[SYCL][ESIMD][EMU] Marking ESIMD kernels for esimd_emulator backend

SYCL/ESIMD/BitonicSortK.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_oword_ld_unaligned
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/accessor.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_oword_ld_unaligned
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl -D_CRT_SECURE_NO_WARNINGS=1 %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/accessor_gather_scatter.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated memory intrinsic
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/accessor_load_store.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated memory intrinsic
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/ballot.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated memory intrinsic
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/api/bin_and_cmp_ops_heavy.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/api/esimd_bit_ops.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'single_task()' method
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -fsycl-device-code-split=per_kernel -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/replicate_smoke.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/saturation_smoke.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/simd_any_all.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -fsycl-device-code-split=per_kernel -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/simd_binop_integer_promotion.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated memory intrinsic
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/simd_copy_to_from.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/api/simd_memory_access.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/simd_negation_operator.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_oword_ld_unaligned
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/simd_subscript_operator.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated memory intrinsic
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/simd_view_copy_move_assign.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/api/simd_view_negation_operator.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_oword_ld_unaligned
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/simd_view_subscript_operator.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_oword_ld_unaligned
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/slm_gather_scatter.cpp

@@ -1,5 +1,7 @@
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_scatter_scaled
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/slm_gather_scatter_heavy.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated memory intrinsic
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/slm_gather_scatter_rgba.cpp

@@ -1,5 +1,7 @@
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_scatter_scaled
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //

SYCL/ESIMD/api/unary_ops_heavy.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/esimd_check_vc_codegen.cpp

@@ -0,0 +1,104 @@
+//==-------- esimd_check_vc_codegen.cpp - DPC++ ESIMD on-device test -------==//
+//
+// Part of the LLVM 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
+//
+//===----------------------------------------------------------------------===//
+// REQUIRES: gpu
+// UNSUPPORTED: cuda || hip
+// esimd_emulator does not support online-compiler that invokes 'piProgramBuild'
+// UNSUPPORTED: esimd_emulator
+// RUN: %clangxx -fsycl %s -o %t.out
+// RUN: env SYCL_PI_TRACE=-1 %GPU_RUN_PLACEHOLDER %t.out 2>&1 %GPU_CHECK_PLACEHOLDER
+
+#include "esimd_test_utils.hpp"
+
+#include <CL/sycl.hpp>
+#include <iostream>
+#include <sycl/ext/intel/experimental/esimd.hpp>
+
+using namespace cl::sycl;
+
+int main(void) {
+  constexpr unsigned Size = 1024 * 128;
+  constexpr unsigned VL = 16;
+
+  float *A = new float[Size];
+  float *B = new float[Size];
+  float *C = new float[Size];
+
+  for (unsigned i = 0; i < Size; ++i) {
+    A[i] = B[i] = i;
+    C[i] = 0.0f;
+  }
+
+  try {
+    buffer<float, 1> bufa(A, range<1>(Size));
+    buffer<float, 1> bufb(B, range<1>(Size));
+    buffer<float, 1> bufc(C, range<1>(Size));
+
+    // We need that many workgroups
+    range<1> GlobalRange{Size / VL};
+
+    // We need that many threads in each group
+    range<1> LocalRange{1};
+
+    queue q(esimd_test::ESIMDSelector{}, esimd_test::createExceptionHandler());
+
+    auto dev = q.get_device();
+    std::cout << "Running on " << dev.get_info<info::device::name>() << "\n";
+
+    auto e = q.submit([&](handler &cgh) {
+      auto PA = bufa.get_access<access::mode::read>(cgh);
+      auto PB = bufb.get_access<access::mode::read>(cgh);
+      auto PC = bufc.get_access<access::mode::write>(cgh);
+      cgh.parallel_for<class Test>(
+          GlobalRange * LocalRange, [=](id<1> i) SYCL_ESIMD_KERNEL {
+            using namespace sycl::ext::intel::experimental::esimd;
+            unsigned int offset = i * VL * sizeof(float);
+            simd<float, VL> va;
+            va.copy_from(PA, offset);
+            simd<float, VL> vb;
+            vb.copy_from(PB, offset);
+            simd<float, VL> vc = va + vb;
+            vc.copy_to(PC, offset);
+          });
+    });
+    e.wait();
+  } catch (sycl::exception const &e) {
+    std::cout << "SYCL exception caught: " << e.what() << '\n';
+
+    delete[] A;
+    delete[] B;
+    delete[] C;
+    return 1;
+  }
+
+  int err_cnt = 0;
+
+  for (unsigned i = 0; i < Size; ++i) {
+    if (A[i] + B[i] != C[i]) {
+      if (++err_cnt < 10) {
+        std::cout << "failed at index " << i << ", " << C[i] << " != " << A[i]
+                  << " + " << B[i] << "\n";
+      }
+    }
+  }
+  if (err_cnt > 0) {
+    std::cout << "  pass rate: "
+              << ((float)(Size - err_cnt) / (float)Size) * 100.0f << "% ("
+              << (Size - err_cnt) << "/" << Size << ")\n";
+  }
+
+  delete[] A;
+  delete[] B;
+  delete[] C;
+
+  std::cout << (err_cnt > 0 ? "FAILED\n" : "Passed\n");
+  return err_cnt > 0 ? 1 : 0;
+}
+
+// CHECK: ---> piProgramBuild(
+// CHECK: <const char *>: {{.*}}-vc-codegen
+// CHECK: ) ---> pi_result : PI_SUCCESS

