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
@@ -2486,6 +2486,52 @@
   }
 };
 
+class ScatterConverter : public OpConversionPattern<tosa::ScatterOp> {
+public:
+  using OpConversionPattern<tosa::ScatterOp>::OpConversionPattern;
+  LogicalResult
+  matchAndRewrite(tosa::ScatterOp op, ArrayRef<Value> args,
+                  ConversionPatternRewriter &rewriter) const final {
+    auto initial = op.values_in();
+    auto indices = op.indices();
+    auto values = op.input();
+
+    ShapedType initialTy = initial.getType().cast<ShapedType>();
+
+    AffineExpr m, n, o, p;
+    bindDims(rewriter.getContext(), m, n, o, p);
+    SmallVector<AffineMap> affineMaps = AffineMap::inferFromExprList({
+        ArrayRef<AffineExpr>({m, p}),
+        ArrayRef<AffineExpr>({m, p, o}),
+        ArrayRef<AffineExpr>({m, n, o}),
+    });
+
+    SmallVector<StringRef, 4> iteratorTypes = {
+        getParallelIteratorTypeName(), getParallelIteratorTypeName(),
+        getParallelIteratorTypeName(), getReductionIteratorTypeName()};
+
+    auto loc = op.getLoc();
+    auto genericOp = rewriter.create<linalg::GenericOp>(
+        loc, ArrayRef<Type>({initialTy}), ValueRange{indices, values},
+        ValueRange{initial}, affineMaps, iteratorTypes,
+        [&](OpBuilder &b, Location loc, ValueRange args) {
+          auto thisIndex = b.create<linalg::IndexOp>(loc, 1);
+          auto indicesIndex =
+              b.create<IndexCastOp>(loc, b.getIndexType(), args[0]);
+
+          auto eq = rewriter.create<mlir::CmpIOp>(loc, CmpIPredicate::eq,
+                                                  thisIndex, indicesIndex);
+
+          auto select =
+              rewriter.create<mlir::SelectOp>(loc, eq, args[1], args[2]);
+          b.create<linalg::YieldOp>(loc, select.result());
+        });
+
+    rewriter.replaceOp(op, genericOp.getResult(0));
+    return success();
+  }
+};
+
 // Lowerings the TableOp to a series of gathers and numerica operations. This
 // includes interpolation between the high/low values. For the I8 varient, this
 // simplifies to a single gather operation.
@@ -2934,6 +2980,7 @@
       RescaleConverter,
       ResizeConverter,
       ReverseConverter,
+      ScatterConverter,
       TableConverter,
       TileConverter,
       TransposeConverter,
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
@@ -1165,6 +1165,20 @@
 
 // -----
 
+// CHECK-LABEL: @scatter_float
+func @scatter_float(%arg0: tensor<2x8x16xf32>, %arg1: tensor<2x3xi32>, %arg2: tensor<2x3x16xf32>) -> (tensor<2x8x16xf32>) {
+  // CHECK: %0 = linalg.generic {indexing_maps = [#map0, #map1, #map2], iterator_types = ["parallel", "parallel", "parallel", "reduction"]} ins(%arg1, %arg2 : tensor<2x3xi32>, tensor<2x3x16xf32>) outs(%arg0 : tensor<2x8x16xf32>)
+  // CHECK:   %[[VAL1:.+]] = linalg.index 1
+  // CHECK:   %[[VAL2:.+]] = index_cast %arg3
+  // CHECK:   %[[VAL3:.+]] = cmpi eq, %[[VAL1]], %[[VAL2]]
+  // CHECK:   %[[VAL4:.+]] = select %[[VAL3]], %arg4, %arg5
+  // CHECK:   linalg.yield %[[VAL4]]
+  %result = "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<2x8x16xf32>, tensor<2x3xi32>, tensor<2x3x16xf32>) -> tensor<2x8x16xf32>
+  return %result : tensor<2x8x16xf32>
+}
+
+// -----
+
 // CHECK-LABEL: @table8
 func @table8(%arg0: tensor<6xi8>, %arg1: tensor<512xi8>) -> () {
   // CHECK: %[[INIT:.+]] = linalg.init_tensor [6]