Update to Pytorch 2.3.0

platform/pom.xml

@@ -292,7 +292,7 @@
     <dependency>
       <groupId>org.bytedeco</groupId>
       <artifactId>pytorch-platform</artifactId>
-      <version>2.2.2-${project.version}</version>
+      <version>2.3.0-${project.version}</version>
     </dependency>
     <dependency>
       <groupId>org.bytedeco</groupId>

pytorch/cppbuild.sh

@@ -35,7 +35,7 @@ if [[ $PLATFORM == windows* ]]; then
     export PYTHON_BIN_PATH=$(which python.exe)
 fi
 
-PYTORCH_VERSION=2.2.2
+PYTORCH_VERSION=2.3.0
 
 export PYTORCH_BUILD_VERSION="$PYTORCH_VERSION"
 export PYTORCH_BUILD_NUMBER=1
@@ -53,10 +53,6 @@ git checkout v$PYTORCH_VERSION
 git submodule update --init --recursive
 git submodule foreach --recursive 'git reset --hard'
 
-# Fix version of this submodule to allow compilation on windows.
-# Probably could be removed when we upgrade to 2.3
-(cd third_party/pocketfft; git checkout 9d3ab05a7fffbc71a492bc6a17be034e83e8f0fe)
-
 CPYTHON_HOST_PATH="$INSTALL_PATH/../../../cpython/cppbuild/$PLATFORM/host/"
 CPYTHON_PATH="$INSTALL_PATH/../../../cpython/cppbuild/$PLATFORM/"
 OPENBLAS_PATH="$INSTALL_PATH/../../../openblas/cppbuild/$PLATFORM/"

pytorch/include_list.pl

@@ -52,9 +52,7 @@ sub go {
 
 chdir "cppbuild/linux-x86_64-gpu/pytorch/torch/include";
 
-# Doesn't compile on Windows. Waiting for 2.2.1.
-#go('torch/csrc/api/include/torch/torch.h', 'torch/script.h', 'torch/csrc/inductor/aoti_model_container_runner.h');
-go('torch/csrc/api/include/torch/torch.h', 'torch/script.h');
+go('torch/csrc/api/include/torch/torch.h', 'torch/script.h', 'torch/csrc/inductor/aoti_runner/model_container_runner_cpu.h');
 
 print <<EOF;
 
@@ -64,6 +62,4 @@ sub go {
 // c10/cuda/CUDAGuard.h
 EOF
 
-# Doesn't compile on Windows. Waiting for 2.2.1.
-#go('ATen/cudnn/Descriptors.h', 'ATen/cudnn/Types.h', 'c10/cuda/CUDAGuard.h', '-I/opt/cuda/targets/x86_64-linux/include/', 'torch/csrc/inductor/aoti_model_container_runner_cuda.h');
-go('ATen/cudnn/Descriptors.h', 'ATen/cudnn/Types.h', 'c10/cuda/CUDAGuard.h', '-I/opt/cuda/targets/x86_64-linux/include/');
+go('ATen/cudnn/Descriptors.h', 'ATen/cudnn/Types.h', 'c10/cuda/CUDAGuard.h', '-I/opt/cuda/targets/x86_64-linux/include/', 'torch/csrc/inductor/aoti_runner/model_container_runner_cuda.h');

pytorch/platform/gpu/pom.xml

@@ -12,7 +12,7 @@
 
   <groupId>org.bytedeco</groupId>
   <artifactId>pytorch-platform-gpu</artifactId>
-  <version>2.2.2-${project.parent.version}</version>
+  <version>2.3.0-${project.parent.version}</version>
   <name>JavaCPP Presets Platform GPU for PyTorch</name>
 
   <properties>

pytorch/platform/pom.xml

@@ -12,7 +12,7 @@
 
   <groupId>org.bytedeco</groupId>
   <artifactId>pytorch-platform</artifactId>
-  <version>2.2.2-${project.parent.version}</version>
+  <version>2.3.0-${project.parent.version}</version>
   <name>JavaCPP Presets Platform for PyTorch</name>
 
   <properties>

pytorch/pom.xml

@@ -11,7 +11,7 @@
 
   <groupId>org.bytedeco</groupId>
   <artifactId>pytorch</artifactId>
-  <version>2.2.2-${project.parent.version}</version>
+  <version>2.3.0-${project.parent.version}</version>
   <name>JavaCPP Presets for PyTorch</name>
 
   <dependencies>

pytorch/samples/pom.xml

@@ -12,14 +12,14 @@
         <dependency>
             <groupId>org.bytedeco</groupId>
             <artifactId>pytorch-platform</artifactId>
-            <version>2.2.2-1.5.11-SNAPSHOT</version>
+            <version>2.3.0-1.5.11-SNAPSHOT</version>
         </dependency>
 
