diff --git a/Android.bp b/Android.bp
index 871dacf..b8a4fcd 100644
--- a/Android.bp
+++ b/Android.bp
@@ -647,6 +647,7 @@
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwCast.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwDirectConv2d.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwElementwiseBinary.cpp",
+        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwResize.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwStore.cpp",
         "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentActivation.cpp",
diff --git a/arm_compute/dynamic_fusion/sketch/gpu/operators/GpuPool2d.h b/arm_compute/dynamic_fusion/sketch/gpu/operators/GpuPool2d.h
index 6e1bcdb..65a092c 100644
--- a/arm_compute/dynamic_fusion/sketch/gpu/operators/GpuPool2d.h
+++ b/arm_compute/dynamic_fusion/sketch/gpu/operators/GpuPool2d.h
@@ -21,8 +21,8 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef ARM_COMPUTE_DYNAMIC_FUSION_SKETCH_GPU_OPERATORS_GPUPOOL2D
-#define ARM_COMPUTE_DYNAMIC_FUSION_SKETCH_GPU_OPERATORS_GPUPOOL2D
+#ifndef ACL_ARM_COMPUTE_DYNAMIC_FUSION_SKETCH_GPU_OPERATORS_GPUPOOL2D_H
+#define ACL_ARM_COMPUTE_DYNAMIC_FUSION_SKETCH_GPU_OPERATORS_GPUPOOL2D_H
 
 #include "arm_compute/core/Error.h"
 #include "arm_compute/dynamic_fusion/sketch/attributes/Pool2dAttributes.h"
@@ -83,27 +83,22 @@
      *
      * @param[in,out] sketch     Workload sketch into which the operator will be fused
      * @param[in]     src        Source tensor
-     * @param[out]    dst        Destination tensor
      * @param[in]     attributes Operator attributes
      * @param[in]     settings   Operator settings
      */
-    static void create_op(GpuWorkloadSketch &sketch,
-                          ITensorInfo       *src,
-                          ITensorInfo       *dst,
-                          const Attributes &attributes,
-                          const Settings    &settings);
+    static ITensorInfo *create_op(GpuWorkloadSketch &sketch,
+                                  ITensorInfo       *src,
+                                  const Attributes  &attributes,
+                                  const Settings    &settings);
     /** Check if the operator configuration is supported, irrespective of fusion
      *
-     * @param[in]  context    Workload context within which the operator is running
-     * @param[in]  src        Left hand side tensor info. Data types supported: F16/F32.
-     * @param[out] dst        Destination tensor info. Data types supported: F16/F32.
-     *                        If an uninitialized ITensorInfo is passed in, it will be auto-initialized
-     * @param[in]  attributes Operator attributes
-     * @param[in]  settings   Operator settings
+     * @param[in] context    Workload context within which the operator is running
+     * @param[in] src        Left hand side tensor info. Data types supported: F16/F32.
+     * @param[in] attributes Operator attributes
+     * @param[in] settings   Operator settings
      */
     static Status is_supported_op(const GpuWorkloadContext &context,
                                   const ITensorInfo        *src,
-                                  const ITensorInfo        *dst,
                                   const Attributes         &attributes,
                                   const Settings           &settings);
     /** Validate the operator and check if it can be fused into the workload sketch.
@@ -111,11 +106,10 @@
      */
     static Status validate_op(const GpuWorkloadSketch &sketch,
                               const ITensorInfo       *src,
-                              const ITensorInfo       *dst,
                               const Attributes        &attributes,
                               const Settings          &settings);
 };
 } // namespace dynamic_fusion
 } // namespace experimental
 } // namespace arm_compute
-#endif /* ARM_COMPUTE_DYNAMIC_FUSION_SKETCH_GPU_OPERATORS_GPUPOOL2D */
+#endif // ACL_ARM_COMPUTE_DYNAMIC_FUSION_SKETCH_GPU_OPERATORS_GPUPOOL2D_H
diff --git a/compute_kernel_writer/prototype/include/ckw/types/Operators.h b/compute_kernel_writer/prototype/include/ckw/types/Operators.h
index 14a88c9..4324117 100644
--- a/compute_kernel_writer/prototype/include/ckw/types/Operators.h
+++ b/compute_kernel_writer/prototype/include/ckw/types/Operators.h
@@ -34,6 +34,7 @@
 {
     LogicalNot = 0x0000, // !
     BitwiseNot = 0x0001, // ~
+    Negate     = 0x0002, // -
 };
 
 /* Binary operations
diff --git a/compute_kernel_writer/prototype/src/Prototype.h b/compute_kernel_writer/prototype/src/Prototype.h
index 88d6e89..eb9d719 100644
--- a/compute_kernel_writer/prototype/src/Prototype.h
+++ b/compute_kernel_writer/prototype/src/Prototype.h
@@ -1583,6 +1583,8 @@
             return "!";
         case UnaryOp::BitwiseNot:
             return "~";
+        case UnaryOp::Negate:
+            return "-";
         default:
             assert(false);
             return "";
diff --git a/compute_kernel_writer/prototype/src/TileOperand.cpp b/compute_kernel_writer/prototype/src/TileOperand.cpp
index bf6a15b..0eb2ca6 100644
--- a/compute_kernel_writer/prototype/src/TileOperand.cpp
+++ b/compute_kernel_writer/prototype/src/TileOperand.cpp
@@ -85,6 +85,9 @@
                 case DataType::Fp32:
                     return prototype::Operand(_value[0][0], prototype::OperandType::ScalarFp32);
 
