diff --git a/src/ATen/native/xpu/Activation.cpp b/src/ATen/native/xpu/Activation.cpp
index fcbbe99db..38aa44dc6 100644
--- a/src/ATen/native/xpu/Activation.cpp
+++ b/src/ATen/native/xpu/Activation.cpp
@@ -11,12 +11,13 @@
 #include <ATen/native/xpu/sycl/ActivationLeakyReluKernels.h>
 #include <ATen/native/xpu/sycl/ActivationLogSigmoidKernels.h>
 #include <ATen/native/xpu/sycl/ActivationMishKernels.h>
+#include <ATen/native/xpu/sycl/ActivationPreluKernels.h>
 #include <ATen/native/xpu/sycl/ActivationSiluKernels.h>
 #include <ATen/native/xpu/sycl/ActivationSoftplusKernels.h>
 #include <ATen/native/xpu/sycl/ActivationSoftshrinkKernels.h>
 #include <ATen/native/xpu/sycl/ActivationThresholdKernel.h>
-namespace at {
 
+namespace at {
 Tensor XPUNativeFunctions::relu(const Tensor& self) {
   TORCH_CHECK(
       self.scalar_type() != at::kBool, "Boolean inputs not supported for relu");
@@ -633,6 +634,37 @@ Tensor& XPUNativeFunctions::softshrink_backward_out(
   return grad_input;
 }
 
+Tensor XPUNativeFunctions::_prelu_kernel(
+    const Tensor& self,
+    const Tensor& weight) {
+  // Weight broadcasts over self and they have the same dtype
+  auto result = at::empty_like(self);
+  auto iter = TensorIteratorConfig()
+                  .add_output(result)
+                  .add_const_input(self)
+                  .add_const_input(weight)
+                  .build();
+  native::xpu::prelu_kernel(iter);
+  return result;
+}
+
+std::tuple<Tensor, Tensor> XPUNativeFunctions::_prelu_kernel_backward(
+    const Tensor& grad_out,
+    const Tensor& self,
+    const Tensor& weight) {
+  Tensor grad_self = at::empty({0}, self.options());
+  Tensor grad_weight = at::empty({0}, weight.options());
+  auto iter = TensorIteratorConfig()
+                  .add_output(grad_self)
+                  .add_output(grad_weight)
+                  .add_const_input(self)
+                  .add_const_input(weight)
+                  .add_const_input(grad_out)
+                  .build();
+  native::xpu::prelu_backward_kernel(iter);
+  return {grad_self, grad_weight};
+}
+
 std::tuple<Tensor&, Tensor&> XPUNativeFunctions::log_sigmoid_forward_out(
     const Tensor& input,
     Tensor& result,
diff --git a/src/ATen/native/xpu/BinaryOps.cpp b/src/ATen/native/xpu/BinaryOps.cpp
index 14bc49990..e17309841 100644
--- a/src/ATen/native/xpu/BinaryOps.cpp
+++ b/src/ATen/native/xpu/BinaryOps.cpp
@@ -10,6 +10,7 @@
 #include <ATen/native/xpu/sycl/BinaryLogicalOpsKernels.h>
 #include <ATen/native/xpu/sycl/BinaryMiscBackwardOpsKernels.h>
 #include <ATen/native/xpu/sycl/BinaryRemainderKernel.h>
+#include <ATen/native/xpu/sycl/BinaryShiftOpsKernels.h>
 #include <ATen/native/xpu/sycl/CopysignKernel.h>
 #include <ATen/native/xpu/sycl/GcdLcmKernels.h>
 #include <ATen/native/xpu/sycl/LogAddExpKernels.h>
@@ -301,6 +302,80 @@ Tensor& XPUNativeFunctions::bitwise_xor_out(
   return out;
 }
 
+Tensor XPUNativeFunctions::__lshift__(const Tensor& self, const Tensor& other) {
+  Tensor result;
+  auto iter = TensorIterator::binary_op(result, self, other);
+  native::xpu::lshift_kernel(iter);
+  return iter.output();
+}
+
+Tensor XPUNativeFunctions::__lshift__(const Tensor& self, const Scalar& other) {
+  Tensor result;
+  auto wrapper = native::wrapped_scalar_tensor(other);
+  auto iter = TensorIterator::binary_op(result, self, wrapper);
+  native::xpu::lshift_kernel(iter);
+  return iter.output();
+}
+
+Tensor& XPUNativeFunctions::__ilshift__(Tensor& self, const Tensor& other) {
+  auto iter = TensorIterator::binary_op(self, self, other);
+  native::xpu::lshift_kernel(iter);
+  return self;
+}
+
+Tensor& XPUNativeFunctions::__ilshift__(Tensor& self, const Scalar& other) {
+  auto wrapper = native::wrapped_scalar_tensor(other);
+  auto iter = TensorIterator::binary_op(self, self, wrapper);
