[Pipeline][GPUToSPIRV][LevelZeroRuntimeWrappers] Use upstream pipeline for VC path. (#1124)

[Pipeline][GPUToSPIRV][LevelZeroRuntimeWrappers] Use upstream pipeline for VC path.

    - Add a one-shot SPIR-V conversion pass (ConvertToSPIRV).
      It combines upstream and downstream patterns.
    - Modify pass pipeline to for host and gpu code lowering.
    - Modify test cases to use upstream level-zero-runtime.

Remove `spirv.target_env` from individual test cases.
Use SPIRVAttatchTarget pass to set `targetEnv`.

More details are available in internal issue: 1205, 1206, 1239.

Author: mshahneo

include/imex/Conversion/ConvertToSPIRV/ConvertToSPIRV.h

@@ -0,0 +1,36 @@
+//===- ConvertToSPIRV.h - Converts everything to SPIR-V dialect - *-C++ -*-===//
+//
+// Copyright 2025 Intel Corporation
+// Part of the IMEX Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef IMEX_CONVERSION_CONVERTTOSPIRV_H
+#define IMEX_CONVERSION_CONVERTTOSPIRV_H
+
+#include "imex/Utils/XeCommon.h"
+#include "mlir/Dialect/SPIRV/Transforms/SPIRVConversion.h"
+#include "mlir/IR/Operation.h"
+#include "mlir/Pass/Pass.h"
+#include "mlir/Transforms/DialectConversion.h"
+
+namespace mlir {
+class ConversionTarget;
+class SPIRVTypeConverter;
+class Pass;
+class Operation;
+class RewritePatternSet;
+template <typename T> class OperationPass;
+} // namespace mlir
+
+namespace imex {
+#define GEN_PASS_DECL_CONVERTTOSPIRV
+#include "imex/Conversion/Passes.h.inc"
+
+std::unique_ptr<::mlir::OperationPass<::mlir::ModuleOp>>
+createConvertToSPIRVPass();
+
+} // namespace imex
+#endif

include/imex/Conversion/Passes.h

@@ -18,6 +18,7 @@
 #include "mlir/Pass/Pass.h"
 
 #include <imex/Conversion/ArithToVC/ArithToVC.h>
+#include <imex/Conversion/ConvertToSPIRV/ConvertToSPIRV.h>
 #include <imex/Conversion/DropRegions/DropRegions.h>
 #include <imex/Conversion/GPUToGPUX/GPUToGPUX.h>
 #include <imex/Conversion/GPUToSPIRV/GPUToSPIRVPass.h>

include/imex/Conversion/Passes.td

@@ -258,6 +258,43 @@ memref, arith and math.
   let dependentDialects = ["::mlir::spirv::SPIRVDialect"];
 }
 
+
+//===----------------------------------------------------------------------===//
+// ConvertToSPIRV
+//===----------------------------------------------------------------------===//
+
+def ConvertToSPIRV : Pass<"imex-convert-to-spirv", "::mlir::ModuleOp"> {
+  let summary = "Convert to SPIR-V dialect by using all the 'to SPIR-V' conversion patterns from all the dialect conversions";
+  let description = [{
+    This is a one-shot pass to convert to SPIR-V dialect by using all the 'to SPIR-V' conversion patterns from all the dialect conversions.
+    It includes the GPU to SPIR-V conversion as well as other dialects like SCF, Math, Arith etc.
+
+  }];
+  let dependentDialects = ["::mlir::spirv::SPIRVDialect"];
+  let constructor = "imex::createConvertToSPIRVPass()";
+  let options = [
+    // arith, cf, func, tensor to SPIR-V options
+    Option<"emulateLT32BitScalarTypes", "emulate-lt-32-bit-scalar-types",
+           "bool", /*default=*/"true",
+           "Emulate narrower scalar types with 32-bit ones if not supported by "
+           "the target">,
+    Option<"emulateUnsupportedFloatTypes", "emulate-unsupported-float-types",
+           "bool", /*default=*/"true",
+           "Emulate unsupported float types by representing them with integer "
+           "types of same bit width">,
+    // gpu, Index to SPIR-V options
+    Option<"use64bitIndex", "use-64bit-index",
+           "bool", /*default=*/"false",
+           "Use 64-bit integers to convert index types">,
+    // memref to SPIR-V options
+    Option<"boolNumBits", "bool-num-bits",
+           "int", /*default=*/"8",
+           "The number of bits to store a boolean value">,
+
+  ];
+}
+
+
 //===----------------------------------------------------------------------===//
 // GPUToGPUX
 //===----------------------------------------------------------------------===//

lib/Conversion/CMakeLists.txt

@@ -1,4 +1,5 @@
 add_subdirectory(ArithToVC)
+add_subdirectory(ConvertToSPIRV)
 add_subdirectory(NDArrayToLinalg)
 add_subdirectory(DropRegions)
 add_subdirectory(RegionParallelLoopToGpu)

lib/Conversion/ConvertToSPIRV/CMakeLists.txt

@@ -0,0 +1,24 @@
+add_imex_conversion_library(IMEXConvertToSPIRV
+  ConvertToSPIRVPass.cpp
+
+  ADDITIONAL_HEADER_DIRS
+  ${MLIR_MAIN_INCLUDE_DIR}/imex/Conversion/ConvertToSPIRV
+
+  DEPENDS
+  IMEXConversionPassIncGen
+
+  LINK_LIBS PUBLIC
+  MLIRArithToSPIRV
+  MLIRControlFlowToSPIRV
+  MLIRFuncToSPIRV
+  MLIRGPUDialect
+  MLIRGPUToSPIRV
+  MLIRIR
+  MLIRMathToSPIRV
+  MLIRPass
+  MLIRSCFToSPIRV
+  MLIRSPIRVDialect
+  MLIRSPIRVConversion
+  MLIRSupport
+  MLIRTransforms
+  )