+                case DataType::Fp16:
+                    return prototype::Operand(_value[0][0], prototype::OperandType::ScalarFp16);
+
                 default:
                     CKW_ASSERT(false);
             }
diff --git a/docs/user_guide/release_version_and_change_log.dox b/docs/user_guide/release_version_and_change_log.dox
index 3cb1ac4..04ee10b 100644
--- a/docs/user_guide/release_version_and_change_log.dox
+++ b/docs/user_guide/release_version_and_change_log.dox
@@ -50,6 +50,8 @@
  - Update OpenCL™ API headers to v2023.04.17.
  - Performance optimizations:
    - Optimize @ref cpu::CpuReshape
+ - Port the following kernels in the experimental Dynamic Fusion interface to use the new Compute Kernel Writer interface with support for FP16/FP32 only:
+   - @ref experimental::dynamic_fusion::GpuCkwPool2d
 
 v23.08 Public major release
  - Deprecate the legacy 'libarm_compute_core' library. This library is an artifact of Compute Library's legacy library architecture and no longer serves any purpose.
diff --git a/filelist.json b/filelist.json
index 22d5cba..668feb9 100644
--- a/filelist.json
+++ b/filelist.json
@@ -2353,6 +2353,7 @@
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwCast.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwDirectConv2d.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwElementwiseBinary.cpp",
+        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwResize.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwStore.cpp",
         "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwKernelWriter.cpp",
diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.cpp b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.cpp
new file mode 100644
index 0000000..9c9a298
--- /dev/null
+++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.cpp
@@ -0,0 +1,345 @@
+/*
+ * Copyright (c) 2023 Arm Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.h"
+
+#include "arm_compute/core/Error.h"
+#include "arm_compute/core/Validate.h"
+#include "arm_compute/core/utils/helpers/AdjustVecSize.h"
+#include "ckw/TensorTileSampler.h"
+#include "src/core/helpers/WindowHelpers.h"
+#include "src/dynamic_fusion/sketch/gpu/GpuKernelArgument.h"
+#include "src/dynamic_fusion/sketch/gpu/GpuKernelComponentGroup.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwKernelWriter.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwScopedKernelWriter.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwVariableTable.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/WriterHelper.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.h"
+
+using namespace ckw;
+
+namespace arm_compute
+{
+namespace experimental
+{
+namespace dynamic_fusion
+{
+GpuCkwPool2d::GpuCkwPool2d(ComponentId                      id,
+                           const ArgumentPack<ITensorInfo> &tensors,
+                           const Attributes                &attributes,
+                           const Settings                  &settings)
+    : IGpuCkwComponentDriver{ id, tensors },
+      _src{},
+      _dst{},
+      _attributes{ attributes },
+      _settings{ settings }
+
+{
+    _src = this->tensors().get_const_tensor(TensorType::ACL_SRC_0);
+    _dst = this->tensors().get_const_tensor(TensorType::ACL_DST_0);
+    ARM_COMPUTE_ERROR_ON_NULLPTR(_src, _dst);
+}
+
+void GpuCkwPool2d::write_component_code(const ComponentGroup &comp_group, GpuCkwVariableTable &vtable, GpuCkwScopedKernelWriter writer) const
+{
+    const auto         root_window = comp_group.get_root_component()->ckw_component_driver()->get_window();
+    const unsigned int n0          = root_window.x().step();
+    const unsigned int m0          = root_window.y().step();
+
+    GpuCkwComponentArgument *src = vtable.declare_variable(comp_group, writer, _src, TensorStorageType::ClBufferUint8Ptr, "src");
+    GpuCkwComponentArgument *dst = vtable.declare_variable(comp_group, writer, _dst, TensorStorageType::ClBufferUint8Ptr, "dst");
+
+    TileOperand &gid_0 = writer->declare_tile("gid_0", ckw::DataType::Int32);
+    TileOperand &gid_1 = writer->declare_tile("gid_1", ckw::DataType::Int32);
+    TileOperand &gid_2 = writer->declare_tile("gid_2", ckw::DataType::Int32);
+
+    writer->op_get_global_id(gid_0, 0);
+    writer->op_get_global_id(gid_1, 1);
+    writer->op_get_global_id(gid_2, 2);
+
+    // Data Layout is NHWC
+    constexpr int width_idx  = 1;
+    constexpr int height_idx = 2;
+
+    const int32_t pool_size_x   = static_cast<int32_t>(_attributes.pool_size().x());
+    const int32_t pool_size_y   = static_cast<int32_t>(_attributes.pool_size().y());
+    const int32_t pad_x         = static_cast<int32_t>(_attributes.pad().left);
+    const int32_t pad_y         = static_cast<int32_t>(_attributes.pad().top);
+    const int32_t src_width     = static_cast<int32_t>(_src->dimension(width_idx));
+    const int32_t src_height    = static_cast<int32_t>(_src->dimension(height_idx));
+    const auto    src_data_type = _src->data_type();
+
+    // Check if this is global pooling path
+    const bool is_global_pooling = (pool_size_x == src_width) && (pool_size_y == src_height) && (pad_x == 0) && (pad_y == 0);
+    // Check if this a case of FP_MIXED_PRECISION
+    const bool use_fp_mixed_precision = (src_data_type == DataType::F16) && _settings.mixed_precision() && _attributes.pool_type() != PoolingType::MAX;
