[pull] master from tensorflow:master - #8798

Merged

pull[bot] merged 15 commits into

Cache-Cloud:masterfrom

tensorflow:master

Sep 1, 2026

Commits on Sep 1, 2026

Verify that multiple buffers at a position only occur when at least one buffer's position belongs to a computation that does not have a linear (single) caller chain up to the entry computation.

subhankarshah
authored and
tensorflower-gardener
committed
6bfdbe3
Implement byte-based memory budgeting and telemetry in ContinuousProfilerOrchestrator

sannidhyachauhan
authored and
tensorflower-gardener
committed
3a46aa2
Add converter pass debugging and crash reproduction infrastructure

majiddadashi
authored and
tensorflower-gardener
committed
aa3e471
Add XLA compilation start and end time streamz metrics to track device compiler performance and invocation boundaries.
tensorflower-gardener
committed
3c39650
PR #48045: [XLA:GPU] Propagate concatenate unroll factor to LLVM lowering

dlcompilers-infra-bot
authored and
tensorflower-gardener
committed
9bac9f2
PR #47846: Fix while loop range analysis for nonconstant steps

982945902
authored and
tensorflower-gardener
committed
5340b57
Remove deprecated MLIR pass wrappers in mlir_hlo

akuegel
authored and
tensorflower-gardener
committed
75a19c8
[XLA]: Fix -1 ^ x for negative exponents in elemental_ir_emitter

Ryo-not-rio
authored and
tensorflower-gardener
committed
21f5b88
PR #48107: Fix ambiguous conversion in `ExpandIntegerPower` on GCC

Flamefire
authored and
tensorflower-gardener
committed
3ff526a
Automated Code Change
tensorflower-gardener
committed
e93bbd7
Support excluding backends from selection in autotuner config_selector.

shyamliagrawal
authored and
tensorflower-gardener
committed
84a910c
Automated Code Change
tensorflower-gardener
committed
42c8474
Remove HoistFusedBitcasts from Triton GEMM fusion tests.

vwbaker
authored and
tensorflower-gardener
committed
87ba17e
[XLA:GPU] Reland: Use symmetric memory and peer address lookup for collective kernel thunks.

PatriosTheGreat
authored and
tensorflower-gardener
committed
f7b4581
Replace TF_ASSERT_OK macros with ASSERT_OK in triton_gemm_fusion_test.cc.

vwbaker
authored and
tensorflower-gardener
committed
52ec721

Navigation Menu

[pull] master from tensorflow:master - #8798

[pull] master from tensorflow:master#8798
pull[bot] merged 15 commits into
Cache-Cloud:masterfrom
tensorflow:master

Commits on Sep 1, 2026

Verify that multiple buffers at a position only occur when at least one buffer's position belongs to a computation that does not have a linear (single) caller chain up to the entry computation.

Implement byte-based memory budgeting and telemetry in ContinuousProfilerOrchestrator

Add converter pass debugging and crash reproduction infrastructure

Add XLA compilation start and end time streamz metrics to track device compiler performance and invocation boundaries.

PR #48045: [XLA:GPU] Propagate concatenate unroll factor to LLVM lowering

PR #47846: Fix while loop range analysis for nonconstant steps

Remove deprecated MLIR pass wrappers in mlir_hlo

[XLA]: Fix -1 ^ x for negative exponents in elemental_ir_emitter

PR #48107: Fix ambiguous conversion in `ExpandIntegerPower` on GCC

Automated Code Change

Support excluding backends from selection in autotuner config_selector.

Automated Code Change

Remove HoistFusedBitcasts from Triton GEMM fusion tests.

[XLA:GPU] Reland: Use symmetric memory and peer address lookup for collective kernel thunks.

Replace TF_ASSERT_OK macros with ASSERT_OK in triton_gemm_fusion_test.cc.