[Adreno] Change compute/schedule for ToMixedPrecision pass #12537
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
csullivan
reviewed
Aug 23, 2022
| @@ -437,7 +437,7 @@ def test_conv2d_vgg16_winograd_4d(): | |||
| stat_file = temp.relpath("stat.log") | |||
| with open(stat_file, "w") as f: | |||
| f.write( | |||
| '{"input": ["opencl -keys=adreno,opencl,gpu -device=adreno -max_num_threads=256", "conv2d_nchw_winograd_acc32.image2d", [["TENSOR", [1, 512, 28, 28], "float16"], ["TENSOR", [512, 512, 3, 3], "float16"], [1, 1], [1, 1, 1, 1], [1, 1], "float16"], {}], "config": {"index": 1591, "code_hash": null, "entity": [["auto_unroll_max_step", "ot", 4], ["tile_y", "sp", [-1, 1, 32]], ["tile_x", "sp", [-1, 4, 2]], ["tile_rc", "sp", [-1, 8]]]}, "result": [[0.0037244], 0, 7.06374192237854, 1653898629.7427933], "version": 0.2, "tvm_version": "0.8.dev0"}\n' | |||
| '{"input": ["opencl -keys=adreno,opencl,gpu -device=adreno -max_num_threads=256", "conv2d_nchw_winograd.image2d", [["TENSOR", [1, 512, 28, 28], "float16"], ["TENSOR", [512, 512, 3, 3], "float16"], [1, 1], [1, 1, 1, 1], [1, 1], "float16"], {}], "config": {"index": 1591, "code_hash": null, "entity": [["auto_unroll_max_step", "ot", 4], ["tile_y", "sp", [-1, 1, 32]], ["tile_x", "sp", [-1, 4, 2]], ["tile_rc", "sp", [-1, 8]]]}, "result": [[0.0037244], 0, 7.06374192237854, 1653898629.7427933], "version": 0.2, "tvm_version": "0.8.dev0"}\n' | |||
|
Sorry, I will take a look tomorrow |
csullivan
approved these changes
Aug 30, 2022
elvin-n
force-pushed
the
amalyshe/mixed_precision
branch
from
81075fd
to
02381d0
Compare
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Outdated suggestions cannot be applied.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.
The current support of mixed precision in adreno schedules was implemented as standalone schedules having "fp32" suffix. Such kernels can be selected during compilation due to two reasons
The tune flow in its turn was not able to distinguish and point only fp16 or fp16_acc32. Both schedules are tuned and during the compilation the schedule having best time is selected. I.e. by fact without artificial approach we are not able to tune and compile pure fp16 or fp16_acc32. Only manual selection of tune statistics causes currently to execute one of this mode.
In addition to this, the conversion function to fp16 was custom made in the user's script that isnot available to the public tvm user.
To address above issues we are proposing to use
ToMixedPrecision()pass. It supports mixed precision (fp16 compute with fp32 accumulation) as well.Current PR changes