Update base for Update on "[ET-VK] Clean up api::vTensor class"

## Context

Now that we have forked the `api/` directory from PyTorch Vulkan, we can clean up the `vTensor` class and remove functionality that is not necessary for the ExecuTorch Vulkan delegate.

The following changes are made:

* Remove unused member variables and member functions from `vTensor` and `vTensorStorage`
* Remove all quantization related member variables, member functions, and the `vTensor` constructor for quantized tensors. The Quantization API will be reworked from the ground up.
* Rename `view_` (which is an instance of `vTensorStorage`) to `storage_`

Finally, the critical change that is introduced is that we now store `storage_` as a direct `vTensorStorage` member variable in `vTensor` instead of storing it as a `std::shared_ptr<vTensorStorage>`.

For context, the reason `storage_` was stored as a shared pointer is to be compliant with ATen Tensors, which needs to enable copy construction to enable the following:

```
at::Tensor b = at::rand(...);
// Oftentimes this will create a "view" of the tensor. a and b will point the the same underlying storage, but with different metadata.
at::Tensor a = b;
```

However, in the ExecuTorch delegate this is no longer necessary. Each Tensor is associated with it's own independent storage and is responsible for managing it's own memory. **By getting rid of `std::shared_ptr`, we can avoid a heap allocation and avoid chasing pointers whenever we need to access the resources of a `vTensor`.**

Differential Revision: [D55811279](https://our.internmc.facebook.com/intern/diff/D55811279/)

[ghstack-poisoned]

Author: SS-JIA

backends/qualcomm/builders/node_visitor.py

@@ -29,6 +29,7 @@
     QNN_uint16: PyQnnWrapper.Qnn_DataType_t.QNN_DATATYPE_UFIXED_POINT_16,
 }
 QNN_TENSOR_TYPE_MAP = {
+    torch.bool: PyQnnWrapper.Qnn_DataType_t.QNN_DATATYPE_BOOL_8,
     torch.float32: PyQnnWrapper.Qnn_DataType_t.QNN_DATATYPE_FLOAT_32,
     torch.int8: PyQnnWrapper.Qnn_DataType_t.QNN_DATATYPE_INT_8,
     torch.int16: PyQnnWrapper.Qnn_DataType_t.QNN_DATATYPE_INT_16,

backends/qualcomm/partition/common_defs.py

@@ -13,6 +13,8 @@
     exir_ops.edge.aten.clone.default,
     exir_ops.edge.aten.index.Tensor,
     exir_ops.edge.aten.full.default,
+    exir_ops.edge.aten.slice_scatter.default,
+    exir_ops.edge.aten.index_put.default,
 ]
 
 allow_list_operator = [

examples/models/llama2/README.md

@@ -17,9 +17,9 @@ Please note that the models are subject to the [acceptable use policy](https://g
 
 # Results
 
-Since 7B Llama2 model needs at least 4-bit quantization to fit even within some of the highend phones, results presented here correspond to 4-bit groupwise post-training quantized model. 
+Since 7B Llama2 model needs at least 4-bit quantization to fit even within some of the highend phones, results presented here correspond to 4-bit groupwise post-training quantized model.
 
-For Llama3, we can use the same process. Note that it's only supported in the ExecuTorch main branch. 
+For Llama3, we can use the same process. Note that it's only supported in the ExecuTorch main branch.
 
 ## Quantization:
 We employed 4-bit groupwise per token dynamic quantization of all the linear layers of the model. Dynamic quantization refers to quantizating activations dynamically, such that quantization parameters for activations are calculated, from min/max range, at runtime. Here we quantized activations with 8bits (signed integer). Furthermore, weights are statically quantized. In our case weights were per-channel groupwise quantized with 4bit signed integer. For more information refer to this [page](https://github.com/pytorch-labs/ao/).
@@ -57,7 +57,7 @@ Performance was measured on Samsung Galaxy S22, S24, One Plus 12 and iPhone 15 m
 - For Llama7b, your device may require at least 32GB RAM. If this is a constraint for you, please try the smaller stories model.
 
 ## Step 1: Setup
-1. Follow the [tutorial](https://pytorch.org/executorch/main/getting-started-setup) to set up ExecuTorch
+1. Follow the [tutorial](https://pytorch.org/executorch/main/getting-started-setup) to set up ExecuTorch. For installation run `./install_requirements.sh --pybind xnnpack`
 2. Run `examples/models/llama2/install_requirements.sh` to install a few dependencies.
 
