Update name from xtensa to cadence (#2982)

Summary:
Pull Request resolved: #2982

As titled.

Reviewed By: cccclai

Differential Revision: D55998135

fbshipit-source-id: a57bd233afe170290c7def4406d6d6e769d467ed

Author: mcremon-meta

docs/source/build-run-xtensa.md

@@ -64,7 +64,7 @@ Step 2. Make sure you have completed the ExecuTorch setup tutorials linked to at
 The working tree is:
 
 ```
-examples/xtensa/
+examples/cadence/
 ├── aot
 ├── kernels
 ├── ops
@@ -75,7 +75,7 @@ examples/xtensa/
 
 ***AoT (Ahead-of-Time) Components***:
 
-The AoT folder contains all of the python scripts and functions needed to export the model to an ExecuTorch `.pte` file. In our case, [export_example.py](https://github.com/pytorch/executorch/blob/main/examples/xtensa/aot/export_example.py) is an API that takes a model (nn.Module) and representative inputs and runs it through the quantizer (from [quantizer.py](https://github.com/pytorch/executorch/blob/main/examples/xtensa/aot/quantizer.py)). Then a few compiler passes, also defined in [quantizer.py](https://github.com/pytorch/executorch/blob/main/examples/xtensa/aot/quantizer.py), will replace operators with custom ones that are supported and optimized on the chip. Any operator needed to compute things should be defined in [meta_registrations.py](https://github.com/pytorch/executorch/blob/main/examples/xtensa/aot/meta_registrations.py) and have corresponding implemetations in the other folders.
+The AoT folder contains all of the python scripts and functions needed to export the model to an ExecuTorch `.pte` file. In our case, [export_example.py](https://github.com/pytorch/executorch/blob/main/examples/cadence/aot/export_example.py) is an API that takes a model (nn.Module) and representative inputs and runs it through the quantizer (from [quantizer.py](https://github.com/pytorch/executorch/blob/main/examples/cadence/aot/quantizer.py)). Then a few compiler passes, also defined in [quantizer.py](https://github.com/pytorch/executorch/blob/main/examples/cadence/aot/quantizer.py), will replace operators with custom ones that are supported and optimized on the chip. Any operator needed to compute things should be defined in [meta_registrations.py](https://github.com/pytorch/executorch/blob/main/examples/cadence/aot/meta_registrations.py) and have corresponding implemetations in the other folders.
 
 ***Operators***:
 
@@ -101,14 +101,14 @@ python3 -m examples.portable.scripts.export --model_name="add"
 ***Quantized Operators***:
 
 The other, more complex model are custom operators, including:
-  - a quantized [linear](https://pytorch.org/docs/stable/generated/torch.nn.Linear.html) operation. The model is defined [here](https://github.com/pytorch/executorch/blob/main/examples/xtensa/tests/quantized_linear_example.py#L28). Linear is the backbone of most Automatic Speech Recognition (ASR) models.
-  - a quantized [conv1d](https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html) operation. The model is defined [here](https://github.com/pytorch/executorch/blob/main/examples/xtensa/tests/quantized_conv1d_example.py#L36). Convolutions are important in wake word and many denoising models.
+  - a quantized [linear](https://pytorch.org/docs/stable/generated/torch.nn.Linear.html) operation. The model is defined [here](https://github.com/pytorch/executorch/blob/main/examples/cadence/tests/quantized_linear_example.py#L28). Linear is the backbone of most Automatic Speech Recognition (ASR) models.
+  - a quantized [conv1d](https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html) operation. The model is defined [here](https://github.com/pytorch/executorch/blob/main/examples/cadence/tests/quantized_conv1d_example.py#L36). Convolutions are important in wake word and many denoising models.
 
 In both cases the generated file is called `XtensaDemoModel.pte`.
 