+    const auto acc_data_type          = (use_fp_mixed_precision) ? (DataType::F32) : (src_data_type);
+
+    TileOperand       &const_0            = writer->declare_tile("0", 0);
+    const TileOperand &const_1            = writer->declare_tile("1", 1);
+    const TileOperand &const_lowest_value = writer->declare_tile("LOWEST_VALUE", std::numeric_limits<float>::lowest());
+    const TileOperand &pool_size_x_tile   = writer->declare_tile("POOL_SIZE_X", pool_size_x);
+    const TileOperand &pool_size_y_tile   = writer->declare_tile("POOL_SIZE_Y", pool_size_y);
+    const TileOperand &stride_x_tile      = writer->declare_tile("STRIDE_X", static_cast<int32_t>(_attributes.stride().x()));
+    const TileOperand &stride_y_tile      = writer->declare_tile("STRIDE_Y", static_cast<int32_t>(_attributes.stride().y()));
+    const TileOperand &pad_x_tile         = writer->declare_tile("PAD_X", pad_x);
+    const TileOperand &pad_y_tile         = writer->declare_tile("PAD_Y", pad_y);
+    const TileOperand &dst_height_tile    = writer->declare_tile("DST_HEIGHT", static_cast<int32_t>(_dst->dimension(height_idx)));
+    const TileOperand &src_height_tile    = writer->declare_tile("SRC_HEIGHT", src_height);
+    const TileOperand &src_width_tile     = writer->declare_tile("SRC_WIDTH", src_width);
+
+    TileOperand &idx_out_n = writer->declare_tile("idx_out_n", ckw::DataType::Int32);
+    TileOperand &idx_out_h = writer->declare_tile("idx_out_h", ckw::DataType::Int32);
+    TileOperand &idx_out_w = writer->declare_tile("idx_out_w", ckw::DataType::Int32);
+    TileOperand &idx_out_c = writer->declare_tile("idx_out_c", ckw::DataType::Int32);
+
+    const int32_t dst_partial_n0_v = _dst->tensor_shape()[0] % n0;
+
+    get_coord(writer, idx_out_c, gid_0, n0, dst_partial_n0_v, "dst_x_", const_0);
+    get_coord(writer, idx_out_w, gid_1, 1, 0, "dst_y_", const_0);
+
+    writer->op_binary_expression(idx_out_h, gid_2, BinaryOp::Mod, dst_height_tile); // gid_2 % h
+    writer->op_binary_expression(idx_out_n, gid_2, BinaryOp::Div, dst_height_tile); // gid_2 / h
+
+    TensorTileSampler src_sampler;
+    src_sampler.width(n0);
+    src_sampler.height(m0);
+    src_sampler.format(TensorSamplerFormat::C_W_H);
+    src_sampler.address_mode_x(TensorSamplerAddressModeX::None);
+    src_sampler.address_mode_y(TensorSamplerAddressModeY::None);
+    src_sampler.address_mode_z(TensorSamplerAddressModeZ::None);
+    src_sampler.x(idx_out_c);
+    src_sampler.b(idx_out_n);
+
+    TensorTileSampler dst_sampler;
+    dst_sampler.width(n0);
+    dst_sampler.height(m0);
+    dst_sampler.format(TensorSamplerFormat::C_W_H);
+    dst_sampler.address_mode_x(TensorSamplerAddressModeX::OverlappingMin);
+    dst_sampler.address_mode_y(TensorSamplerAddressModeY::None);
+    dst_sampler.address_mode_z(TensorSamplerAddressModeZ::None);
+    dst_sampler.x(idx_out_c);
+    dst_sampler.y(idx_out_w);
+    dst_sampler.z(idx_out_h);
+    dst_sampler.b(idx_out_n);
+
+    // Prepare dst tensor and tile
+    TileInfo dst_tile_info = TileInfo(to_ckw(src_data_type), m0, n0);
+    if(!dst->has_tile())
+    {
+        TileOperand &dst_tile = writer->declare_tile("dst_tile", dst_tile_info);
+        dst->init_virtual_tensor(dst_tile, dst_sampler);
+    }
+    const TileOperand &dst_tile = dst->tile();
+
+    // A tile used to temporarily store results or as an accumulator in case of AVG and L2 pooling.
+    const TileOperand &res_tile = writer->declare_tile("res_tile", TileInfo(to_ckw(acc_data_type), m0, n0));
+
+    // Initialise result tile with appropriate value
+    if(_attributes.pool_type() == PoolingType::MAX)
+    {
+        if(_settings.use_inf_as_limit())
+        {
+            TileContainer            minus_inf_tile_container;
+            std::vector<std::string> value = std::vector<std::string>(n0, "(-INFINITY)");
+            minus_inf_tile_container.push_back({ value });
+            const TileOperand &minus_inf = writer->declare_tile("minus_inf_const", minus_inf_tile_container, to_ckw(acc_data_type));
+            writer->op_assign(res_tile, minus_inf);
+        }
+        else
+        {
+            writer->op_assign(res_tile, const_lowest_value);
+        }
+    }
+    else
+    {
+        writer->op_assign(res_tile, const_0);
+    }
+
+    // idx_in_w = idx_out_w * STRIDE_X - PAD_X
+    TileOperand &idx_in_w = writer->declare_tile("idx_in_w", ckw::DataType::Int32);
+    writer->op_binary_expression(idx_in_w, idx_out_w, BinaryOp::Mul, stride_x_tile);
+    writer->op_binary_expression(idx_in_w, idx_in_w, BinaryOp::Sub, pad_x_tile);
+
+    // idx_in_h = idx_out_h * STRIDE_Y - PAD_Y
+    TileOperand &idx_in_h = writer->declare_tile("idx_in_h", ckw::DataType::Int32);
+    writer->op_binary_expression(idx_in_h, idx_out_h, BinaryOp::Mul, stride_y_tile);
+    writer->op_binary_expression(idx_in_h, idx_in_h, BinaryOp::Sub, pad_y_tile);
+
+    TileOperand &minus_idx_in_w = writer->declare_tile("minus_idx_in_w", ckw::DataType::Int32);
+    TileOperand &minus_idx_in_h = writer->declare_tile("minus_idx_in_h", ckw::DataType::Int32);
+
+    writer->op_unary_expression(minus_idx_in_w, UnaryOp::Negate, idx_in_w);
+    writer->op_unary_expression(minus_idx_in_h, UnaryOp::Negate, idx_in_h);
+