 ## Step 2: Prepare model
@@ -103,6 +103,16 @@ If you want to deploy and run a smaller model for educational purposes. From `ex
     python -m examples.models.llama2.tokenizer.tokenizer -t tokenizer.model -o tokenizer.bin
     ```
 
+### Option C: Download and export Llama3 8B model
+
+You can export and run the original Llama3 8B model.
+
+1. Llama3 pretrained parameters can be downloaded from [Meta's official llama3 repository](https://github.com/meta-llama/llama3/).
+
+2. Export model and generate `.pte` file
+    ```
+    python -m examples.models.llama2.export_llama --checkpoint <consolidated.00.pth> -p <params.json> -d=fp32 -X -qmode 8da4w -kv --use_sdpa_with_kv_cache --output_name="llama3_kv_sdpa_xnn_qe_4_32.pte" group_size 128 --metadata '{"get_bos_id":128000, "get_eos_id":128001}' --embedding-quantize 4,32
+    ```
 
 ## (Optional) Finetuning
 
@@ -148,6 +158,7 @@ The Uncyclotext results generated above used: `{max_seq_len: 2048, limit: 1000}`
         -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
         -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
         -DEXECUTORCH_BUILD_XNNPACK=ON \
+        -DEXECUTORCH_BUILD_QUANTIZED=ON \
         -DEXECUTORCH_BUILD_OPTIMIZED=ON \
         -DEXECUTORCH_BUILD_CUSTOM=ON \
         -Bcmake-out .
@@ -163,17 +174,22 @@ The Uncyclotext results generated above used: `{max_seq_len: 2048, limit: 1000}`
         -DEXECUTORCH_BUILD_CUSTOM=ON \
         -DEXECUTORCH_BUILD_OPTIMIZED=ON \
         -DEXECUTORCH_BUILD_XNNPACK=ON \
+        -DEXECUTORCH_BUILD_QUANTIZED=ON \
         -Bcmake-out/examples/models/llama2 \
         examples/models/llama2
 
     cmake --build cmake-out/examples/models/llama2 -j16 --config Release
     ```
 
+For Llama3, add `-DEXECUTORCH_USE_TIKTOKEN=ON` option when building the llama runner.
+
 3. Run model. Run options available [here](https://github.com/pytorch/executorch/blob/main/examples/models/llama2/main.cpp#L18-L40).
     ```
     cmake-out/examples/models/llama2/llama_main --model_path=<model pte file> --tokenizer_path=<tokenizer.bin> --prompt=<prompt>
     ```
 
+For Llama3, you can pass the original `tokenizer.model` (without converting to `.bin` file).
+
 ## Step 5: Run benchmark on Android phone
 
 **1. Build llama runner binary for Android**
@@ -271,7 +287,7 @@ This example tries to reuse the Python code, with minimal modifications to make
 ```
 git clean -xfd
 pip uninstall executorch
-./install_requirements.sh <options>
+./install_requirements.sh --pybind xnnpack
 
 rm -rf cmake-out
 ```

examples/models/llama2/export_llama_lib.py

@@ -355,6 +355,13 @@ def build_args_parser() -> argparse.ArgumentParser:
     parser.add_argument(
         "--pt2e_quantize",
         default=None,
+        choices=[
+            "xnnpack_dynamic",
+            "xnnpack_dynamic_qc4",
+            "qnn_8a8w",
+            "qnn_16a16w",
+            "qnn_16a4w",
+        ],
         help="Use PT2E quantization. Comma separated options. e.g. xnnpack_dynamic (for per channel 8 bit weight), xnnpack_dynamic_qc4 (for per channel 4 bit weight), embedding.",
     )
     parser.add_argument(
@@ -627,6 +634,9 @@ def _prepare_for_llama_export(modelname: str, args) -> LlamaEdgeManager:
     if args.use_sdpa_with_kv_cache:
         transforms.append(replace_sdpa_with_custom_op)
 
+    if args.qnn and args.use_kv_cache:
+        transforms.append(replace_sdpa_with_simple_sdpa)
+        transforms.append(replace_causal_mask)
     return (
         load_llama_model(
             modelname=modelname,
@@ -650,13 +660,16 @@ def _export_llama(modelname, args) -> str:  # noqa: C901
     # export_to_edge
     pt2e_quant_params = _get_pt2e_quantization_params(args)
     quantizers = get_pt2e_quantizers(pt2e_quant_params, args)
-    if args.qnn:
-        assert (
-            args.quantization_mode is None
-        ), "Currently qnn backend only supports QnnQuantizer via pt2e flow"
+    quant_dtype = None
+    if args.qnn and args.pt2e_quantize:
         try:
             # pyre-ignore: Undefined import [21]: Could not find a module corresponding to import `executorch.backends.qualcomm.quantizer.quantizer`
-            from executorch.backends.qualcomm.quantizer.quantizer import QnnQuantizer
+            from executorch.backends.qualcomm.quantizer.quantizer import (
+                get_16a4w_qnn_ptq_config,
+                get_default_16bit_qnn_ptq_config,
+                QnnQuantizer,
+                QuantDtype,
+            )
 
