experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
STT-tensorflow/tensorflow/compiler/xla/service/cpu
History
Benjamin Kramer 15e0f259b1 [XLA:CPU] Move elemental conv emitter to generic code path so it can share the multiply-add logic with dot 
This fixes some edge cases when it comes to complex numbers and also would
allow using it for ints if we want that. GPU doesn't use this code path.

PiperOrigin-RevId: 351549079
Change-Id: I7a2f9e9758e270b62814c7e3e0419342c5b58196
2021-01-13 03:07:21 -08:00
..
tests
Allow outfeed of size 0 for XLA:CPU.
2020-10-12 01:43:20 -07:00
buffer_info_util.cc
…
buffer_info_util.h
…
BUILD
[XLA] IntegralUpcaster->OperandUpcaster to allow auto upcasting floating point types.
2020-12-29 11:14:04 -08:00
build_defs.bzl
…
compiler_functor.cc
[XLA:CPU] Port XLA:CPU JIT to OrcV2
2020-10-20 17:24:56 -07:00
compiler_functor.h
[XLA:CPU] Port XLA:CPU JIT to OrcV2
2020-10-20 17:24:56 -07:00
conv_canonicalization_test.cc
…
conv_canonicalization.cc
…
conv_canonicalization.h
…
cpu_compiler.cc
[XLA:CPU] Move bf16->f32 conversion before fusion
2021-01-07 05:07:58 -08:00
cpu_compiler.h
[XLA] Pack some parameters in xla::Compiler member functions into a struct. This is for later expanding for more parameters, like a thread pool for parallel compilation.
2020-12-08 01:54:36 -08:00
cpu_eigen_tensor_alignment_test.cc
…
cpu_executable.cc
[XLA:CPU] Port XLA:CPU JIT to OrcV2
2020-10-20 17:24:56 -07:00
cpu_executable.h
[XLA/CPU] Support buffer aliasing on XLA:CPU
2020-06-09 22:54:46 -07:00
cpu_instruction_fusion_test.cc
[XLA:CPU] Fuse reduce-window
2020-05-08 08:34:10 -07:00
cpu_instruction_fusion.cc
Evaluate the maximum code duplication in fusion nodes instead of average.
2020-09-02 07:36:23 -07:00
cpu_instruction_fusion.h
…
cpu_layout_assignment_test.cc
…
cpu_layout_assignment.cc
…
cpu_layout_assignment.h
…
cpu_options.cc
[XLA:CPU] Lower small matmul through MLIR linalg by default
2020-07-16 11:49:08 -07:00
cpu_options.h
[XLA:CPU] Lower small matmul through MLIR linalg by default
2020-07-16 11:49:08 -07:00
cpu_runtime_test.cc
…
cpu_runtime.cc
[XLA] NFC: Remove unnecessary ParticipantImplOutput type from Rendezvous.
2020-12-09 14:38:44 -08:00
cpu_runtime.h
[XLA:CPU] Add a runtime function for F32 TopK and use TopkRewriter to target it
2020-08-05 02:20:45 -07:00
cpu_transfer_manager.cc
Allow outfeed of size 0 for XLA:CPU.
2020-10-12 01:43:20 -07:00
cpu_transfer_manager.h
Add overload to check for device memory access semantics, akin to
2020-04-17 09:09:52 -07:00
dot_op_emitter_internal.h
…
dot_op_emitter.cc
[XLA:CPU] Allow more types to be used in dot
2021-01-07 04:32:51 -08:00
dot_op_emitter.h
[XLA:CPU] Plumb through a minimal emitter for matmuls using the mlir linalg dialect
2020-05-20 06:43:17 -07:00
elemental_ir_emitter.cc
[XLA:CPU] Move elemental conv emitter to generic code path so it can share the multiply-add logic with dot
2021-01-13 03:07:21 -08:00
elemental_ir_emitter.h
[XLA:CPU] Move elemental conv emitter to generic code path so it can share the multiply-add logic with dot
2021-01-13 03:07:21 -08:00
ir_emission_utils_test.cc
…
ir_emission_utils.cc
…
ir_emission_utils.h
…
ir_emitter.cc
[XLA:CPU] Move elemental conv emitter to generic code path so it can share the multiply-add logic with dot
2021-01-13 03:07:21 -08:00
ir_emitter.h
[XLA:CPU] Move elemental conv emitter to generic code path so it can share the multiply-add logic with dot
2021-01-13 03:07:21 -08:00
ir_function.cc
[XLA:CPU] Support multi-operand all-reduce on CPU
2020-04-06 19:50:31 -07:00
ir_function.h
[XLA:CPU] Support multi-operand all-reduce on CPU
2020-04-06 19:50:31 -07:00
llvm_ir_runtime.cc
Integrate LLVM at llvm/llvm-project@0581c0b0ee
2020-08-15 11:47:56 -07:00
llvm_ir_runtime.h
…
mlir_emitter.cc
Integrate LLVM at llvm/llvm-project@1b97cdf885
2020-12-21 23:55:25 -08:00
mlir_emitter.h
[mlir][NFC] Replace usages of Function.h and Module.h with BuiltinOps.h
2020-12-01 13:23:48 -08:00
orc_jit_memory_mapper.cc
Prefixing TensorFlow thread annotation macros with TF_.
2020-03-05 08:42:01 -08:00
orc_jit_memory_mapper.h
…
parallel_loop_emitter.cc
[XLA] Speed up. Make XLA faster by making PW kernel use the right number of block and loops.
2020-09-02 12:57:32 -07:00
parallel_loop_emitter.h
Fix code style from comments.