 ```bash
 cd executorch
-python3 -m examples.xtensa.tests.quantized_<linear,conv1d>_example
+python3 -m examples.cadence.tests.quantized_<linear,conv1d>_example
 ```
 
 ***Small Model: RNNT predictor***:
@@ -118,7 +118,7 @@ The predictor is a sequence of basic ops (embedding, ReLU, linear, layer norm) a
 
 ```bash
 cd executorch
-python3 -m examples.xtensa.tests.rnnt_predictor_quantized_example
+python3 -m examples.cadence.tests.rnnt_predictor_quantized_example
 ```
 
 The generated file is called `XtensaDemoModel.pte`.
@@ -131,7 +131,7 @@ In this step, you'll be building the DSP firmware image that consists of the sam
 ***Step 1***. Configure the environment variables needed to point to the Xtensa toolchain that you have installed in the previous step. The three environment variables that need to be set include:
 ```bash
 # Directory in which the Xtensa toolchain was installed
-export XTENSA_TOOLCHAIN=/home/user_name/xtensa/XtDevTools/install/tools
+export XTENSA_TOOLCHAIN=/home/user_name/cadence/XtDevTools/install/tools
 # The version of the toolchain that was installed. This is essentially the name of the directory
 # that is present in the XTENSA_TOOLCHAIN directory from above.
 export TOOLCHAIN_VER=RI-2021.8-linux
@@ -151,7 +151,7 @@ cd executorch
 rm -rf cmake-out
 # prebuild and install executorch library
 cmake -DBUCK2=buck2 \
-    -DCMAKE_TOOLCHAIN_FILE=<path_to_executorch>/examples/xtensa/xtensa.cmake \
+    -DCMAKE_TOOLCHAIN_FILE=<path_to_executorch>/examples/cadence/cadence.cmake \
     -DCMAKE_INSTALL_PREFIX=cmake-out \
     -DCMAKE_BUILD_TYPE=Debug \
     -DPYTHON_EXECUTABLE=python3 \
@@ -165,18 +165,18 @@ cmake -DBUCK2=buck2 \
     -Bcmake-out .
 
 cmake --build cmake-out -j8 --target install --config Debug
-# build xtensa runner
+# build cadence runner
 cmake -DCMAKE_BUILD_TYPE=Debug \
-    -DCMAKE_TOOLCHAIN_FILE=<path_to_executorch>/examples/xtensa/xtensa.cmake \
+    -DCMAKE_TOOLCHAIN_FILE=<path_to_executorch>/examples/cadence/cadence.cmake \
     -DCMAKE_PREFIX_PATH=<path_to_executorch>/cmake-out \
     -DMODEL_PATH=<path_to_program_file_generated_in_previous_step> \
     -DNXP_SDK_ROOT_DIR=<path_to_nxp_sdk_root> -DEXECUTORCH_BUILD_FLATC=0 \
     -DFLATC_EXECUTABLE="$(which flatc)" \
     -DNN_LIB_BASE_DIR=<path_to_nnlib_cloned_in_step_2> \
-    -Bcmake-out/examples/xtensa \
-    examples/xtensa
+    -Bcmake-out/examples/cadence \
+    examples/cadence
 