             # reset quantizers and pt2e_quant_params from xnnpack backend
             pt2e_quant_params = None
@@ -666,10 +679,41 @@ def _export_llama(modelname, args) -> str:  # noqa: C901
                 "Please install the Qualcomm backend follwing https://pytorch.org/executorch/main/build-run-qualcomm.html"
             )
 
+        backend, quant_config = args.pt2e_quantize.split("_")
+        assert (
+            backend == "qnn"
+        ), f"The quantization config is for backend {backend} instead of qnn."
         # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
         qnn_quantizer = QnnQuantizer()
         # more custom quantization are supported including 16a4w etc. default to 8bit quantized
         custom_annotations = ()
+        if quant_config == "8a8w":
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            quant_dtype = QuantDtype.use_8a8w
+            pass
+        elif quant_config == "16a16w":
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            quant_dtype = QuantDtype.use_16a16w
+            qnn_quantizer.add_16bit_quant_ops(qnn_quantizer.SUPPORTED_OPS)
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            qnn_quantizer.set_bit16_op_quant_config(get_default_16bit_qnn_ptq_config())
+        elif quant_config == "16a4w":
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            quant_dtype = QuantDtype.use_16a4w
+            qnn_quantizer.add_16bit_quant_ops(qnn_quantizer.SUPPORTED_OPS)
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            qnn_quantizer.set_bit16_op_quant_config(get_16a4w_qnn_ptq_config())
+            qnn_quantizer.set_per_channel_weight_dtype(
+                weight_dtype_for_16bit_act="int4"
+            )
+        else:
+            raise AssertionError(
+                f"No support for quant type {quant_config}. Support 8a8w, 16a16w and 16a4w."
+            )
+
+        assert (
+            args.quantization_mode is None
+        ), "Currently qnn backend only supports QnnQuantizer via pt2e flow"
         qnn_quantizer.add_custom_quant_annotations(custom_annotations)
         quantizers.append(qnn_quantizer)
 
@@ -786,25 +830,38 @@ def _export_llama(modelname, args) -> str:  # noqa: C901
                 "Please install the Qualcomm backend follwing https://pytorch.org/executorch/main/build-run-qualcomm.html"
             )
 
-        # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`
-        backend_options = generate_htp_compiler_spec(use_fp16=False)
+        use_fp16 = True
+        skip_node_op_set = {}
+        if args.pt2e_quantize:
+            use_fp16 = False
+            # TODO: fix the lowering error without skipping nodes
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            if quant_dtype == QuantDtype.use_8a8w:
+                raise NotImplementedError("8a8w for llama is still under development")
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            elif quant_dtype == QuantDtype.use_16a16w:
+                raise NotImplementedError("16a16w for llama is still under development")
+            # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+            elif quant_dtype == QuantDtype.use_16a4w:
+                raise NotImplementedError("16a4w for llama is still under development")
         partitioners.append(
             # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`
             QnnPartitioner(
                 # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`
                 generate_qnn_executorch_compiler_spec(
                     # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
                     soc_model=QcomChipset.SM8650,  # default to SM8650
-                    backend_options=backend_options,
+                    # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`.
+                    backend_options=generate_htp_compiler_spec(use_fp16=use_fp16),
                     debug=False,
                     saver=False,
                 ),
                 skip_node_id_set={},
-                skip_node_op_set={},
+                skip_node_op_set=skip_node_op_set,
             )
         )
         # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`
-        _transform(builder_exported_to_edge.export_program())
+        _transform(builder_exported_to_edge.edge_manager.exported_program())
 
     if args.generate_etrecord:
         if not builder_exported_to_edge.edge_manager:

exir/passes/_quant_patterns_and_replacements.py

@@ -179,7 +179,7 @@ def embedding_byte_dtype_out_meta(
 
 quantized_decomposed_lib.define(
     "embedding_4bit.dtype(Tensor weight, Tensor weight_scales, Tensor? weight_zero_points, "
-    "int weight_quant_min, int weight_quant_max, Tensor indices, ScalarType? dtype=None) -> Tensor",
+    "int weight_quant_min, int weight_quant_max, Tensor indices, *, ScalarType? dtype=None) -> Tensor",
 )
 
 quantized_decomposed_lib.define(

extension/module/module.cpp

@@ -51,8 +51,8 @@ Module::Module(
     std::unique_ptr<EventTracer> event_tracer)
     : data_loader_(std::move(data_loader)),
       memory_allocator_(
-          std::move(memory_allocator)
-              ?: std::make_unique<util::MallocMemoryAllocator>()),
+          memory_allocator ? std::move(memory_allocator)
+                           : std::make_unique<util::MallocMemoryAllocator>()),
       event_tracer_(std::move(event_tracer)) {
   runtime_init();
 }