2020-10-01 07:47:11 -07:00
parallel_task_assignment_test.cc
[XLA:CPU] Don't parallelize constant()
2020-09-14 12:00:14 -07:00
parallel_task_assignment.cc
[XLA:CPU] Don't parallelize constant()
2020-09-14 12:00:14 -07:00
parallel_task_assignment.h
…
runtime_conv2d_impl.h
[XLA] Fix ODR violation coming from Eigen contraction kernels
2020-07-16 12:40:11 -07:00
runtime_conv2d_mkl.cc
[Intel MKL] Fix memory leak in dnnl 0.21.2
2020-02-20 14:42:03 +08:00
runtime_conv2d_mkl.h
…
runtime_conv2d.cc
Fully qualifying uses of tensorflow::int64.
2020-07-15 11:33:34 -07:00
runtime_conv2d.h
…
runtime_fft_impl.h
[XLA:CPU] [XLA:GPU] Add support for double precision FFTs on CPU and GPU.
2020-06-01 19:25:40 -07:00
runtime_fft_test.cc
Add standalone XLA AOT runtime target + relevant .cc sources to pip package.
2020-03-02 10:56:42 -08:00
runtime_fft.cc
[XLA:CPU] [XLA:GPU] Add support for double precision FFTs on CPU and GPU.
2020-06-01 19:25:40 -07:00
runtime_fft.h
[XLA:CPU] [XLA:GPU] Add support for double precision FFTs on CPU and GPU.
2020-06-01 19:25:40 -07:00
runtime_fork_join.cc
Add standalone XLA AOT runtime target + relevant .cc sources to pip package.
2020-03-02 10:56:42 -08:00
runtime_fork_join.h
…
runtime_fp16.cc
…
runtime_fp16.h
…
runtime_key_value_sort.cc
Fully qualifying uses of tensorflow::int64.
2020-07-15 11:33:34 -07:00
runtime_key_value_sort.h
…
runtime_lightweight_check.h
…
runtime_matmul_mkl.cc
…
runtime_matmul_mkl.h
…
runtime_matmul.cc
Fully qualifying uses of tensorflow::int64.
2020-07-15 11:33:34 -07:00
runtime_matmul.h
[XLA:CPU] Wire up C64/C128 matmul to Eigen
2020-05-27 09:56:35 -07:00
runtime_pow.cc
[XLA:CPU] Add our own __powisf2 and __powidf2 which can be generated by LLVM
2020-05-05 11:53:15 -07:00
runtime_pow.h
[XLA:CPU] Add our own __powisf2 and __powidf2 which can be generated by LLVM
2020-05-05 11:53:15 -07:00
runtime_single_threaded_conv2d.cc
Fully qualifying uses of tensorflow::int64.
2020-07-15 11:33:34 -07:00
runtime_single_threaded_conv2d.h
…
runtime_single_threaded_fft.cc
[XLA:CPU] [XLA:GPU] Add support for double precision FFTs on CPU and GPU.
2020-06-01 19:25:40 -07:00
runtime_single_threaded_fft.h
[XLA:CPU] [XLA:GPU] Add support for double precision FFTs on CPU and GPU.
2020-06-01 19:25:40 -07:00
runtime_single_threaded_matmul.cc
Fully qualifying uses of tensorflow::int64.
2020-07-15 11:33:34 -07:00
runtime_single_threaded_matmul.h
[XLA:CPU] Wire up C64/C128 matmul to Eigen
2020-05-27 09:56:35 -07:00
runtime_topk.cc
[XLA:CPU] Add a runtime function for F32 TopK and use TopkRewriter to target it
2020-08-05 02:20:45 -07:00
runtime_topk.h
[XLA:CPU] Add a runtime function for F32 TopK and use TopkRewriter to target it
2020-08-05 02:20:45 -07:00
sample_harness.cc
…
shape_partition_test.cc
…
shape_partition.cc
…
shape_partition.h
…
simple_orc_jit.cc
Integrate LLVM at llvm/llvm-project@c30ab6c2a3
2020-11-13 12:03:51 -08:00
simple_orc_jit.h
Integrate LLVM at llvm/llvm-project@c89447b659
2020-10-22 15:30:54 -07:00
target_machine_features_fake.h
[XLA:CPU] Teach dot_op_emitter how to tile&vectorize linalg matmuls
2020-07-08 06:32:43 -07:00
target_machine_features.cc
[XLA:CPU] Reconsider alignment heuristics
2020-04-24 02:51:55 -07:00
target_machine_features.h
[XLA:CPU] Teach dot_op_emitter how to tile&vectorize linalg matmuls
2020-07-08 06:32:43 -07:00
test_target_triple_helper.h
Add s390x support for compiler related TCs
2020-11-18 08:34:42 -08:00
tiled_dot_emitter.cc
…
tiled_dot_emitter.h
…
vector_support_library.cc
Integrate LLVM at llvm/llvm-project@0581c0b0ee
2020-08-15 11:47:56 -07:00
vector_support_library.h
Integrate LLVM at llvm/llvm-project@131b3b9ed4
2020-08-20 09:01:58 -07:00
vectorized_reduce_with_no_vector_registers_test.cc
Merge pull request #38980 from npanpaliya:tf-cpu-xla-test-fixes
2020-06-18 02:37:49 -07:00
windows_compatibility.cc
…
windows_compatibility.h
…
xfeed_manager_test.cc
Allow outfeed of size 0 for XLA:CPU.
2020-10-12 01:43:20 -07:00
xfeed_manager.cc
…
xfeed_manager.h
…