- 09 8月, 2019 40 次提交
-
-
由 Benoit Jacob 提交于
See the comment. PiperOrigin-RevId: 262564280
-
由 Dero Gharibian 提交于
Also, minor cleanup of methods. PiperOrigin-RevId: 262564087
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 262562281
-
由 Benoit Jacob 提交于
PiperOrigin-RevId: 262558564
-
由 Nicolas Vasilache 提交于
This CL introduces the ability to generate the external library name for Linalg operations. The problem is that neither mlir or C support overloading and we want a simplified form of name mangling that is still reasonable to read. This CL creates the name of the external call that Linalg expects from the operation name and the type of its arguments. The interface library names are updated and use new cases are added for FillOp. PiperOrigin-RevId: 262556833
-
由 Nicolas Vasilache 提交于
This CL adds the ability for linalg.view to act as a bitcast operation. This will be used when promoting views into faster memory and casting to vector types. In the process, linalg.view is moved to ODS. PiperOrigin-RevId: 262556246
-
由 Nicolas Vasilache 提交于
This CL is step 2/n towards building a simple, programmable and portable vector abstraction in MLIR that can go all the way down to generating assembly vector code via LLVM's opt and llc tools. This CL adds the vector.outerproduct operation to the MLIR vector dialect as well as the appropriate roundtrip test. Lowering to LLVM will occur in the following CL. PiperOrigin-RevId: 262552027
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 262551455
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 262550835
-
由 Nicolas Vasilache 提交于
This CL is step 2/n towards building a simple, programmable and portable vector abstraction in MLIR that can go all the way down to generating assembly vector code via LLVM's opt and llc tools. This CL adds the vector.extractelement operation to the MLIR vector dialect as well as the appropriate roundtrip test. Lowering to LLVM will occur in the following CL. PiperOrigin-RevId: 262545089
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 262544523
-
由 Nicolas Vasilache 提交于
This CL is step 1/n towards building a simple, programmable and portable vector abstraction in MLIR that can go all the way down to generating assembly vector code via LLVM's opt and llc tools. This CL adds the 3 instructions `llvm.extractelement`, `llvm.insertelement` and `llvm.shufflevector` as documented in the LLVM LangRef "Vector Instructions" section. The "Experimental Vector Reduction Intrinsics" are left out for now and can be added in the future on a per-need basis. Appropriate roundtrip and LLVM Target tests are added. PiperOrigin-RevId: 262542095
-
由 A. Unique TensorFlower 提交于
Introduce an operation that defines global constants and variables in the LLVM dialect, to reflect the corresponding LLVM IR capability. This operation is expected to live in the top-level module and behaves similarly to llvm.constant. It currently does not model many of the attributes supported by the LLVM IR for global values (memory space, alignment, thread-local, linkage) and will be extended as the relevant use cases appear. PiperOrigin-RevId: 262539445
-
由 Sergei Lebedev 提交于
The default layout is still ICFO, but V2 version of the op will use IFCO to match CuDNN-RNN. PiperOrigin-RevId: 262538325
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 262533270
-
由 Sergei Lebedev 提交于
CPU/GPU ops are now declared consistently using the same macro names. This change helps minimize the diff for the followup which generalizes the kernels/ops wrt the gate layout. PiperOrigin-RevId: 262523295
-
由 Sergei Lebedev 提交于
This makes it simpler to port it to C++ later. PiperOrigin-RevId: 262522805
-
由 Alexander Belyaev 提交于
PiperOrigin-RevId: 262521118
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262521085
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262521068
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 262512394
-
由 Brian Zhao 提交于
PiperOrigin-RevId: 262505241
-
由 Brian Zhao 提交于
PiperOrigin-RevId: 262501165
-
由 Ayush Dubey 提交于
PiperOrigin-RevId: 262498505
-
由 Yifei Feng 提交于
PiperOrigin-RevId: 262494606
-
由 Amit Patankar 提交于
Rename `tf_pybind_extension` to `pybind_extension` as it doesn't necessarily build always build TF related things. Add a specific `tf_python_pybind_extension` for pybind11 that links to libtensorflow_framework.so and pywrap_tensorflow_internal.so. This is a precursor for our migration to pybind11 as documented in https://github.com/tensorflow/community/blob/master/rfcs/20190208-pybind11.md PiperOrigin-RevId: 262491770
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262488994
-
由 Nupur Garg 提交于
PiperOrigin-RevId: 262486343
-
由 Kibeom Kim 提交于
Tensors will be unhashable starting TF 2.0, so disallow a dictionary argument with Tensor as a key for tf.case PiperOrigin-RevId: 262483688
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262483401
-
由 Juho Ha 提交于
SIN, COS, LOG, SQRT, RSQRT, SQUARE PiperOrigin-RevId: 262482624
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262482235
-
由 Katherine Wu 提交于
PiperOrigin-RevId: 262481484
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262480689
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262479408
-
由 Jaesung Chung 提交于
PiperOrigin-RevId: 262478980
-
由 George Karpenkov 提交于
Returning `bool` is ambiguous, error-prone, and hides the failure message. PiperOrigin-RevId: 262477427
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 262476676
-
由 richardbrks 提交于
-
由 Chao Mei 提交于
PiperOrigin-RevId: 262475573
-