Statement instance order and lex order map #479
Draft
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
…ts for all relevant insn_ids simultaneously; then afterward process pairs individually; (also eliminate more unnecessary lex dims)
… that are not in {0,1} with {0,1} (we're only describing relative order, so higher int values are unnecessary and confusing)
…es in schedule outline creation since these are replaced with {0, 1} in the map simplification step. Further reduce number of lex dims in pairwise maps with the following strategy: once a lex tuple dimension is found where both tuples have non-matching integer values, remove any faster-updating lex dimensions where both tuples have integer values, even if the integers don't match.
…id_before, insn_id_after) tuples to (schedule_before, schedule_after) tuples
…ent_pairs(); don't hardcode expected lex order maps, instead create them to match expected dim size
…nsion is found where both tuples have non-matching integer values, remove all faster-updating lex dimensions.
…ts, keep all pairs in lists (of length 2) and operate on the lists with loops
…rder-and-lex-order-map
…est; probably the wrong way to do this... (but it's not time to add this to tutorial yet)
…rder-and-lex-order-map
…ements map to the exact same point(s) in the lex ordering, which is okay, but to represent this, our lex tuple cannot be empty, so map to (0))
…rder-and-lex-order-map
…e_symbolic_map_from_tuples(), all dims will always be present by construction; rename insert_missing_dims_and_reorder_by_name()->reorder_dims_by_name()
…rder-and-lex-order-map
…ged with the same parallel iname tag
…ple inames tagged with the same parallel iname tag
…ple dims to rename in a dict
…pairwise_statement_orderings_inner
…concurrent loops; enable removal of redundant pass over instructions by computing max_depth_of_barrier_loop in first pass
…nested_loops in _gather_blex_ordering_info()
…_info; rename loops_with_barriers->seq_loops_with_barriers for clarity
…ion _pad_tuples_and_assign_integer_vals_to_map_template
…ow computed by starting with TOP/BOTTOM and then applying the appropriate loop bound
…make our mechanisms that account for the effect of *local* barriers on execution order view *global* barriers as also having that effect
…so syncronize threads *within* a work-group, make our mechanisms that account for the effect of *local* barriers on execution order view *global* barriers as also having that effect)
4 tasks
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.
cc @jdsteve2
Subsumes #202 and #169.
See prior review progress