SYCL/ESIMD/ext_math.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented 'half' type
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/fp_in_phi.cpp

@@ -14,6 +14,8 @@
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // UNSUPPORTED: cuda || hip
 // UNSUPPORTED: ze_debug-1,ze_debug4
+// TODO: esimd_emulator fails due to unimplemented __esimd_scatter_scaled
+// XFAIL: esimd_emulator
 //
 // The test checks that ESIMD kernels correctly handle function pointers as
 // arguments of LLVM's PHI function.

SYCL/ESIMD/histogram.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated __esimd_media_ld
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/ESIMD/histogram_256_slm.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_scatter_scaled
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 

SYCL/ESIMD/histogram_256_slm_spec_2020.cpp

@@ -1,6 +1,8 @@
 // TODO enable on Windows
 // REQUIRES: linux && gpu
 // UNSUPPORTED: cuda || hip
+// TODO online_compiler check fails for esimd_emulator
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out 16
 

SYCL/ESIMD/histogram_2d.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated __esimd_media_ld
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/ESIMD/histogram_raw_send.cpp

@@ -9,6 +9,8 @@
 // REQUIRES: gpu
 // UNSUPPORTED: gpu-intel-dg1,cuda,hip
 // UNSUPPORTED: ze_debug-1,ze_debug4
+// TODO: esimd_emulator fails due to outdated __esimd_media_ld
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/ESIMD/linear/linear.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated __esimd_media_ld
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -I%S/.. -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out %S/linear_in.bmp %S/linear_gold_hw.bmp
 // RUN: %GPU_RUN_PLACEHOLDER %t.out %S/linear_in.bmp %S/linear_gold_hw.bmp

SYCL/ESIMD/mandelbrot/mandelbrot.cpp

@@ -8,6 +8,8 @@
 
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to outdated __esimd_media_st
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -I%S/.. -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out %T/output.ppm %S/golden_hw.ppm
 

SYCL/ESIMD/mandelbrot/mandelbrot_spec.cpp

@@ -9,6 +9,8 @@
 // TODO enable on Windows
 // REQUIRES: linux && gpu
 // UNSUPPORTED: cuda || hip
+// TODO online_compiler check fails for esimd_emulator
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -I%S/.. -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out %T/output_spec.ppm %S/golden_hw.ppm 512 -2.09798 -1.19798 0.004 4.0
 

SYCL/ESIMD/matrix_transpose.cpp

@@ -7,6 +7,8 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// TODO: esimd_emulator fails due to unimplemented __esimd_oword_ld_unaligned
+// XFAIL: esimd_emulator
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out