
address/release/bound check in
20 matches
Mail list logo