-cmake --build cmake-out/examples/xtensa -j8 -t xtensa_executorch_example
+cmake --build cmake-out/examples/cadence -j8 -t cadence_executorch_example
 ```
 
 After having succesfully run the above step you should see two binary files in their CMake output directory.
@@ -213,6 +213,6 @@ First 20 elements of output 0
 
 In this tutorial, you have learned how to export a quantized operation, build the ExecuTorch runtime and run this model on the Xtensa HiFi4 DSP chip.
 
-The (quantized linear) model in this tutorial is a typical operation appearing in ASR models, and can be extended to a complete ASR model by creating the model as a new test and adding the needed operators/kernels to [operators](https://github.com/pytorch/executorch/blob/main/examples/xtensa/ops) and [kernels](https://github.com/pytorch/executorch/blob/main/examples/xtensa/kernels).
+The (quantized linear) model in this tutorial is a typical operation appearing in ASR models, and can be extended to a complete ASR model by creating the model as a new test and adding the needed operators/kernels to [operators](https://github.com/pytorch/executorch/blob/main/examples/cadence/ops) and [kernels](https://github.com/pytorch/executorch/blob/main/examples/cadence/kernels).
 
 Other models can be created following the same structure, always assuming that operators and kernels are available.

examples/xtensa/CMakeLists.txt renamed to examples/cadence/CMakeLists.txt

@@ -12,7 +12,7 @@ if(NOT CMAKE_CXX_STANDARD)
 endif()
 
 # Set the project name.
-project(xtensa_executorch_example)
+project(cadence_executorch_example)
 
 # Source root directory for executorch.
 if(NOT EXECUTORCH_ROOT)
@@ -100,21 +100,21 @@ add_custom_command(
 
 add_custom_target(gen_model_header DEPENDS ${CMAKE_BINARY_DIR}/model_pte.h)
 
-add_executable(xtensa_executorch_example executor_runner.cpp)
-add_dependencies(xtensa_executorch_example gen_model_header)
+add_executable(cadence_executorch_example executor_runner.cpp)
+add_dependencies(cadence_executorch_example gen_model_header)
 
 # lint_cmake: -linelength
-target_include_directories(xtensa_executorch_example PUBLIC ${ROOT_DIR}/..
+target_include_directories(cadence_executorch_example PUBLIC ${ROOT_DIR}/..
                                                             ${CMAKE_BINARY_DIR}
                                                             ${_common_include_directories})
 
-target_link_options(xtensa_executorch_example PRIVATE
+target_link_options(cadence_executorch_example PRIVATE
                     -mlsp=${NXP_SDK_ROOT_DIR}/devices/MIMXRT685S/xtensa/min-rt)
-target_link_libraries(xtensa_executorch_example dsp_mu_polling_libs
-                      xtensa_ops_lib extension_runner_util executorch)
+target_link_libraries(cadence_executorch_example dsp_mu_polling_libs
+                      cadence_ops_lib extension_runner_util executorch)
 
 add_custom_command(
-  TARGET xtensa_executorch_example
+  TARGET cadence_executorch_example
   POST_BUILD
   COMMAND
     ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_LIST_DIR}/utils/post_compilation.py

examples/xtensa/aot/compiler.py renamed to examples/cadence/aot/compiler.py

@@ -17,20 +17,20 @@
 
 from .compiler import export_to_edge
 from .quantizer import (
+    CadenceBaseQuantizer,
     QuantFusion,
-    ReplacePT2DequantWithXtensaDequant,
-    ReplacePT2QuantWithXtensaQuant,
-    XtensaBaseQuantizer,
+    ReplacePT2DequantWithCadenceDequant,
+    ReplacePT2QuantWithCadenceQuant,
 )
 
 
 FORMAT = "[%(levelname)s %(asctime)s %(filename)s:%(lineno)s] %(message)s"
 logging.basicConfig(level=logging.INFO, format=FORMAT)
 
 
-def export_xtensa_model(model, example_inputs):
+def export_model(model, example_inputs):
     # Quantizer
-    quantizer = XtensaBaseQuantizer()
+    quantizer = CadenceBaseQuantizer()
 
     # Export
     model_exp = capture_pre_autograd_graph(model, example_inputs)
@@ -42,24 +42,24 @@ def export_xtensa_model(model, example_inputs):
     # Convert
     converted_model = convert_pt2e(prepared_model)
 
-    # pyre-fixme[16]: Pyre doesn't get that XtensaQuantizer has a patterns attribute
+    # pyre-fixme[16]: Pyre doesn't get that CadenceQuantizer has a patterns attribute
     patterns = [q.pattern for q in quantizer.quantizers]
     QuantFusion(patterns)(converted_model)
 
