[Ansor][AutoTVM v2.0] Phase 2: Update heavy operations with parallel_for #6348
Conversation
This file contains 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
jcf94
changed the title
comaniac
approved these changes
Aug 27, 2020
jcf94
force-pushed
the
parallel_for_update
branch
from
412ae14
to
9c8dca0
Compare
|
@merrymercy @FrozenGene Plz have a look when you're free. :) |
merrymercy
requested changes
Aug 30, 2020
src/auto_scheduler/compute_dag.cc
Outdated
| out_states.push_back(std::move(out_state)); | ||
| } | ||
| if (out_state.defined()) { | ||
| std::unique_lock<std::mutex> l(m); |
There was a problem hiding this comment.
Can we use low-level APIs from ArrayNode to remove this mutex lock?
Because we know the number of states, we can allocate n empty State() in advance.
Do something like
std::vector<State> out_states(n, State());
parallel_for i in 0...n:
out_state[i] = process(states[i])There was a problem hiding this comment.
Our original implementation in Ansor repo just worked in this way, while if there's any inferbound failure the out_state will contains some empy State().
There was a problem hiding this comment.
Added a note in the doc string of ComputeDAG::Inferbound, added necessary PruneInvalidState to the search policy.
merrymercy
approved these changes
Aug 31, 2020
kevinthesun
pushed a commit
to kevinthesun/tvm
that referenced
this pull request
Sep 17, 2020
…for (apache#6348) * Update auto_scheduler with parallel_for * Update * Update * Update * Update inferbound
kevinthesun
pushed a commit
to kevinthesun/tvm
that referenced
this pull request
Sep 18, 2020
…for (apache#6348) * Update auto_scheduler with parallel_for * Update * Update * Update * Update inferbound
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Sep 18, 2020
…for (apache#6348) * Update auto_scheduler with parallel_for * Update * Update * Update * Update inferbound
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
For the full upstream plan, see Ansor RFC.
This PR contains some small fix:
Currently add for:
SketchPolicyNode::SampleInitPopulation(some rules used the SketchPolicyNode->rand_gen, so this part still has to be protected by a std::mutex)Do we need to add parallel_for for EvolutionarySearch after [Ansor][AutoTVM v2.0] Phase 2: Evolutionary Search #6310 has been merged?cc @merrymercy @comaniac