+    // Pooling starting/ending offsets for X dim
+    TileOperand &pool_x_s = writer->declare_tile("pool_x_s", ckw::DataType::Int32);
+    TileOperand &pool_x_e = writer->declare_tile("pool_x_e", ckw::DataType::Int32);
+
+    writer->op_binary_elementwise_function(pool_x_s, BinaryFunction::Max, const_0, minus_idx_in_w);
+    writer->op_binary_expression(pool_x_e, src_width_tile, BinaryOp::Add, minus_idx_in_w);
+    writer->op_binary_elementwise_function(pool_x_e, BinaryFunction::Min, pool_size_x_tile, pool_x_e);
+
+    // Pooling starting/ending offsets for Y dim
+    TileOperand &pool_y_s = writer->declare_tile("pool_y_s", ckw::DataType::Int32);
+    TileOperand &pool_y_e = writer->declare_tile("pool_y_e", ckw::DataType::Int32);
+
+    writer->op_binary_elementwise_function(pool_y_s, BinaryFunction::Max, const_0, minus_idx_in_h);
+    writer->op_binary_expression(pool_y_e, src_height_tile, BinaryOp::Add, minus_idx_in_h);
+    writer->op_binary_elementwise_function(pool_y_e, BinaryFunction::Min, pool_size_y_tile, pool_y_e);
+
+    const TileOperand &filter_size = writer->declare_tile("filter_size", ckw::DataType::Int32);
+    if(_attributes.exclude_padding())
+    {
+        const TileOperand &y_diff = writer->declare_tile("y_diff", ckw::DataType::Int32);
+        const TileOperand &x_diff = writer->declare_tile("x_diff", ckw::DataType::Int32);
+
+        writer->op_binary_expression(y_diff, pool_y_e, BinaryOp::Sub, pool_y_s);
+        writer->op_binary_expression(x_diff, pool_x_e, BinaryOp::Sub, pool_x_s);
+
+        writer->op_binary_expression(filter_size, y_diff, BinaryOp::Mul, x_diff);
+    }
+    else
+    {
+        writer->op_binary_expression(filter_size, pool_size_x_tile, BinaryOp::Mul, pool_size_y_tile);
+    }
+
+    const TileOperand &x = writer->declare_tile("x", ckw::DataType::Int32);
+    const TileOperand &y = writer->declare_tile("y", ckw::DataType::Int32);
+
+    if(is_global_pooling)
+    {
+        writer->op_assign(x, const_0);
+        writer->op_assign(y, const_0);
+
+        writer->op_assign(pool_y_e, pool_size_y_tile);
+        writer->op_assign(pool_x_e, pool_size_x_tile);
+    }
+    else
+    {
+        writer->op_assign(x, pool_x_s);
+        writer->op_assign(y, pool_y_s);
+    }
+
+    // Y dim for-loop
+    writer->op_for_loop(y, BinaryOp::Less, pool_y_e, y, AssignmentOp::Increment, const_1, [&]()
+    {
+        // Reset the iterator for the inner loop
+        if(is_global_pooling)
+        {
+            writer->op_assign(x, const_0);
+        }
+        else
+        {
+            writer->op_assign(x, pool_x_s);
+        }
+
+        TileOperand &a_y = writer->declare_tile("a_y", ckw::DataType::Int32);
+        writer->op_binary_expression(a_y, idx_in_h, BinaryOp::Add, y);
+
+        // X dim for-loop
+        writer->op_for_loop(x, BinaryOp::Less, pool_x_e, x, AssignmentOp::Increment, const_1, [&]()
+        {
+            TileOperand &a_x = writer->declare_tile("a_x", ckw::DataType::Int32);
+            writer->op_binary_expression(a_x, idx_in_w, BinaryOp::Add, x);
+
+            TileOperand &src_tile = writer->declare_tile("src_tile", TileInfo(to_ckw(acc_data_type), m0, n0));
+
+            src_sampler.y(a_x);
+            src_sampler.z(a_y);
+
+            // Load src tile
+            if(use_fp_mixed_precision)
+            {
+                TileOperand &src_uncasted_tile = writer->declare_tile("uncasted_src_tile", dst_tile_info);
+                writer->op_load(src_uncasted_tile, src->tensor(), src_sampler);
+                writer->op_cast_expression(src_tile, src_uncasted_tile, ckw::ConvertPolicy::None);
+            }
+            else
+            {
+                writer->op_load(src_tile, src->tensor(), src_sampler);
+            }
+
+            // Take the square of the input, for L2 Pooling
+            if(_attributes.pool_type() == PoolingType::L2)
+            {
+                writer->op_binary_expression(src_tile, src_tile, BinaryOp::Mul, src_tile);
+            }
+
+            // Perfom Pooling op
+            if(_attributes.pool_type() == PoolingType::MAX)
+            {
+                writer->op_binary_elementwise_function(res_tile, BinaryFunction::Max, res_tile, src_tile);
+            }
+            else
+            {
+                writer->op_binary_expression(res_tile, res_tile, BinaryOp::Add, src_tile);
+            }
+        });
+    });
+
+    if((_attributes.pool_type() == PoolingType::AVG) || (_attributes.pool_type() == PoolingType::L2))
+    {
+        // filter_size is automatically broadcasted in the operation
+        writer->op_binary_expression(res_tile, res_tile, BinaryOp::Div, filter_size);
+    }
+
+    // Take square root of the result in L2 pooling
+    if(_attributes.pool_type() == PoolingType::L2)
+    {
+        writer->op_unary_elementwise_function(res_tile, UnaryFunction::Sqrt, res_tile);
+    }
+
+    // Store the results and do casting if FP_MIXED_PRECISION
+    if(use_fp_mixed_precision)
+    {
+        writer->op_cast_expression(dst_tile, res_tile, ckw::ConvertPolicy::None);
+    }
+    else
+    {
+        writer->op_assign(dst_tile, res_tile);
+    }
+}
+
+Window GpuCkwPool2d::get_window() const
+{
+    ARM_COMPUTE_ERROR_ON_MSG(_dst->tensor_shape().total_size() == 0U, "Destination tensor is not initialized");
+
