Float16 design doc #5313
Merged
Float16 design doc #5313
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
hedaoyuan
reviewed
Nov 6, 2017
doc/design/float16.md
Outdated
| float16 float_to_half_rn(float f); // convert to half precision in round-to-nearest-even mode | ||
| float half_to_float(float16 h); | ||
| ``` | ||
| which provides one-to-one conversion between float32 and float16. These twos functions will do different conversion routines based on the current hardware. CUDA/ARM instrinsics will be used when the corresonding hardware is available. When the hardware falls back to non-ARM cpu, software emulation will be performed to do the conversion. |
Contributor
There was a problem hiding this comment.
When the hardware falls back to non-ARM cpu -> If the hardware or compiler level does not support float32 to float16 conversion.
doc/design/float16.md
Outdated
|
|
||
| A brief survey of float16 support on different hardwares can be found [here](https://github.com/PaddlePaddle/Paddle/issues/4853). A brief survey of existing float16 implementations can be found [here](https://github.com/Xreki/Xreki.github.io/blob/master/multi_data_types_in_dl_framework/ppt/float16_and_quantized_type.md). | ||
|
|
||
| There are various natively supported float16 implementations on different hardwares/linear algebra libraries including half on cuda, float16_t on ARM processor, and Eigen::half on Eigen. |
Contributor
There was a problem hiding this comment.
I think a more detailed description is needed here. Need to describe the support of float16 from the three levels of hardware, compiler, and library respectively.
- For nvcc compiler, after CUDA 7.5, the __half type is supported.
- For NVIDIA GPU, maybe after sm_6.0(or sm_5.3?)
- For gcc/clang? In the mobile, usually, clang compiler.
- For libraries. Currently, what libraries support float16 calculations?
This is important because of if we upgrade the environment, we need to know what the minimum environment to support is.
Contributor
Author
There was a problem hiding this comment.
Great point! A detailed description has been added. Thanks!
hedaoyuan
approved these changes
Nov 7, 2017
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
No description provided.