[JIT] Support multiple outputs in subgraph matcher.#48992
Closed
ZolotukhinM wants to merge 7 commits intogh/ZolotukhinM/388/basefrom
Closed
[JIT] Support multiple outputs in subgraph matcher.#48992ZolotukhinM wants to merge 7 commits intogh/ZolotukhinM/388/basefrom
ZolotukhinM wants to merge 7 commits intogh/ZolotukhinM/388/basefrom
Conversation
[ghstack-poisoned]
facebook-github-bot
added
cla signed
oncall: jit
💊 CI failures summary and remediationsAs of commit be3a47c (more details on the Dr. CI page):
This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions to the (internal) Dr. CI Users group. This comment has been revised 19 times. |
Differential Revision: [D25388100](https://our.internmc.facebook.com/intern/diff/D25388100) [ghstack-poisoned]
Codecov Report
@@ Coverage Diff @@
## gh/ZolotukhinM/388/base #48992 +/- ##
===========================================================
+ Coverage 80.56% 80.59% +0.03%
===========================================================
Files 1874 1873 -1
Lines 202776 202755 -21
===========================================================
+ Hits 163362 163406 +44
+ Misses 39414 39349 -65 |
hlu1
pushed a commit
to hlu1/pytorch
that referenced
this pull request
Dec 9, 2020
Summary: Pull Request resolved: pytorch#48992 Differential Revision: D25388100 Test Plan: Imported from OSS Pulled By: ZolotukhinM fbshipit-source-id: 94a178680bcaf4ab7eece90dcb1f31a2ca9aad30
Differential Revision: [D25388100](https://our.internmc.facebook.com/intern/diff/D25388100) [ghstack-poisoned]
ZolotukhinM
changed the title
Differential Revision: [D25388100](https://our.internmc.facebook.com/intern/diff/D25388100) [ghstack-poisoned]
bwasti
approved these changes
Dec 14, 2020
| %z = my::op2(%x) | ||
| return (%y, %z))IR", | ||
| &pattern2); | ||
| // Not supported multi-output pattern |
Author
There was a problem hiding this comment.
Because not the whole pattern is covered by a traversal up from the first output (%z = ... is not visited). I will add a comment.
| } | ||
|
|
||
| bool SubgraphMatcher::isPatternOutput(const Value* v) const { | ||
| if (v->owningGraph() != &pattern_) { |
Contributor
There was a problem hiding this comment.
I don't think it will hurt, but is this check necessary?
| * but the latter is taken as non-const so that Match may contain non-const | ||
| * pointers. This enables clients of this API to use Match to drive mutations. | ||
| * | ||
| * Note [Multi-output Patterns] |
Contributor
There was a problem hiding this comment.
the non-backtracking algorithm (O(M*N) I think: BFS at each node in the graph) or endowing an output ordering (via some kind of topo-sort) might be a good bootcamp task :)
Author
There was a problem hiding this comment.
Mm, I don't know if such an algorithm exists, maybe I'm missing something? Will be happy to discuss it :)
| * output and assume that it is sufficient to traverse up from it to match the | ||
| * entire pattern. | ||
| * | ||
| * Corrolary 1: the order of outputs in the pattern matters! |
Contributor
There was a problem hiding this comment.
does the order in which internal outputs are consumed matter?
graph(a, b):
c = node0(a)
d = node1(b)
return c + d, d
vs
graph(a, b):
d = node1(b)
c = node0(a)
return c + d, d
Differential Revision: [D25388100](https://our.internmc.facebook.com/intern/diff/D25388100) [ghstack-poisoned]
Differential Revision: [D25388100](https://our.internmc.facebook.com/intern/diff/D25388100) [ghstack-poisoned]
Differential Revision: [D25388100](https://our.internmc.facebook.com/intern/diff/D25388100) [ghstack-poisoned]
Contributor
|
@ZolotukhinM merged this pull request in 38a59a6. |
hwangdeyu
pushed a commit
to hwangdeyu/pytorch
that referenced
this pull request
Jan 6, 2021
Summary: Pull Request resolved: pytorch#48992 Differential Revision: D25388100 Test Plan: Imported from OSS Reviewed By: heitorschueroff Pulled By: ZolotukhinM fbshipit-source-id: d95713af2220cf4f99ac92f59f8e5b902f2f3822
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stack from ghstack:
Differential Revision: D25388100