+  native::xpu::lshift_kernel(iter);
+  return self;
+}
+
+Tensor& XPUNativeFunctions::bitwise_left_shift_out(
+    const Tensor& self,
+    const Tensor& other,
+    Tensor& result) {
+  auto iter = TensorIterator::borrowing_binary_op(result, self, other);
+  native::xpu::lshift_kernel(iter);
+  return result;
+}
+
+Tensor XPUNativeFunctions::__rshift__(const Tensor& self, const Tensor& other) {
+  Tensor result;
+  auto iter = TensorIterator::binary_op(result, self, other);
+  native::xpu::rshift_kernel(iter);
+  return iter.output();
+}
+
+Tensor XPUNativeFunctions::__rshift__(const Tensor& self, const Scalar& other) {
+  Tensor result;
+  auto wrapper = native::wrapped_scalar_tensor(other);
+  auto iter = TensorIterator::binary_op(result, self, wrapper);
+  native::xpu::rshift_kernel(iter);
+  return iter.output();
+}
+
+Tensor& XPUNativeFunctions::__irshift__(Tensor& self, const Tensor& other) {
+  auto iter = TensorIterator::binary_op(self, self, other);
+  native::xpu::rshift_kernel(iter);
+  return self;
+}
+
+Tensor& XPUNativeFunctions::__irshift__(Tensor& self, const Scalar& other) {
+  auto wrapper = native::wrapped_scalar_tensor(other);
+  auto iter = TensorIterator::binary_op(self, self, wrapper);
+  native::xpu::rshift_kernel(iter);
+  return self;
+}
+
+Tensor& XPUNativeFunctions::bitwise_right_shift_out(
+    const Tensor& self,
+    const Tensor& other,
+    Tensor& result) {
+  auto iter = TensorIterator::borrowing_binary_op(result, self, other);
+  native::xpu::rshift_kernel(iter);
+  return result;
+}
+
 Tensor XPUNativeFunctions::gcd(const Tensor& self, const Tensor& other) {
   Tensor out;
   auto iter = TensorIterator::borrowing_binary_op(out, self, other);
diff --git a/src/ATen/native/xpu/XPUFallback.template b/src/ATen/native/xpu/XPUFallback.template
index 5b2d6e5ff..75cbd3c1c 100644
--- a/src/ATen/native/xpu/XPUFallback.template
+++ b/src/ATen/native/xpu/XPUFallback.template
@@ -236,11 +236,8 @@ TORCH_LIBRARY_IMPL(aten, XPU, m) {
     "ormqr",
     "_pdist_backward",
     "_pdist_forward",
-    "_prelu_kernel",
-    "_prelu_kernel_backward",
     "put_",
     "rrelu_with_noise",
-    "__rshift__.Scalar",
     "_scaled_dot_product_efficient_attention",
     "_scaled_mm",
     "segment_reduce",
diff --git a/src/ATen/native/xpu/sycl/ActivationPreluKernels.cpp b/src/ATen/native/xpu/sycl/ActivationPreluKernels.cpp
new file mode 100644
index 000000000..8a2c7012f
--- /dev/null
+++ b/src/ATen/native/xpu/sycl/ActivationPreluKernels.cpp
@@ -0,0 +1,43 @@
+#include <ATen/ATen.h>
+#include <ATen/Dispatch.h>
+#include <ATen/native/TensorIterator.h>
+
+#include <ATen/native/xpu/sycl/Loops.h>
+
+namespace at::native::xpu {
+
+template <typename scalar_t>
+struct PreluFunctor {
+  scalar_t operator()(scalar_t input, scalar_t weight) const {
+    return (input > 0) ? input : weight * input;
+  }
+};
+
+template <typename scalar_t>
+struct PreluBackwardFunctor {
+  std::tuple<scalar_t, scalar_t> operator()(
+      scalar_t input,
+      scalar_t weight,
+      scalar_t grad) const {
+    auto mask = input > 0;
+    auto grad_input = mask ? grad : weight * grad;
+    auto grad_weight = mask ? scalar_t{0} : input * grad;
+    return std::tuple<scalar_t, scalar_t>{grad_input, grad_weight};
+  }
+};
+
+void prelu_kernel(TensorIterator& iter) {
+  AT_DISPATCH_FLOATING_TYPES_AND2(
+      kBFloat16, kHalf, iter.dtype(), "prelu_xpu", [&] {
+        gpu_kernel(iter, PreluFunctor<scalar_t>());
+      });
+}
+
+void prelu_backward_kernel(TensorIterator& iter) {
+  AT_DISPATCH_FLOATING_TYPES_AND2(
+      kBFloat16, kHalf, iter.dtype(), "prelu_backward_xpu", [&] {
+        gpu_kernel_multiple_outputs(iter, PreluBackwardFunctor<scalar_t>());
+      });
+}
+
+} // namespace at::native::xpu
\ No newline at end of file
