diff --git a/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp b/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp
--- a/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp
+++ b/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp
@@ -566,6 +566,19 @@
       return rewriter.create<mlir::SIToFPOp>(loc, resultTypes, args,
                                              mlir::None);
 
+    // Unsigned integers need an unrealized cast so that they can be passed
+    // to UIToFP.
+    if (srcTy.isUnsignedInteger() && dstTy.isa<FloatType>()) {
+      auto unrealizedCast =
+          rewriter
+              .create<UnrealizedConversionCastOp>(
+                  loc, rewriter.getIntegerType(srcTy.getIntOrFloatBitWidth()),
+                  args[0])
+              .getResult(0);
+      return rewriter.create<mlir::UIToFPOp>(loc, resultTypes[0],
+                                             unrealizedCast);
+    }
+
     // Casting to boolean, floats need to only be checked as not-equal to zero.
     if (srcTy.isa<FloatType>() && dstTy.isInteger(1)) {
       Value zero =
diff --git a/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir b/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
--- a/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
+++ b/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
@@ -432,6 +432,18 @@
 
 // -----
 
+// CHECK-LABEL: @test_simple_ui8
+func @test_simple_ui8(%arg0: tensor<1xui8>) -> () {
+
+  // CHECK: linalg.generic
+  // CHECK: uitofp
+  %0 = "tosa.cast"(%arg0) : (tensor<1xui8>) -> tensor<1xf32>
+
+  return
+}
+
+// -----
+
 // CHECK-LABEL: @test_i8
 func @test_i8(%arg0: tensor<1xi8>) -> () {
   // CHECK: linalg.generic