Fix as_strided/slice symint (#1401)

* Fix as_strided symint

* Re-enable LTC tests

* Re-enable LTC

* Add hardtanh shape inference function

* Fix slice symint

build_tools/autogen_ltc_backend.yaml

@@ -78,8 +78,13 @@ symint:
 - expand
 - expand_copy
 - narrow_copy
+- slice_backward
+- slice_copy.Tensor
+- slice_scatter
 - view
 - view_copy
+- as_strided_copy
+- as_strided_scatter
 
 
 additional_ops:

build_tools/python_deploy/build_linux_packages.sh

@@ -177,7 +177,7 @@ function build_in_tree() {
       -DLLVM_EXTERNAL_TORCH_MLIR_DIALECTS_SOURCE_DIR="/main_checkout/torch-mlir/externals/llvm-external-projects/torch-mlir-dialects" \
       -DLLVM_TARGETS_TO_BUILD=host \
       -DMLIR_ENABLE_BINDINGS_PYTHON=ON \
-      -DTORCH_MLIR_ENABLE_LTC=OFF \
+      -DTORCH_MLIR_ENABLE_LTC=ON \
       -DTORCH_MLIR_USE_INSTALLED_PYTORCH="$torch_from_src" \
       -DPython3_EXECUTABLE="$(which python3)" \
       /main_checkout/torch-mlir/externals/llvm-project/llvm
@@ -240,9 +240,8 @@ function test_in_tree() {
   # - AvgPool2dFloatModule_basic,AvgPool2dCeilModeTrueModule_basic: https://github.com/llvm/torch-mlir/issues/1361
   python -m e2e_testing.main --config=tosa -v --crashing_tests_to_not_attempt_to_run_and_a_bug_is_filed AvgPool2dFloatModule_basic AvgPool2dCeilModeTrueModule_basic
 
-  # Temporarily disabled in top of main (https://github.com/llvm/torch-mlir/pull/1292)
-  #echo ":::: Run Lazy Tensor Core e2e integration tests"
-  #python -m e2e_testing.torchscript.main --config=lazy_tensor_core -v
+  echo ":::: Run Lazy Tensor Core e2e integration tests"
+  python -m e2e_testing.main --config=lazy_tensor_core -v
 }
 
 function setup_venv() {
@@ -291,7 +290,7 @@ function build_out_of_tree() {
       -DLLVM_DIR="/main_checkout/torch-mlir/llvm-build/lib/cmake/llvm/" \
       -DMLIR_DIR="/main_checkout/torch-mlir/llvm-build/lib/cmake/mlir/" \
       -DMLIR_ENABLE_BINDINGS_PYTHON=OFF \
-      -DTORCH_MLIR_ENABLE_LTC=OFF \
+      -DTORCH_MLIR_ENABLE_LTC=ON \
       -DTORCH_MLIR_USE_INSTALLED_PYTORCH="$torch_from_src" \
       -DPython3_EXECUTABLE="$(which python3)" \
       /main_checkout/torch-mlir

e2e_testing/xfail_sets.py

@@ -450,6 +450,7 @@ LTC_XFAIL_SET = {
     "_Convolution2DTF32Module_basic",
     "_ConvolutionDeprecated2DAllFalseModule_basic",
     "_ConvolutionDeprecated2DBenchmarkModule_basic",
+    "_ConvolutionDeprecated2DCudnnModule_basic",
     "_ConvolutionDeprecated2DDeterministicModule_basic",
     "AdaptiveAvgPool2dNonUnitOutputSizeDynamicModule_basic",
     "AdaptiveAvgPool2dNonUnitOutputSizeStaticModule_basic",
@@ -496,8 +497,6 @@ LTC_XFAIL_SET = {
     "GtFloatIntModule_basic",
     "GtIntModule_basic",
     "HBC_basic",
-    "HardTanhIntModule_basic",
-    "HardTanhModule_basic",
     "IndexPut1DFloatAccumulateModule_basic",
     "IndexPut1DFloatNonAccumulateModule_basic",
     "IndexPut1DIntAccumulateModule_basic",
@@ -545,6 +544,7 @@ LTC_XFAIL_SET = {
     "IndexTensorHackedTwinModule_basic",
     "IndexTensorHackedTwinModule3dInput_basic",
     "IndexTensorHackedTwinMultiInputNonContiguousMultipleStaticDims_basic",
+    "LiftFreshCopyModule_basic",
     "Matmul_dot",
     "Matmul_matvec",
     "MulIntModule_basic",

python/torch_mlir/csrc/base_lazy_backend/mlir_native_functions.cpp

@@ -419,9 +419,13 @@ at::Tensor LazyNativeFunctions::select_backward(
   return at::functionalization::functionalize_aten_op<ATEN_OP(
       select_backward)>::call(grad_output, input_sizes, dim, index);
 }
-at::Tensor LazyNativeFunctions::slice_backward(
-    const at::Tensor& grad_output, at::IntArrayRef input_sizes, int64_t dim,
-    int64_t start, int64_t end, int64_t step) {
+at::Tensor LazyNativeFunctions::slice_backward_symint(
+    const at::Tensor& grad_output,
+    at::SymIntArrayRef input_sizes,
+    int64_t dim,
+    c10::SymInt start,
+    c10::SymInt end,
+    c10::SymInt step) {
   return at::functionalization::functionalize_aten_op<ATEN_OP(
       slice_backward)>::call(grad_output, input_sizes, dim, start, end, step);
 }

python/torch_mlir/csrc/base_lazy_backend/shape_inference.cpp

@@ -36,5 +36,12 @@ std::vector<torch::lazy::Shape> compute_shape_var(
   return {Shape(self.scalar_type(), {})};
 }
 
+std::vector<torch::lazy::Shape> compute_shape_hardtanh(
+    const at::Tensor& self, const at::Scalar& min_val, const at::Scalar& max_val
+) {
+  return {Shape(self.scalar_type(), self.sizes().vec())};
+}
+
+
 } // namespace lazy
 } // namespace torch

setup.py

@@ -46,7 +46,7 @@ import torch
 PACKAGE_VERSION = os.environ.get("TORCH_MLIR_PYTHON_PACKAGE_VERSION") or "0.0.1"
 
 # If true, enable LTC build by default
-TORCH_MLIR_ENABLE_LTC_DEFAULT = False
+TORCH_MLIR_ENABLE_LTC_DEFAULT = True
 
 # Build phase discovery is unreliable. Just tell it what phases to run.
 class CustomBuild(_build):