diff --git a/src/ATen/native/xpu/sycl/ActivationPreluKernels.h b/src/ATen/native/xpu/sycl/ActivationPreluKernels.h
new file mode 100644
index 000000000..23f60c242
--- /dev/null
+++ b/src/ATen/native/xpu/sycl/ActivationPreluKernels.h
@@ -0,0 +1,11 @@
+#pragma once
+
+#include <ATen/native/TensorIterator.h>
+
+namespace at::native::xpu {
+
+void prelu_kernel(TensorIterator& iter);
+
+void prelu_backward_kernel(TensorIterator& iter);
+
+} // namespace at::native::xpu
diff --git a/src/ATen/native/xpu/sycl/BinaryShiftOpsKernels.cpp b/src/ATen/native/xpu/sycl/BinaryShiftOpsKernels.cpp
new file mode 100644
index 000000000..64adba17e
--- /dev/null
+++ b/src/ATen/native/xpu/sycl/BinaryShiftOpsKernels.cpp
@@ -0,0 +1,48 @@
+#include <ATen/ATen.h>
+#include <ATen/Dispatch.h>
+#include <ATen/native/TensorIterator.h>
+
+#include <ATen/native/xpu/sycl/Loops.h>
+
+namespace at::native::xpu {
+
+template <typename scalar_t>
+struct LshiftFunctor {
+  scalar_t operator()(scalar_t a, scalar_t b) const {
+    constexpr scalar_t max_shift = sizeof(scalar_t) * CHAR_BIT;
+    if ((static_cast<std::make_signed_t<scalar_t>>(b) < 0) ||
+        (b >= max_shift)) {
+      return 0;
+    }
+    return static_cast<std::make_unsigned_t<scalar_t>>(a) << b;
+  }
+};
+
+void lshift_kernel(TensorIteratorBase& iter) {
+  AT_DISPATCH_INTEGRAL_TYPES(iter.dtype(), "lshift_xpu", [&]() {
+    gpu_kernel_with_scalars(iter, LshiftFunctor<scalar_t>());
+  });
+}
+
+template <typename scalar_t>
+struct RshiftFunctor {
+  scalar_t operator()(scalar_t a, scalar_t b) const {
+    // right shift value to retain sign bit for signed and no bits for
+    // unsigned
+    constexpr scalar_t max_shift =
+        sizeof(scalar_t) * CHAR_BIT - std::is_signed_v<scalar_t>;
+    if ((static_cast<std::make_signed_t<scalar_t>>(b) < 0) ||
+        (b >= max_shift)) {
+      return a >> max_shift;
+    }
+    return a >> b;
+  }
+};
+
+void rshift_kernel(TensorIteratorBase& iter) {
+  AT_DISPATCH_INTEGRAL_TYPES(iter.dtype(), "rshift_xpu", [&]() {
+    gpu_kernel_with_scalars(iter, RshiftFunctor<scalar_t>());
+  });
+}
+
+} // namespace at::native::xpu
\ No newline at end of file
diff --git a/src/ATen/native/xpu/sycl/BinaryShiftOpsKernels.h b/src/ATen/native/xpu/sycl/BinaryShiftOpsKernels.h
new file mode 100644
index 000000000..7e661d919
--- /dev/null
+++ b/src/ATen/native/xpu/sycl/BinaryShiftOpsKernels.h
@@ -0,0 +1,11 @@
+#pragma once
+
+#include <ATen/native/TensorIterator.h>
+
+namespace at::native::xpu {
+
+void lshift_kernel(TensorIteratorBase& iter);
+
+void rshift_kernel(TensorIteratorBase& iter);
+
+} // namespace at::native::xpu
diff --git a/test/xpu/xpu_test_utils.py b/test/xpu/xpu_test_utils.py
index c3a268024..3bc809743 100644
--- a/test/xpu/xpu_test_utils.py
+++ b/test/xpu/xpu_test_utils.py
@@ -54,6 +54,8 @@
     "bitwise_not",
     "bitwise_or",
     "bitwise_xor",
+    "bitwise_left_shift",
+    "bitwise_right_shift",
     "addcmul",
     "addcdiv",
     "clamp",
@@ -112,6 +114,7 @@
     "nn.functional.glu",
     "nn.functional.pad",
     "nn.functional.leaky_relu",
+    "nn.functional.prelu",
     "nn.functional.threshold",
     "nn.functional.silu",
     "nn.functional.hardsigmoid",
diff --git a/yaml/xpu_functions.yaml b/yaml/xpu_functions.yaml
index 263c522d0..8d92fa60e 100644
--- a/yaml/xpu_functions.yaml
+++ b/yaml/xpu_functions.yaml
@@ -118,6 +118,18 @@ supported:
   - relu
   - relu_
   - relu.out
+  - _prelu_kernel
+  - _prelu_kernel_backward
+  - __lshift__.Scalar
+  - __lshift__.Tensor
+  - __ilshift__.Scalar
+  - __ilshift__.Tensor
+  - __rshift__.Scalar
+  - __rshift__.Tensor
+  - __irshift__.Scalar
+  - __irshift__.Tensor
+  - bitwise_left_shift.Tensor_out
+  - bitwise_right_shift.Tensor_out
   - threshold
   - threshold_
   - threshold.out