[ARITH] Simplify casts of constants 0 and 1 #3758
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
tqchen
requested changes
Aug 12, 2019
| } | ||
| } | ||
|
|
||
| inline bool is_const_int(const Expr& x, int64_t value) { |
There was a problem hiding this comment.
keep this impl, as the is_const_value involves recursions and could be more complicated than this one.
| } | ||
|
|
||
| Expr RewriteSimplifier::Impl:: | ||
| Mutate_(const Cast* op, const Expr& self) { |
There was a problem hiding this comment.
Let us just add cast between (uint/float/int here), ignore vector types for now as they are not as common.
include/tvm/expr_operator.h
Outdated
| } | ||
|
|
||
| template <typename ValueType> | ||
| inline bool is_const_value(const Expr& e, ValueType value) { |
There was a problem hiding this comment.
Let us not introduce this for now(see comments in the Mutator).
xqdan
suggested changes
Aug 13, 2019
src/arithmetic/rewrite_simplify.cc
Outdated
| } | ||
| if (is_const_value(op->value, 1)) { | ||
| return make_const(op->type, 1); | ||
| } |
There was a problem hiding this comment.
could you support float(16) --> 16.000, int32(16.000) --> 16 ?
There was a problem hiding this comment.
With the implementation via tvm::cast cases like these should work now.
This reverts commit 7e1b346.
|
Turned out there is the |
xqdan
approved these changes
Aug 13, 2019
tqchen
approved these changes
Aug 13, 2019
wweic
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Aug 16, 2019
* [ARITH] Simplify casts of constants 0 and 1 * [EXPR] is_const_value to check whether non-ints are consts * Revert "[EXPR] is_const_value to check whether non-ints are consts" This reverts commit 7e1b346. * Use tvm::cast
anijain2305
pushed a commit
to anijain2305/tvm
that referenced
this pull request
Aug 22, 2019
* [ARITH] Simplify casts of constants 0 and 1 * [EXPR] is_const_value to check whether non-ints are consts * Revert "[EXPR] is_const_value to check whether non-ints are consts" This reverts commit 7e1b346. * Use tvm::cast
wweic
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Sep 6, 2019
* [ARITH] Simplify casts of constants 0 and 1 * [EXPR] is_const_value to check whether non-ints are consts * Revert "[EXPR] is_const_value to check whether non-ints are consts" This reverts commit 7e1b346. * Use tvm::cast
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.
This PR implements simplification of some casts of constants in the rewrite simplifier. The use case I have in mind is conversion from boolean expressions to floats (
float32(i < j)instead ofselect(i < k, 1.0, 0.0)). I restricted consts to be just 0 and 1 because I'm not sure about usefulness and safety of other values (0 and 1 should be safe for all types, except maybe for some custom types).This PR also introduces the
is_const_valuefunctions which is likeis_const_int, but works for float expressions too. Besides this PR, it is also used in the tensor expression autodiff implementation.