-    # Get edge program (note: the name will change to export_to_xtensa in future PRs)
+    # Get edge program (note: the name will change to export_to_cadence in future PRs)
     edge_prog_manager = export_to_edge(converted_model, example_inputs, pt2_quant=True)
 
     # Run a couple required passes for quant/dequant ops
-    xtensa_prog_manager = edge_prog_manager.transform(
-        [ReplacePT2QuantWithXtensaQuant(), ReplacePT2DequantWithXtensaDequant()],
+    cadence_prog_manager = edge_prog_manager.transform(
+        [ReplacePT2QuantWithCadenceQuant(), ReplacePT2DequantWithCadenceDequant()],
         check_ir_validity=False,
     )
 
-    exec_prog = xtensa_prog_manager.to_executorch()
+    exec_prog = cadence_prog_manager.to_executorch()
 
     logging.info(
         f"Final exported graph module:\n{exec_prog.exported_program().graph_module}"
     )
 
-    # Save the program as XtensaDemoModel.pte
-    save_pte_program(exec_prog, "XtensaDemoModel")
+    # Save the program as CadenceDemoModel.pte
+    save_pte_program(exec_prog, "CadenceDemoModel")

examples/xtensa/aot/export_example.py renamed to examples/cadence/aot/export_example.py

@@ -12,7 +12,7 @@
 
 from .utils import get_conv1d_output_size
 
-lib = Library("xtensa", "DEF")
+lib = Library("cadence", "DEF")
 
 lib.define(
     "quantize_per_tensor(Tensor input, float scale, int zero_point, int quant_min, int quant_max, ScalarType dtype) -> (Tensor Z)"
@@ -56,7 +56,7 @@
     "quantized_conv.out(Tensor input, Tensor weight, Tensor bias, int[] stride, SymInt[] padding, int[] dilation, int groups, int input_zero_point, Tensor weight_zero_point, Tensor bias_scale, float out_scale, int out_zero_point, Tensor out_multiplier, Tensor out_shift, bool channel_last=False, *, Tensor(a!) out) -> Tensor(a!)"
 )
 
-m = Library("xtensa", "IMPL", "Meta")
+m = Library("cadence", "IMPL", "Meta")
 
 
 @impl(m, "quantize_per_tensor")

examples/xtensa/aot/meta_registrations.py renamed to examples/cadence/aot/meta_registrations.py

@@ -437,7 +437,7 @@ def get_anchors(
         )
 
     def replacement_op(self):
-        return torch.ops.xtensa.quantized_linear.default
+        return torch.ops.cadence.quantized_linear.default
 
 
 class LinearFunctionalPattern(QuantizationPattern):
@@ -457,7 +457,7 @@ def get_anchors(
         )
 
     def replacement_op(self):
-        return torch.ops.xtensa.quantized_linear.default
+        return torch.ops.cadence.quantized_linear.default
 
 
 class LayerNormPattern(QuantizationPattern):
@@ -476,7 +476,7 @@ def get_anchors(self, gm, fused_partition) -> PartitionAnchors:
         )
 
     def replacement_op(self):
-        return torch.ops.xtensa.quantized_layer_norm.default
+        return torch.ops.cadence.quantized_layer_norm.default
 
 
 class Conv1dPattern(QuantizationPattern):
@@ -503,7 +503,7 @@ def get_anchors(
         )
 
     def replacement_op(self):
-        return torch.ops.xtensa.quantized_conv.default
+        return torch.ops.cadence.quantized_conv.default
 
 
 class Conv2dPattern(QuantizationPattern):
@@ -530,7 +530,7 @@ def get_anchors(
         )
 
     def replacement_op(self):
-        return torch.ops.xtensa.quantized_conv.default
+        return torch.ops.cadence.quantized_conv.default
 
 
 class AddmmPattern(QuantizationPattern):
@@ -550,7 +550,7 @@ def get_anchors(
         )
 
     def replacement_op(self):
-        return torch.ops.xtensa.quantized_linear.default
+        return torch.ops.cadence.quantized_linear.default
 
 
 class ReluPattern(QuantizationPattern):
@@ -573,7 +573,7 @@ def get_anchors(
         )
 
     def replacement_op(self):
-        return torch.ops.xtensa.quantized_relu.default
+        return torch.ops.cadence.quantized_relu.default
 
 
 class GenericQuantizer(Quantizer):
@@ -657,7 +657,7 @@ def get_supported_operators(cls) -> List[OperatorConfig]:
 )
 
 
-class XtensaBaseQuantizer(ComposableQuantizer):
+class CadenceBaseQuantizer(ComposableQuantizer):
     def __init__(self):
         static_qconfig = QuantizationConfig(
             act_qspec,
@@ -821,34 +821,34 @@ def mark_fused(cls, nodes) -> bool:
             n.meta["QuantFusion"] = True
 
 
-class ReplacePT2QuantWithXtensaQuant(ExportPass):
+class ReplacePT2QuantWithCadenceQuant(ExportPass):
     """
-    Replace the pt2 quantization ops with custom xtensa quantization ops.
