[Mlir-commits] [mlir] [mlir] Add `mlir_arm_runner_utils` library for use in integration tests (PR #78583)

[Andrzej Warzyński]

================
@@ -0,0 +1,34 @@
+// DEFINE: %{entry_point} = main
+// DEFINE: %{compile} = mlir-opt %s -convert-arm-sme-to-llvm \
+// DEFINE:   -cse -canonicalize -test-lower-to-llvm
+// DEFINE: %{run} = %mcr_aarch64_cmd \
+// DEFINE:  -march=aarch64 -mattr=+sve,+sme \
+// DEFINE:  -e %{entry_point} -entry-point-result=void \
+// DEFINE:  -shared-libs=%mlir_runner_utils,%mlir_c_runner_utils,%mlir_arm_runner_utils
+
+// RUN: %{compile} | %{run} | FileCheck %s
+
+func.func @checkVScale() {
+  %vscale = vector.vscale
+  vector.print str "vscale"
+  vector.print %vscale : index
+  return
+}
+
+func.func @main() {
+  //      CHECK: vscale
+  // CHECK-NEXT: 1
+  %c128 = arith.constant 128 : i32
+  func.call @setArmVLBits(%c128) : (i32) -> ()
+  func.call @checkVScale() : () -> ()
+
+  //      CHECK: vscale
+  // CHECK-NEXT: 2
+  %c256 = arith.constant 256 : i32
----------------
banach-space wrote:

[nit] I'd try the architecture max - 2048.

https://github.com/llvm/llvm-project/pull/78583