[Mlir-commits] [mlir] [mlir][x86vector] AVX Convert/Broadcast F16 to F32 instructions (PR #137917)

[Adam Siemieniuk]

================
@@ -527,4 +527,126 @@ def BcstBF16ToPackedF32Op : AVX_Op<"bcst.bf16_to_f32.packed", [MemoryEffects<[Me
 
 }
 
+//----------------------------------------------------------------------------//
+// AVX: Convert packed F16 even-indexed/odd-indexed elements into packed F32
+//----------------------------------------------------------------------------//
+
+def CvtPackedEvenIndexedF16ToF32Op : AVX_Op<"cvt.packed.even.indexed.f16_to_f32", [MemoryEffects<[MemRead]>, 
+  DeclareOpInterfaceMethods<OneToOneIntrinsicOpInterface>]> {
+  let summary = "AVX: Convert packed F16 even-indexed elements into packed F32 Data.";
+  let description = [{
+
+    #### From the Intel Intrinsics Guide:
+
+    Convert packed F16 (16-bit) floating-point even-indexed elements stored at
+    memory locations starting at location `__A` to packed single-precision
+    (32-bit) floating-point elements, and store the results in `dst`.
+
+    Example:
+    ```mlir
+    %dst = x86vector.avx.cvt.packed.even.indexed.f16_to_f32 %a : memref<16xbf16> -> vector<8xf32>
----------------
adam-smnk wrote:

copy pasta wrong input type

https://github.com/llvm/llvm-project/pull/137917