         <!-- Additional dependencies required to use CUDA, cuDNN, and NCCL -->
         <dependency>
             <groupId>org.bytedeco</groupId>
             <artifactId>pytorch-platform-gpu</artifactId>
-            <version>2.2.2-1.5.11-SNAPSHOT</version>
+            <version>2.3.0-1.5.11-SNAPSHOT</version>
         </dependency>
 
         <!-- Additional dependencies to use bundled CUDA, cuDNN, and NCCL -->

pytorch/src/gen/java/org/bytedeco/pytorch/AOTIModelContainerRunner.java

@@ -0,0 +1,48 @@
+// Targeted by JavaCPP version 1.5.11-SNAPSHOT: DO NOT EDIT THIS FILE
+
+package org.bytedeco.pytorch;
+
+import org.bytedeco.pytorch.Allocator;
+import org.bytedeco.pytorch.Function;
+import org.bytedeco.pytorch.functions.*;
+import org.bytedeco.pytorch.Module;
+import org.bytedeco.javacpp.annotation.Cast;
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.bytedeco.javacpp.presets.javacpp.*;
+import static org.bytedeco.openblas.global.openblas_nolapack.*;
+import static org.bytedeco.openblas.global.openblas.*;
+
+import static org.bytedeco.pytorch.global.torch.*;
+
+
+@Namespace("torch::inductor") @NoOffset @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
+public class AOTIModelContainerRunner extends Pointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public AOTIModelContainerRunner(Pointer p) { super(p); }
+
+
+
+
+
+
+
+  public native @ByVal TensorVector run(
+        @ByRef TensorVector inputs);
+
+  public native @ByVal ExtraFilesMap getConstantNamesToOriginalFQNs();
+  public native @ByVal StringIntMap getConstantNamesToDtypes();
+  public native void update_inactive_constant_buffer(@Cast("const torch::inductor::TensorConstantMap*") @ByRef HashAliasedIValueMap const_map);
+  public native void update_constant_buffer(
+        @Cast("const torch::inductor::TensorConstantMap*") @ByRef HashAliasedIValueMap const_map,
+        @Cast("bool") boolean use_inactive,
+        @Cast("bool") boolean validate_full_updates);
+  public native void run_const_fold(
+        @Cast("bool") boolean use_inactive);
+  public native void swap_constant_buffer();
+
+  public native @ByVal StringVector get_call_spec();
+}

pytorch/src/gen/java/org/bytedeco/pytorch/AOTIModelContainerRunnerCpu.java

@@ -0,0 +1,48 @@
+// Targeted by JavaCPP version 1.5.11-SNAPSHOT: DO NOT EDIT THIS FILE
+
+package org.bytedeco.pytorch;
+
+import org.bytedeco.pytorch.Allocator;
+import org.bytedeco.pytorch.Function;
+import org.bytedeco.pytorch.functions.*;
+import org.bytedeco.pytorch.Module;
+import org.bytedeco.javacpp.annotation.Cast;
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.bytedeco.javacpp.presets.javacpp.*;
+import static org.bytedeco.openblas.global.openblas_nolapack.*;
+import static org.bytedeco.openblas.global.openblas.*;
+
+import static org.bytedeco.pytorch.global.torch.*;
+
+@Namespace("torch::inductor") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
+public class AOTIModelContainerRunnerCpu extends AOTIModelContainerRunner {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public AOTIModelContainerRunnerCpu(Pointer p) { super(p); }
+
+  public AOTIModelContainerRunnerCpu(
+        @StdString BytePointer model_so_path,
+        @Cast("size_t") long num_models/*=1*/) { super((Pointer)null); allocate(model_so_path, num_models); }
+  private native void allocate(
+        @StdString BytePointer model_so_path,
+        @Cast("size_t") long num_models/*=1*/);
+  public AOTIModelContainerRunnerCpu(
+        @StdString BytePointer model_so_path) { super((Pointer)null); allocate(model_so_path); }
+  private native void allocate(
+        @StdString BytePointer model_so_path);
+  public AOTIModelContainerRunnerCpu(
+        @StdString String model_so_path,
+        @Cast("size_t") long num_models/*=1*/) { super((Pointer)null); allocate(model_so_path, num_models); }
+  private native void allocate(
+        @StdString String model_so_path,
+        @Cast("size_t") long num_models/*=1*/);
+  public AOTIModelContainerRunnerCpu(
+        @StdString String model_so_path) { super((Pointer)null); allocate(model_so_path); }
+  private native void allocate(
+        @StdString String model_so_path);
+
+  public native @ByVal TensorVector run(@ByRef TensorVector inputs);
+}

