experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
STT-tensorflow/tensorflow/compiler/xla
History
Stephan Herhut 1a44a1f9dd Drop the implementation of the mlir_gpu plugin prototype in XLA. 
This leaves some passes and the kernel_lowering in place for now. Those are pending removal,, too.

PiperOrigin-RevId: 348621555
Change-Id: I4ff2ed11fb646bf76ceaa780956c6257d89f54ae
2020-12-22 06:05:01 -08:00
..
client
[XLA] Avoid reshape to R1 in NormalFloatingPointDistribution
2020-12-15 11:13:11 -08:00
experimental/xla_sharding
[XLA:SPMD] Minor fixes and utils for manual sharding
2020-12-01 13:37:36 -08:00
g3doc
[TF2XLA] Support asserts as no-ops for tf.function(jit_compile=True)
2020-12-17 18:53:10 -08:00
pjrt
Mark Execute methods non-const allow more flexibility, e.g., lazily load programs on first execution.
2020-12-16 12:13:08 -08:00
python
Have the C++ path raise a cleaner error.
2020-12-16 07:39:51 -08:00
python_api
…
rpc
…
service
Drop the implementation of the mlir_gpu plugin prototype in XLA.
2020-12-22 06:05:01 -08:00
tests
Optimize calls to std::string::find() and friends for a single char.
2020-12-21 08:54:54 -08:00
tools
Drop the implementation of the mlir_gpu plugin prototype in XLA.
2020-12-22 06:05:01 -08:00
.clang-format
…
array2d_test.cc
…
array2d.h
…
array3d_test.cc
…
array3d.h
…
array4d_test.cc
…
array4d.h
…
array_test.cc
…
array.cc
…
array.h
…
bit_cast_test.cc
…
bit_cast.h
…
BUILD
Move llvm dependency from xla/util.h to xla/service/gpu/ir_emission_utils.h
2020-12-14 16:51:06 -08:00
comparison_util.cc
…
comparison_util.h
…
cpu_function_runtime.cc
…
cpu_function_runtime.h
…
debug_options_flags.cc
[XLA:GPU] Add --xla_gpu_deterministic_ops to force run-to-run determinism.
2020-12-15 12:51:36 -08:00
debug_options_flags.h
…
debug_options_parsers_test.cc
…
debug_options_parsers.h
…
device_util.h
…
error_spec.h
…
executable_run_options.cc
[NFC] Moved GlobalDeviceId into its own file as it is also used by the CPU runtime.
2020-11-24 04:21:38 -08:00
executable_run_options.h
[NFC] Moved GlobalDeviceId into its own file as it is also used by the CPU runtime.
2020-11-24 04:21:38 -08:00
execution_options_util.cc
…
execution_options_util.h
…
index_util_test.cc
…
index_util.cc
…
index_util.h
…
iterator_util_test.cc
…
iterator_util.h
…
layout_test.cc
…
layout_util_test.cc
…
layout_util.cc
…
layout_util.h
…
layout.cc
…
layout.h
…
literal_comparison.cc
fix typos in compiler dir
2020-10-29 01:08:30 +03:00
literal_comparison.h
…
literal_test.cc
…
literal_util.cc
…
literal_util.h
…
literal.cc
Add several fixes for s390x architecture and other Big-Endian systems
2020-10-28 11:14:47 -04:00
literal.h
…
map_util.h
…
metric_table_report.cc
…
metric_table_report.h
…
overflow_util.h
…
packed_literal_reader.cc
…
packed_literal_reader.h
…
parse_flags_from_env_test.cc
…
parse_flags_from_env.cc
…
parse_flags_from_env.h
…
primitive_util_test.cc
…
primitive_util.cc
…
primitive_util.h
…
protobuf_util.cc
…
protobuf_util.h
…
README.md
…
refcounting_hash_map_test.cc
Added RefCountingHashMap::GetOrTryCreateIfAbsent to allow for factory methods that may fail.
2020-11-26 06:10:04 -08:00
refcounting_hash_map.h
[XLA] Fix race condition in RefcountingHashMap that could lead to a use-after-free segfault.
2020-12-15 12:11:48 -08:00
reference_util_test.cc
…
reference_util.cc
[XLA] Add an optional preferred_element_type argument to Dot/Conv builder methods to enable generating HLOs that has wider accumulation type than default shape inference result.
2020-11-20 13:16:56 -08:00
reference_util.h
…
service_interface.h
…
shape_layout.cc
…
shape_layout.h
…
shape_test.cc
…
shape_tree_test.cc
Internal tests cleanup.
2020-10-28 07:40:36 -07:00
shape_tree.h
…
shape_util_test.cc
Speed up Shape creation by avoiding unnecessary validations.
2020-11-22 20:40:07 -08:00
shape_util.cc
[XLA:GPU] Migrate convolution thunk emission to MLIR
2020-12-14 20:55:56 -08:00
shape_util.h
[XLA:GPU] Migrate convolution thunk emission to MLIR
2020-12-14 20:55:56 -08:00
shape.cc
…
shape.h
[libtpu] Reimplement XLA_Shape in C API for performance.
2020-10-19 14:17:22 -07:00
status_macros_test.cc
…
status_macros.cc
…
status_macros.h
…
status.h
…
statusor.h
…
test_helpers.h
…
test.h
…
text_literal_reader_test.cc
…
text_literal_reader.cc
…
text_literal_reader.h
…
text_literal_writer_test.cc
…
text_literal_writer.cc
…
text_literal_writer.h
…
types.h
…
union_find.h
…
util_test.cc
[XLA] Fix a bug in SplitF64ToF32
2020-11-23 11:27:37 -08:00
util.cc
[XLA] Fix a bug in SplitF64ToF32
2020-11-23 11:27:37 -08:00
util.h
Move llvm dependency from xla/util.h to xla/service/gpu/ir_emission_utils.h
2020-12-14 16:51:06 -08:00
window_util_test.cc
…
window_util.cc
…
window_util.h
…
xla_data.proto
[XLA] Add creation_pass_id and dummy op_names to OpMetadata.
2020-12-11 17:48:13 -08:00
xla.bzl
…
xla.proto
[XLA:GPU] Add --xla_gpu_deterministic_ops to force run-to-run determinism.
2020-12-15 12:51:36 -08:00

README.md

XLA (Accelerated Linear Algebra) is a domain-specific compiler for linear algebra that optimizes TensorFlow computations. See the documentation.