+    TensorShape        output_shape = _dst->tensor_shape();
+    const unsigned int vec_size     = adjust_vec_size(((_dst->data_type() == DataType::F32) ? 2 : 4), _dst->dimension(0));
+    // Create and configure kernel window
+    auto win = calculate_max_window(output_shape, Steps(vec_size));
+    win      = win.collapse_if_possible(win, Window::DimZ); // collapse window on batch size.
+    return win;
+}
+
+std::string GpuCkwPool2d::get_name(const ComponentGroup &comp_group) const
+{
+    ARM_COMPUTE_UNUSED(comp_group);
+
+    return "pool2dMxN";
+}
+
+} // namespace dynamic_fusion
+} // namespace experimental
+} // namespace arm_compute
diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.h b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.h
new file mode 100644
index 0000000..2ccf255
--- /dev/null
+++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.h
@@ -0,0 +1,76 @@
+/*
+ * Copyright (c) 2023 Arm Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_GPUCKWPOOL2D_H
+#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_GPUCKWPOOL2D_H
+
+#include "src/core/common/Macros.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/IGpuCkwComponentDriver.h"
+#include "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.h"
+
+#include <string>
+
+namespace arm_compute
+{
+namespace experimental
+{
+namespace dynamic_fusion
+{
+class GpuCkwPool2d : public IGpuCkwComponentDriver
+{
+public:
+    using Attributes = ClComponentPool2d::Attributes;
+    using Settings   = ClComponentPool2d::Settings;
+
+    /** Constructor
+     *
+     * For supported configurations please refer to @ref ClComponentCast::validate()
+     *
+     * @param[in] id         Component id
+     * @param[in] tensors    Tensor arguments to the component
+     * @param[in] attributes Component attributes
+     * @param[in] settings   Component settings
+     */
+    GpuCkwPool2d(ComponentId                      id,
+                 const ArgumentPack<ITensorInfo> &tensors,
+                 const Attributes                &attributes,
+                 const Settings                  &settings);
+    ARM_COMPUTE_DISALLOW_COPY_ALLOW_MOVE(GpuCkwPool2d);
+    /** Destructor */
+    ~GpuCkwPool2d() override = default;
+    // Inherited methods overriden:
+    virtual void write_component_code(const ComponentGroup &comp_group, GpuCkwVariableTable &vtable, GpuCkwScopedKernelWriter writer) const override;
+    Window      get_window() const override;
+    std::string get_name(const ComponentGroup &comp_group) const override;
+
+private:
+    const ITensorInfo *_src;
+    const ITensorInfo *_dst;
+    Attributes         _attributes;
+    Settings           _settings;
+};
+} // namespace dynamic_fusion
+} // namespace experimental
+} // namespace arm_compute
+
+#endif // ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_GPUCKWPOOL2D_H
diff --git a/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.cpp b/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.cpp
index 2b01803..d415769 100644
--- a/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.cpp
+++ b/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.cpp
@@ -28,6 +28,7 @@
 #include "arm_compute/core/utils/misc/ShapeCalculator.h"
 #include "arm_compute/dynamic_fusion/sketch/attributes/Pool2dAttributes.h"
 #include "src/core/CL/CLValidate.h"
+#include "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.h"
 #include "src/dynamic_fusion/sketch/gpu/template_writer/cl/ClTemplatePool2d.h"
 #include "src/dynamic_fusion/utils/Utils.h"
 #include <memory>
@@ -89,16 +90,33 @@
     const Attributes                &attributes,
     const Settings                  &settings)
     : IGpuKernelComponent{ id, properties, tensors },
-      _component_writer{ std::make_unique<ClTemplatePool2d>(id, tensors, attributes, settings) }
+#ifndef ACL_INTERNAL_TEST_CKW_IN_DF
+      _component_writer
+{
+    std::make_unique<ClTemplatePool2d>(id, tensors, attributes, settings)
+}
+#else  //ACL_INTERNAL_TEST_CKW_IN_DF
+      _component_writer
+{
+    std::make_unique<GpuCkwPool2d>(id, tensors, attributes, settings)
+}
+#endif //ACL_INTERNAL_TEST_CKW_IN_DF
 {
 }
 ClComponentPool2d::~ClComponentPool2d()
 {
 }
+#ifndef ACL_INTERNAL_TEST_CKW_IN_DF
 const IGpuTemplateComponentWriter *ClComponentPool2d::template_writer() const
 {
     return _component_writer.get();
 }
+#else  //ACL_INTERNAL_TEST_CKW_IN_DF
+const IGpuCkwComponentDriver *ClComponentPool2d::ckw_component_driver() const
+{
+    return _component_writer.get();
+}
+#endif //ACL_INTERNAL_TEST_CKW_IN_DF
 } // namespace dynamic_fusion
 } // namespace experimental
 } // namespace arm_compute
diff --git a/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.h b/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.h
index 896048e..6814bf9 100644
--- a/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.h
+++ b/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.h
@@ -21,8 +21,8 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef SRC_DYNAMIC_FUSION_SKETCH_GPU_COMPONENTS_CL_CLCOMPONENTPOOL2D
-#define SRC_DYNAMIC_FUSION_SKETCH_GPU_COMPONENTS_CL_CLCOMPONENTPOOL2D
+#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_COMPONENTS_CL_CLCOMPONENTPOOL2D_H
+#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_COMPONENTS_CL_CLCOMPONENTPOOL2D_H
 
