Skip to content

Support dynamic expression clustering and broadcast singletons #28

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
stevenvar wants to merge 3 commits into
base: for-serving-2.20
Choose a base branch
from
Open

Support dynamic expression clustering and broadcast singletons #28

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b388cfc
Port dynamic input expression clustering stack
stevenvar Jul 14, 2026
fa14e79
Add flag to disable dynamic size padding
stevenvar May 20, 2026
df4822c
Handle broadcast-singleton dynamic solve candidates
stevenvar Jul 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions tensorflow/compiler/jit/flags.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,6 +167,15 @@ void AppendMarkForCompilationPassFlagsInternal(std::vector<Flag>* flag_list) {
Flag("tf_xla_enable_dynamic_sizes",
&mark_for_compilation_flags->tf_xla_enable_dynamic_sizes,
"Enable dynamic sizes support."),
Flag("tf_xla_disable_dynamic_size_padding",
&mark_for_compilation_flags->tf_xla_disable_dynamic_size_padding,
"Disable padding dynamic runtime sizes up to an XLA compile batch "
"when dynamic sizes support is enabled."),
Flag("tf_xla_enable_dynamic_solve_majority_vote",
&mark_for_compilation_flags
->tf_xla_enable_dynamic_solve_majority_vote,
"Enable solve-time majority vote when dynamic expression solving "
"observes conflicting candidates."),
Flag("tf_xla_enable_symbolic_content",
&mark_for_compilation_flags->tf_xla_enable_symbolic_content,
"Enable symbolic content propagation."),
Expand Down Expand Up @@ -265,6 +274,9 @@ void AllocateAndParseFlags() {
->tf_xla_disable_resource_variable_safety_checks_for_debugging = false;
mark_for_compilation_flags->tf_xla_deterministic_cluster_names = false;
mark_for_compilation_flags->tf_xla_enable_dynamic_sizes = false;
mark_for_compilation_flags->tf_xla_disable_dynamic_size_padding = false;
mark_for_compilation_flags->tf_xla_enable_dynamic_solve_majority_vote =
false;
mark_for_compilation_flags->tf_xla_enable_symbolic_content = false;
mark_for_compilation_flags->tf_xla_persistent_cache_directory = "";
mark_for_compilation_flags->tf_xla_persistent_cache_device_types = "";
Expand Down
7 changes: 7 additions & 0 deletions tensorflow/compiler/jit/flags.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,6 +102,13 @@ struct MarkForCompilationPassFlags {
// If true enables support of dynamic sizes.
bool tf_xla_enable_dynamic_sizes;

// If true disables padding runtime dynamic sizes up to an XLA compile batch.
bool tf_xla_disable_dynamic_size_padding;

// If true enables solve-time majority vote when dynamic expression solving
// observes conflicting candidates.
bool tf_xla_enable_dynamic_solve_majority_vote;

// If true enables symbolic content propagation.
bool tf_xla_enable_symbolic_content;

Expand Down
Loading