[QNN] Concat - Refactoring to C++ #3819
Merged
Conversation
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
anijain2305
force-pushed
the
qnn_concat
branch
from
41c82e4 to
6b35322
Compare
anijain2305
force-pushed
the
qnn_concat
branch
6 times, most recently
from
ea98355 to
5679038
Compare
|
@vinx13 @zhiics @yzhliu Can you please review this? Refactoring to C++ removes InferType dependency. Relevant comment - #3730 (comment) |
vinx13
requested changes
Aug 30, 2019
src/relay/pass/pattern_util.h
Outdated
| return CallNode::make(op, {fill_value}, Attrs(attrs), {}); | ||
| } | ||
|
|
||
| static inline Expr Concatenate(Expr data, int axis) { |
There was a problem hiding this comment.
this is actually the same as MakeConcatenate below isn't it?
There was a problem hiding this comment.
Yup. Nice observation. Made the changes to use that.
anijain2305
force-pushed
the
qnn_concat
branch
from
5679038 to
b1e43d2
Compare
zhiics
requested changes
Aug 30, 2019
src/relay/qnn/op/concatenate.cc
Outdated
| Array<tvm::Expr> input_zero_points, double output_scale, | ||
| int32_t output_zero_point, int axis) { | ||
| auto attrs = make_node<QnnConcatenateAttrs>(); | ||
| attrs->input_scales = input_scales; |
There was a problem hiding this comment.
Suggested change
| attrs->input_scales = input_scales; | |
| attrs->input_scales = std::move(input_scales); |
src/relay/qnn/op/concatenate.cc
Outdated
| int32_t output_zero_point, int axis) { | ||
| auto attrs = make_node<QnnConcatenateAttrs>(); | ||
| attrs->input_scales = input_scales; | ||
| attrs->input_zero_points = input_zero_points; |
There was a problem hiding this comment.
Suggested change
| attrs->input_zero_points = input_zero_points; | |
| attrs->input_zero_points = std::move(input_zero_points); |
src/relay/qnn/op/concatenate.cc
Outdated
|
|
||
| /* | ||
| * \brief Canonicalizes the QNN concatenate op. | ||
| * \param ref_call The original call that will be lowered. |
There was a problem hiding this comment.
Names in comment mismatch with the arguments.
src/relay/qnn/util.h
Outdated
| const Array<IndexExpr>& input_shape, const DataType& out_dtype); | ||
|
|
||
| static inline Expr Requantize(const Expr& data, const Array<IndexExpr>& input_shape, | ||
| const double& input_scale, const int32_t& input_zero_point, |
There was a problem hiding this comment.
no need to have const xx& for built-in types, just double input_scale, int32_t input_zero_point etc.
anijain2305
force-pushed
the
qnn_concat
branch
from
b1e43d2 to
cc55ce8
Compare
vinx13
approved these changes
Aug 31, 2019
Merged
wweic
pushed a commit
to wweic/tvm
that referenced
this pull request
Sep 16, 2019
wweic
pushed a commit
to wweic/tvm
that referenced
this pull request
Sep 16, 2019
wweic
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Sep 16, 2019
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.
There are 2 reasons for moving to C++
Please review @yzhliu @jackwish @FrozenGene @vinx13 @tqchen @zhiics