 #include "arm_compute/dynamic_fusion/sketch/gpu/operators/GpuPool2d.h"
 #include "src/dynamic_fusion/sketch/gpu/components/IGpuKernelComponent.h"
@@ -41,7 +41,11 @@
 class Pool2dAttributes;
 
 /** Forward declaration */
+#ifndef ACL_INTERNAL_TEST_CKW_IN_DF
 class ClTemplatePool2d;
+#else  // ACL_INTERNAL_TEST_CKW_IN_DF
+class GpuCkwPool2d;
+#endif // ACL_INTERNAL_TEST_CKW_IN_DF
 
 class ClComponentPool2d final : public IGpuKernelComponent
 {
@@ -113,20 +117,28 @@
 
     /** Allow instances of this class to be moved */
     ClComponentPool2d &operator=(ClComponentPool2d &&component) = default;
-
+#ifndef ACL_INTERNAL_TEST_CKW_IN_DF
     /** Get template writer for the component */
     const IGpuTemplateComponentWriter *template_writer() const override;
+#else  // ACL_INTERNAL_TEST_CKW_IN_DF
+    /** Get GPU kernel writer for the component */
+    const IGpuCkwComponentDriver *ckw_component_driver() const override;
+#endif // ACL_INTERNAL_TEST_CKW_IN_DF
 
     /** Get component type */
     GpuComponentType type() const override
     {
-        return GpuComponentType::Unfusable;
+        return GpuComponentType::Complex;
     }
 