pytorch/src/gen/java/org/bytedeco/pytorch/AcceleratorHooksInterface.java

@@ -0,0 +1,37 @@
+// Targeted by JavaCPP version 1.5.11-SNAPSHOT: DO NOT EDIT THIS FILE
+
+package org.bytedeco.pytorch;
+
+import org.bytedeco.pytorch.Allocator;
+import org.bytedeco.pytorch.Function;
+import org.bytedeco.pytorch.functions.*;
+import org.bytedeco.pytorch.Module;
+import org.bytedeco.javacpp.annotation.Cast;
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.bytedeco.javacpp.presets.javacpp.*;
+import static org.bytedeco.openblas.global.openblas_nolapack.*;
+import static org.bytedeco.openblas.global.openblas.*;
+
+import static org.bytedeco.pytorch.global.torch.*;
+
+
+// AcceleratorHooksInterface is a shared interface provided by all
+// accelerators to allow generic code.
+// This inferface is hook-based as it corresponds to all the functions
+// that are going to be called in a generic way from the CPU code.
+
+@Namespace("at") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
+public class AcceleratorHooksInterface extends Pointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public AcceleratorHooksInterface(Pointer p) { super(p); }
+
+  // This should never actually be implemented, but it is used to
+  // squelch -Werror=non-virtual-dtor
+
+  // Whether the device at device_index is fully initialized or not.
+  public native @Cast("bool") boolean hasPrimaryContext(@Cast("c10::DeviceIndex") byte device_index);
+}

pytorch/src/gen/java/org/bytedeco/pytorch/Allocator.java

@@ -46,7 +46,22 @@ public class Allocator extends Pointer {
     public Allocator(Pointer p) { super(p); }
 
 
-  public native @Name("allocate") @Cast({"", "c10::DataPtr&&"}) @StdMove DataPtr _allocate(@Cast("size_t") long n);
+  public native @Name("allocate") @StdMove DataPtr _allocate(@Cast("size_t") long n);
+
+  // Clones an allocation that came from this allocator.
+  //
+  // To perform the copy, this function calls `copy_data`, which
+  // must be implemented by derived classes.
+  //
+  // Note that this explicitly ignores any context that may have been
+  // attached to the input data.
+  //
+  // Requires: input data was allocated by the same allocator.
+  public native @StdMove DataPtr clone(@Const Pointer data, @Cast("std::size_t") long n);
+
+  // Checks if DataPtr has a simple context, not wrapped with any out of the
+  // ordinary contexts.
+  public native @Cast("bool") boolean is_simple_data_ptr(@StdMove DataPtr data_ptr);
 
   // If this returns a non nullptr, it means that allocate()
   // is guaranteed to return a unique_ptr with this deleter attached;
@@ -55,4 +70,13 @@ public class Allocator extends Pointer {
   public native PointerConsumer raw_deleter();
   public native Pointer raw_allocate(@Cast("size_t") long n);
   public native void raw_deallocate(Pointer ptr);
+
+  // Copies data from one allocation to another.
+  // Pure virtual, so derived classes must define behavior.
+  // Derived class implementation can simply call `default_copy_data`
+  // to use `std::memcpy`.
+  //
+  // Requires: src and dest were allocated by this allocator
+  // Requires: src and dest both have length >= count
+  public native void copy_data(Pointer dest, @Const Pointer src, @Cast("std::size_t") long count);
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AnyModule.java

@@ -283,6 +283,8 @@ public class AnyModule extends Pointer {
   private native void allocate(@SharedPtr @Cast({"", "std::shared_ptr<torch::nn::MaxUnpool3dImpl>"}) MaxUnpool3dImpl module);
   public AnyModule(FractionalMaxPool3dImpl module) { super((Pointer)null); allocate(module); }
   private native void allocate(@SharedPtr @Cast({"", "std::shared_ptr<torch::nn::FractionalMaxPool3dImpl>"}) FractionalMaxPool3dImpl module);
+  public AnyModule(LPPool3dImpl module) { super((Pointer)null); allocate(module); }
+  private native void allocate(@SharedPtr @Cast({"", "std::shared_ptr<torch::nn::LPPool3dImpl>"}) LPPool3dImpl module);
   public AnyModule(RNNImpl module) { super((Pointer)null); allocate(module); }
   private native void allocate(@SharedPtr @Cast({"", "std::shared_ptr<torch::nn::RNNImpl>"}) RNNImpl module);
   public AnyModule(LSTMImpl module) { super((Pointer)null); allocate(module); }