[Codegen][cuda-fp16] fallback to fp32 simulation when cuda arch < sm53 #4268
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
|
I am not sure if we really want to enable fp16 on devices that does not have native support. never-the-less. I think we can enable it. However, it would be great to separate this dep lib into a separate const string var that uses string literal to include the files. So we won't have a long string in the content. |
|
I agree better we can put the code in a separate file instead of string concatenation. The code itself is good |
tqchen
requested changes
Nov 8, 2019
src/codegen/codegen_cuda.cc
Outdated
| @@ -50,6 +50,9 @@ void CodeGenCUDA::AddFunction(LoweredFunc f) { | |||
|
|
|||
| std::string CodeGenCUDA::Finish() { | |||
| if (enable_fp16_) { | |||
| static constexpr const char* _cuda_half_t_def = | |||
| #include "literal/cuda_half_t.txt" | |||
There was a problem hiding this comment.
consider still keep it as a header file so we can apply the license header, define a global constant
vinx13
approved these changes
Nov 10, 2019
tqchen
approved these changes
Nov 10, 2019
tqchen
added
status: accepted
and removed
status: need review
status: need update
zxy844288792
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Nov 13, 2019
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Outdated suggestions cannot be applied.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.
This ensures fp16 can still be computed on old devices.
The code is borrowed from mshadow.
An immediate usage will be tvm op support in mxnet - mxnet is able to run fp16 on device which does not support fp16 natively.
@vinx13 @tqchen @Laurawly @reminisce Please advise.