 private:
+#ifndef ACL_INTERNAL_TEST_CKW_IN_DF
     std::unique_ptr<ClTemplatePool2d> _component_writer;
+#else  // ACL_INTERNAL_TEST_CKW_IN_DF
+    std::unique_ptr<GpuCkwPool2d> _component_writer;
+#endif // ACL_INTERNAL_TEST_CKW_IN_DF
 };
 } // namespace dynamic_fusion
 } // namespace experimental
 } // namespace arm_compute
-#endif /* SRC_DYNAMIC_FUSION_SKETCH_GPU_COMPONENTS_CL_CLCOMPONENTPOOL2D */
+#endif // ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_COMPONENTS_CL_CLCOMPONENTPOOL2D_H
diff --git a/src/dynamic_fusion/sketch/gpu/operators/GpuPool2d.cpp b/src/dynamic_fusion/sketch/gpu/operators/GpuPool2d.cpp
index e464be8..7ecfa01 100644
--- a/src/dynamic_fusion/sketch/gpu/operators/GpuPool2d.cpp
+++ b/src/dynamic_fusion/sketch/gpu/operators/GpuPool2d.cpp
@@ -46,7 +46,16 @@
 {
 namespace
 {
-constexpr GpuOperatorType operator_type = GpuOperatorType::Unfusable;
+void calculate_and_init_dst_if_empty(ITensorInfo *dst, const ITensorInfo *src, const Pool2dAttributes &attributes, const GpuPool2dSettings &settings)
+{
+    if(dst->total_size() == 0U)
+    {
+        auto shape = misc::shape_calculator::compute_pool_shape(*src, convert_pool_attr_to_pool_info(attributes, settings.mixed_precision()));
+        auto_init_if_empty(*dst, src->clone()->set_tensor_shape(shape));
+    }
+}
+
+constexpr GpuOperatorType operator_type = GpuOperatorType::Complex;
 } // namespace
 
 GpuPool2dSettings &GpuPool2dSettings::mixed_precision(bool mixed_precision)
@@ -73,19 +82,16 @@
 
 Status GpuPool2d::validate_op(const GpuWorkloadSketch &sketch,
                               const ITensorInfo       *src,
-                              const ITensorInfo       *dst,
                               const Pool2dAttributes &attributes,
                               const GpuPool2dSettings &settings)
 {
-    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(src, dst);
-    ARM_COMPUTE_RETURN_ERROR_ON(!src->has_valid_id() || !dst->has_valid_id());
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(src);
+    ARM_COMPUTE_RETURN_ERROR_ON(!src->has_valid_id());
 
     // Auto initialize dst tensor info
-    TensorInfo dst_info_to_validate = *dst;
-    {
-        auto shape = misc::shape_calculator::compute_pool_shape(*src, convert_pool_attr_to_pool_info(attributes, settings.mixed_precision()));
-        auto_init_if_empty(dst_info_to_validate, src->clone()->set_tensor_shape(shape));
-    }
+    TensorInfo dst_info_to_validate;
+
+    calculate_and_init_dst_if_empty(&dst_info_to_validate, src, attributes, settings);
 
     // Perform fusion test
     // Pack tensor infos
@@ -98,16 +104,15 @@
                                     "Operator fusion test failed. This operator cannot be fused into the workload");
 
     // Check if configuration is supported
-    return is_supported_op(*sketch.gpu_context(), src, &dst_info_to_validate, attributes, settings);
+    return is_supported_op(*sketch.gpu_context(), src, attributes, settings);
 }
 
 Status GpuPool2d::is_supported_op(const GpuWorkloadContext &context,
                                   const ITensorInfo        *src,
-                                  const ITensorInfo        *dst,
                                   const Pool2dAttributes   &attributes,
                                   const GpuPool2dSettings &settings)
 {
-    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(src, dst);
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(src);
     // Data type
     ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(src, 1, DataType::F16, DataType::F32);
     // Data layout
@@ -116,11 +121,9 @@
     ARM_COMPUTE_RETURN_ERROR_ON_MSG(!attributes.exclude_padding(), "Exclude padding must be set to true in Attributes!");
 
     // Auto initialize dst tensor info
-    TensorInfo dst_info_to_validate = *dst;
-    {
-        auto shape = misc::shape_calculator::compute_pool_shape(*src, convert_pool_attr_to_pool_info(attributes, settings.mixed_precision()));
-        auto_init_if_empty(dst_info_to_validate, src->clone()->set_tensor_shape(shape));
-    }
+    TensorInfo dst_info_to_validate;
+
+    calculate_and_init_dst_if_empty(&dst_info_to_validate, src, attributes, settings);
 
     // Check components
     if(context.gpu_language() == GpuLanguage::OpenCL)
@@ -145,21 +148,20 @@
     return Status{};
 }
 
-void GpuPool2d::create_op(GpuWorkloadSketch       &sketch,
-                          ITensorInfo             *src,
-                          ITensorInfo             *dst,
-                          const Pool2dAttributes &attributes,
-                          const GpuPool2dSettings &settings)
+ITensorInfo *GpuPool2d::create_op(GpuWorkloadSketch  &sketch,
+                          ITensorInfo                *src,
+                          const Pool2dAttributes     &attributes,
+                          const GpuPool2dSettings    &settings)
 {
     // Assert validation
-    ARM_COMPUTE_ERROR_THROW_ON(GpuPool2d::validate_op(sketch, src, dst, attributes, settings));
-    ARM_COMPUTE_LOG_PARAMS(src, dst, attributes, settings);
+    ARM_COMPUTE_ERROR_THROW_ON(GpuPool2d::validate_op(sketch, src, attributes, settings));
+    ARM_COMPUTE_LOG_PARAMS(src, attributes, settings);
+
+    ITensorInfo *dst = sketch.implementation().create_virtual_tensor();
+    ARM_COMPUTE_ERROR_ON_NULLPTR(dst);
 
