New feature: Allows the output data of a layer to remain in the low-level memory for later use when possible #25
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
…enough for current layer processing
asyms
requested changes
Oct 9, 2023
| ## do not add the current mem into the modified architecture | ||
| ##################################################### | ||
|
|
||
| class RemoveNoUseMemStage(Stage): |
There was a problem hiding this comment.
This name is not the most informative. Maybe something like "RemoveUnusedMemoryStage"?
| for cme, extra_info in sub_stage.run(): | ||
| yield cme, extra_info | ||
|
|
||
| def generate_accelerator_removing_nouse_mem(self): |
There was a problem hiding this comment.
Similarly rename this function.
| memory_hierarchy = core.memory_hierarchy | ||
|
|
||
| if len(self.layer.constant_operands) == 1: | ||
| act_operand = self.layer.memory_operand_links[[operand for operand in self.layer.input_operands if operand not in self.layer.constant_operands][0]] # act representation |
There was a problem hiding this comment.
Rename all operand variables to reflect if they're "layer operands" or "memory operands" either in the name or in comments
| ## mem_update_weight = current_mem_level | ||
| ##################################################### | ||
|
|
||
| class SearchNoUseMemStage(Stage): |
There was a problem hiding this comment.
Rename this class and documentation to "SearchUnusedMemoryStage" or something similar.
| const_operand = self.layer.memory_operand_links[weight_operand] # weight representation | ||
|
|
||
| # Find target_act/const/output_mem_level | ||
| for pos, ele in enumerate(self.mem_update_list[f"{curr_id}"]): |
There was a problem hiding this comment.
is curr_id not already a string? If not, you should convert a variable to a string using str(...)
| # Find target_act/const/output_mem_level | ||
| for pos, ele in enumerate(self.mem_update_list[f"{curr_id}"]): | ||
| if list(ele.keys())[0] == f"{act_operand}": | ||
| target_act_mem_level = self.mem_update_list[f"{curr_id}"][pos][f"{act_operand}"] |
There was a problem hiding this comment.
Pretty sure act_operand is already a string?
| dataflows = core.dataflows | ||
| if dataflows is not None: | ||
| raise NotImplementedError( | ||
| "Scale your core-defined dataflows accordingly here." |
There was a problem hiding this comment.
Why do the dataflows have to be scaled?
| Update mem_update_list and mem_update_weight according to the algorithm description at the file beginning. | ||
| """ | ||
| """ | ||
| param const_operand: constant operand name (e.g. "W") |
There was a problem hiding this comment.
You specify three parameters but the function has no parameters?
… not passed the test
…elated pytest under tests/main
…_memory -> test_origin
|
New fixes:
Extra note:
|
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.
New feature:
Note:
workload_data_always_from_top_memparameter toTruein therun()function within theSearchNoUseMemStage.pyfile.New stages:
How to use:
Example:
An example function,
get_hardware_performance_zigzag_unused_mem_removingis provided inapi.pyfor reference.