319411158 by A. Unique TensorFlower<gardener@tensorflow.org>:
Integrate LLVM at https://github.com/llvm/llvm-project/commit/d6343e607ac8
--
319410296 by A. Unique TensorFlower<gardener@tensorflow.org>:
[XLA] Implement extra prefetch limit for while uses.
Outstanding prefetch limits can prevent prefetches from being scheduled for
the duration of while loops. Since using alternate memory for the while loops
can be more beneficial, allow specifying additional prefetch limit when the use
is a while HLO.
--
319406145 by A. Unique TensorFlower<gardener@tensorflow.org>:
[XLA:CPU] Teach dot_op_emitter how to tile&vectorize linalg matmuls
And turn them on by default.
This is on-par with the existing emitter, sometimes better and unlocks more
potential. The strategy classes are duplicated right now, but I expect them to
graduate to mlir core soon.
I'm planning to remove the custom LLVM IR emitters if this turns out to be
stable enough.
--
319402982 by A. Unique TensorFlower<gardener@tensorflow.org>:
PR #40327: [ROCm] Enabling optimized FusedBatchNormInferenceMetaKernel for half
Imported from GitHub PR https://github.com/tensorflow/tensorflow/pull/40327
This PR enables optimized FusedBatchNormInferenceMetaKernel for half on ROCm.
Copybara import of the project:
--
5f658e2bc1b20794239658bffe0d7bf9cb89c81f by Eugene Kuznetsov <eugene.kuznetsov@amd.com>:
Enabling optimized FusedBatchNormInferenceMetaKernel for half
--
319393611 by A. Unique TensorFlower<gardener@tensorflow.org>:
Integrate LLVM at https://github.com/llvm/llvm-project/commit/68498ce8af37
--
319374663 by A. Unique TensorFlower<gardener@tensorflow.org>:
compat: Update forward compatibility horizon to 2020-07-02
--
319374662 by A. Unique TensorFlower<gardener@tensorflow.org>:
Update GraphDef version to 450.
--
319371388 by A. Unique TensorFlower<gardener@tensorflow.org>:
Update framework_build_test targets
--
319363982 by A. Unique TensorFlower<gardener@tensorflow.org>:
Resolve the permission denied error on Python 3.7 pip install.
--
319361498 by A. Unique TensorFlower<gardener@tensorflow.org>:
Add an option to only log parameters whose values are parsed from cmdline flags in the benchmark tool.
--
319356677 by A. Unique TensorFlower<gardener@tensorflow.org>:
Fix bug in ReadNonConstantTensor
assigning new value to the reference don't update the reference, so use pointer instead.
--
319350974 by A. Unique TensorFlower<gardener@tensorflow.org>:
Fix the header inclusion path issue with TensorFlowLiteC
--
319342653 by A. Unique TensorFlower<gardener@tensorflow.org>:
Fix the relationship between tpu_executor and tpu_executor_base build targets.
--
319342578 by A. Unique TensorFlower<gardener@tensorflow.org>:
Internal change
319340968 by A. Unique TensorFlower<gardener@tensorflow.org>:
Internal change
PiperOrigin-RevId: 319411158
either sh_test or the path generation of bazel have problems on windows.
PiperOrigin-RevId: 298390766
Change-Id: I2e7f53030d46b3a1a104a140735b1878cd456195
This change to reflect MLIR's recent addition to LLVM's monorepo. Instead of
stripping llvm from the http archive, use build file map to refer to both MLIR
and LLVM within llvm-project and remove read-only copy that was temporarily
there. Also bump the commit version until after the integration of MLIR into
LLVM monorepo.
PiperOrigin-RevId: 287547194
Change-Id: I4459b96702dc5eb19b58c7ceaa6cabb16f530a43
mlir-tblgen invocations for rules generating tests were not considered as test
runs of mlir-tblgen so create extra sh_test to verify their behavior.
PiperOrigin-RevId: 275722547
Change-Id: I30df93f1f9912bd4de36a183383bcabae98375ff
This change adds dialects and tools for TF, TFLite and XLA using MLIR (https://github.com/tensorflow/mlir). This is under active development and not built by default.
PiperOrigin-RevId: 255538027