     // Auto initialize dst tensor
-    {
-        auto shape = misc::shape_calculator::compute_pool_shape(*src, convert_pool_attr_to_pool_info(attributes, settings.mixed_precision())); // use the default DimensionRoundingType
-        auto_init_if_empty(*dst, src->clone()->set_tensor_shape(shape));
-    }
+    calculate_and_init_dst_if_empty(dst, src, attributes, settings);
 
     // Translate into components and add to component graph
     auto &comp_graph = sketch.implementation().component_graph();
@@ -198,6 +200,8 @@
 
     const auto op = sketch.implementation().operator_group().new_operator(operator_type, tensors);
     sketch.implementation().operator_group().add_operator(op);
+
+    return dst;
 }
 
 } // namespace dynamic_fusion
diff --git a/tests/validation/dynamic_fusion/gpu/cl/Pool2d.cpp b/tests/validation/dynamic_fusion/gpu/cl/Pool2d.cpp
index 7f5efd6..b13364c 100644
--- a/tests/validation/dynamic_fusion/gpu/cl/Pool2d.cpp
+++ b/tests/validation/dynamic_fusion/gpu/cl/Pool2d.cpp
@@ -21,8 +21,7 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-
-#ifndef ACL_INTERNAL_TEST_CKW_IN_DF // Do not include this test if ACL_INTERNAL_TEST_CKW_IN_DF and the op has not been ported to ckw
+#ifdef ACL_INTERNAL_TEST_CKW_IN_DF
 #include "arm_compute/dynamic_fusion/sketch/gpu/operators/GpuPool2d.h"
 
 #include "tests/CL/CLAccessor.h"
@@ -111,7 +110,7 @@
     // Validate Pool2d Configuration
     auto                   src_info    = context.create_tensor_info(input_info);
     auto                   dst_info    = context.create_tensor_info(output_info);
-    bool                   res         = bool(GpuPool2d::validate_op(sketch, &src_info, &dst_info, pool2d_attr, settings));
+    bool                   res         = bool(GpuPool2d::validate_op(sketch, &src_info, pool2d_attr, settings));
     ARM_COMPUTE_EXPECT(res == expected, framework::LogLevel::ERRORS);
 }
 
@@ -232,5 +231,4 @@
 }
 }
 }
-
 #endif // ACL_INTERNAL_TEST_CKW_IN_DF
diff --git a/tests/validation/fixtures/dynamic_fusion/gpu/cl/Pool2dFixture.h b/tests/validation/fixtures/dynamic_fusion/gpu/cl/Pool2dFixture.h
index 0efb761..34f2647 100644
--- a/tests/validation/fixtures/dynamic_fusion/gpu/cl/Pool2dFixture.h
+++ b/tests/validation/fixtures/dynamic_fusion/gpu/cl/Pool2dFixture.h
@@ -21,8 +21,8 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef TESTS_VALIDATION_FIXTURES_DYNAMIC_FUSION_GPU_CL_POOL2DFIXTURE
-#define TESTS_VALIDATION_FIXTURES_DYNAMIC_FUSION_GPU_CL_POOL2DFIXTURE
+#ifndef ACL_TESTS_VALIDATION_FIXTURES_DYNAMIC_FUSION_GPU_CL_POOL2DFIXTURE_H
+#define ACL_TESTS_VALIDATION_FIXTURES_DYNAMIC_FUSION_GPU_CL_POOL2DFIXTURE_H
 
 #include "arm_compute/core/CL/CLKernelLibrary.h"
 #include "arm_compute/core/TensorInfo.h"
@@ -33,6 +33,7 @@
 #include "arm_compute/dynamic_fusion/sketch/attributes/Pool2dAttributes.h"
 #include "arm_compute/dynamic_fusion/sketch/gpu/GpuWorkloadSketch.h"
 #include "arm_compute/dynamic_fusion/sketch/gpu/operators/GpuPool2d.h"
+#include "arm_compute/dynamic_fusion/sketch/gpu/operators/GpuOutput.h"
 #include "src/dynamic_fusion/utils/Utils.h"
 
 #include "tests/CL/CLAccessor.h"
@@ -100,7 +101,8 @@
         // Create Pool2dSettings
         GpuPool2dSettings pool_settings = GpuPool2dSettings().mixed_precision(mixed_precision);
 
-        FunctionType::create_op(sketch, &input_info, &dst_info, pool_attr, pool_settings);
+        ITensorInfo *ans_info = FunctionType::create_op(sketch, &input_info, pool_attr, pool_settings);
+        GpuOutput::create_op(sketch, ans_info, &dst_info);
 
         // Configure runtime
         ClWorkloadRuntime runtime;
@@ -184,4 +186,4 @@
 } // namespace test
 } // namespace arm_compute
 
-#endif /* TESTS_VALIDATION_FIXTURES_DYNAMIC_FUSION_GPU_CL_POOL2DFIXTURE */
+#endif // ACL_TESTS_VALIDATION_FIXTURES_DYNAMIC_FUSION_GPU_CL_POOL2DFIXTURE_H