+    Replace the pt2 quantization ops with custom cadence quantization ops.
     """
 
     def call_operator(self, op, args, kwargs, meta):
         if op not in {exir_ops.edge.quantized_decomposed.quantize_per_tensor.default}:
             return super().call_operator(op, args, kwargs, meta)
 
         return super().call_operator(
-            exir_ops.edge.xtensa.quantize_per_tensor.default,
+            exir_ops.edge.cadence.quantize_per_tensor.default,
             args,
             kwargs,
             meta,
         )
 
 
-class ReplacePT2DequantWithXtensaDequant(ExportPass):
+class ReplacePT2DequantWithCadenceDequant(ExportPass):
     """
-    Replace the pt2 dequantization ops with custom xtensa dequantization ops.
+    Replace the pt2 dequantization ops with custom cadence dequantization ops.
     """
 
     def call_operator(self, op, args, kwargs, meta):
         if op not in {exir_ops.edge.quantized_decomposed.dequantize_per_tensor.default}:
             return super().call_operator(op, args, kwargs, meta)
 
         return super().call_operator(
-            exir_ops.edge.xtensa.dequantize_per_tensor.default,
+            exir_ops.edge.cadence.dequantize_per_tensor.default,
             args,
             kwargs,
             meta,

examples/xtensa/aot/quantizer.py renamed to examples/cadence/aot/quantizer.py

@@ -12,7 +12,7 @@
  * This is a simple executor_runner that boots up the DSP, configures the serial
  * port, sends a bunch of test messages to the M33 core and then loads the model
  * defined in model_pte.h. It runs this model using the ops available in
- * xtensa/ops directory.
+ * cadence/ops directory.
  */
 
 #include <fsl_debug_console.h>

examples/xtensa/aot/utils.py renamed to examples/cadence/aot/utils.py

@@ -5,10 +5,10 @@
 # LICENSE file in the root directory of this source tree.
 
 # lint_cmake: -linelength
-add_library(xtensa_kernels kernels.cpp ${EXECUTORCH_ROOT}/examples/xtensa/third-party/nnlib-hifi4/matmul_asym8uxasym8u_asym8u.cpp)
+add_library(cadence_kernels kernels.cpp ${EXECUTORCH_ROOT}/examples/cadence/third-party/nnlib-hifi4/matmul_asym8uxasym8u_asym8u.cpp)
 
 target_include_directories(
-  xtensa_kernels
+  cadence_kernels
   PUBLIC .
          ${NN_LIB_BASE_DIR}/xa_nnlib/algo/common/include/
          ${NN_LIB_BASE_DIR}/xa_nnlib/include/nnlib

examples/xtensa/xtensa.cmake renamed to examples/cadence/cadence.cmake

@@ -31,16 +31,16 @@ set(_aten_ops__srcs
     "${CMAKE_CURRENT_SOURCE_DIR}/op_view_copy.cpp"
     "${EXECUTORCH_ROOT}/kernels/portable/cpu/util/broadcast_util.cpp"
     "${EXECUTORCH_ROOT}/kernels/portable/cpu/util/repeat_util.cpp")
-add_library(aten_ops_xtensa ${_aten_ops__srcs})
-target_link_libraries(aten_ops_xtensa PUBLIC executorch)
-target_link_libraries(aten_ops_xtensa PRIVATE xtensa_kernels)
+add_library(aten_ops_cadence ${_aten_ops__srcs})
+target_link_libraries(aten_ops_cadence PUBLIC executorch)
+target_link_libraries(aten_ops_cadence PRIVATE cadence_kernels)
 
 # Let files say "include <executorch/path/to/header.h>".
 set(_common_include_directories ${EXECUTORCH_ROOT}/..)
 
-target_include_directories(aten_ops_xtensa PUBLIC ${ROOT_DIR}/..
-                                                  ${CMAKE_BINARY_DIR}
-                                                  ${_common_include_directories})
+target_include_directories(aten_ops_cadence PUBLIC ${ROOT_DIR}/..
+                                                ${CMAKE_BINARY_DIR}
+                                                ${_common_include_directories})
 
 # Custom ops that are needed to run the test model.
 add_library(
@@ -52,7 +52,7 @@ target_include_directories(custom_ops PUBLIC ${ROOT_DIR}/..
                                              ${_common_include_directories})
 
 target_link_libraries(custom_ops PUBLIC executorch)
-target_link_libraries(custom_ops PRIVATE xtensa_kernels)
+target_link_libraries(custom_ops PRIVATE cadence_kernels)
 
 # Generate C++ bindings to register kernels into both PyTorch (for AOT) and
 # Executorch (for runtime). Here select all ops in functions.yaml
@@ -62,6 +62,6 @@ generate_bindings_for_kernels(
 message("Generated files ${gen_command_sources}")
 
 gen_operators_lib(
-  "xtensa_ops_lib"
+  "cadence_ops_lib"
   KERNEL_LIBS custom_ops
-  DEPS aten_ops_xtensa)
+  DEPS aten_ops_cadence)