Skye Wanderman-Milne has posted comments on this change.

Change subject: IMPALA-2784: codegen 
PartitionedHashJoinNode::Partition::BuildHashTable()
......................................................................


Patch Set 1:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/2113/1/be/src/exec/partitioned-hash-join-node-ir.cc
File be/src/exec/partitioned-hash-join-node-ir.cc:

Line 289: AddProbeFilters
> What does the variance look like? Can you nail this down to a particular co
I was rebased on a pretty stale trunk (~1 week I think). Here are numbers from 
this morning:

With template:
  PHJ time: 17.4
  Build time: 5.3
  Codegen time: ~480ms

Trunk:
  PHJ time: 20.1s
  Build time: 8.6s
  Codegen time: ~350ms

So these are in line with my original numbers. I think my queries may have been 
running slow before because I was running low on memory, or maybe I pulled in a 
necessary patch.

The codegen time looks pretty variable... for trunk I saw codegen times for the 
fragment between 292ms and 609ms (!!), and for my patch 419ms and 497ms. 
However, trunk seems to usually be around 350ms, and my patch 480ms. So this 
will still be a regression from 2.5.


-- 
To view, visit http://gerrit.cloudera.org:8080/2113
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I616f46a861b4909d7a6e66dcf947b3518556768e
Gerrit-PatchSet: 1
Gerrit-Project: Impala
Gerrit-Branch: cdh5-trunk
Gerrit-Owner: Skye Wanderman-Milne <[email protected]>
Gerrit-Reviewer: Jim Apple <[email protected]>
Gerrit-Reviewer: Skye Wanderman-Milne <[email protected]>
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-HasComments: Yes

Reply via email to