diff --git a/tt_metal/hw/ckernels/blackhole/metal/llk_api/llk_sfpu/ckernel_sfpu_shift.h b/tt_metal/hw/ckernels/blackhole/metal/llk_api/llk_sfpu/ckernel_sfpu_shift.h
new file mode 100644
index 00000000000..ccd4b2e6df2
--- /dev/null
+++ b/tt_metal/hw/ckernels/blackhole/metal/llk_api/llk_sfpu/ckernel_sfpu_shift.h
@@ -0,0 +1,27 @@
+// SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include "ckernel.h"
+#include "ckernel_defs.h"
+#include "sfpi.h"
+
+using namespace sfpi;
+
+namespace ckernel {
+namespace sfpu {
+
+template <bool APPROXIMATION_MODE, int ITERATIONS = 8>
+inline void calculate_binary_left_shift(const uint dst_offset) {
+    _calculate_binary_left_shift_<APPROXIMATION_MODE, ITERATIONS>(dst_offset);
+}
+
+template <bool APPROXIMATION_MODE, int ITERATIONS = 8>
+inline void calculate_binary_right_shift(const uint dst_offset) {
+    _calculate_binary_right_shift_<APPROXIMATION_MODE, ITERATIONS>(dst_offset);
+}
+
+}  // namespace sfpu
+}  // namespace ckernel
diff --git a/tt_metal/hw/ckernels/blackhole/metal/llk_api/llk_sfpu/llk_math_eltwise_binary_sfpu_shift.h b/tt_metal/hw/ckernels/blackhole/metal/llk_api/llk_sfpu/llk_math_eltwise_binary_sfpu_shift.h
new file mode 100644
index 00000000000..337fdd9df5c
--- /dev/null
+++ b/tt_metal/hw/ckernels/blackhole/metal/llk_api/llk_sfpu/llk_math_eltwise_binary_sfpu_shift.h
@@ -0,0 +1,34 @@
+// SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include "llk_math_eltwise_binary_sfpu_init.h"
+#include "llk_math_eltwise_binary_sfpu_params.h"
+#include "ckernel_sfpu_shift.h"
+
+namespace ckernel {
+
+// New LLK SFPU APIs
+
+template <bool APPROXIMATE>
+inline void llk_math_eltwise_binary_sfpu_shift_init() {
+    llk_math_eltwise_binary_sfpu_init<SfpuType::unused, APPROXIMATE>();
+}
+
+template <bool APPROXIMATE>
+inline void llk_math_eltwise_binary_sfpu_left_shift(
+    uint dst_index0, uint32_t dst_index1, int vector_mode = VectorMode::RC) {
+    llk_math_eltwise_binary_sfpu_params<APPROXIMATE>(
+        ckernel::sfpu::calculate_binary_left_shift<APPROXIMATE>, dst_index0, dst_index1, vector_mode);
+}
+
+template <bool APPROXIMATE>
+inline void llk_math_eltwise_binary_sfpu_right_shift(
+    uint dst_index0, uint32_t dst_index1, int vector_mode = VectorMode::RC) {
+    llk_math_eltwise_binary_sfpu_params<APPROXIMATE>(
+        ckernel::sfpu::calculate_binary_right_shift<APPROXIMATE>, dst_index0, dst_index1, vector_mode);
+}
+
+}  // namespace ckernel
diff --git a/tt_metal/hw/ckernels/wormhole_b0/metal/llk_api/llk_sfpu/ckernel_sfpu_shift.h b/tt_metal/hw/ckernels/wormhole_b0/metal/llk_api/llk_sfpu/ckernel_sfpu_shift.h
new file mode 100644
index 00000000000..ccd4b2e6df2
--- /dev/null
+++ b/tt_metal/hw/ckernels/wormhole_b0/metal/llk_api/llk_sfpu/ckernel_sfpu_shift.h
@@ -0,0 +1,27 @@
+// SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include "ckernel.h"
+#include "ckernel_defs.h"
+#include "sfpi.h"
+
+using namespace sfpi;
+
+namespace ckernel {
+namespace sfpu {
+
+template <bool APPROXIMATION_MODE, int ITERATIONS = 8>
+inline void calculate_binary_left_shift(const uint dst_offset) {
+    _calculate_binary_left_shift_<APPROXIMATION_MODE, ITERATIONS>(dst_offset);
+}
+
+template <bool APPROXIMATION_MODE, int ITERATIONS = 8>
+inline void calculate_binary_right_shift(const uint dst_offset) {
+    _calculate_binary_right_shift_<APPROXIMATION_MODE, ITERATIONS>(dst_offset);
+}
+
+}  // namespace sfpu
+}  // namespace ckernel
diff --git a/tt_metal/hw/ckernels/wormhole_b0/metal/llk_api/llk_sfpu/llk_math_eltwise_binary_sfpu_shift.h b/tt_metal/hw/ckernels/wormhole_b0/metal/llk_api/llk_sfpu/llk_math_eltwise_binary_sfpu_shift.h
new file mode 100644
index 00000000000..337fdd9df5c
--- /dev/null
+++ b/tt_metal/hw/ckernels/wormhole_b0/metal/llk_api/llk_sfpu/llk_math_eltwise_binary_sfpu_shift.h
@@ -0,0 +1,34 @@
+// SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include "llk_math_eltwise_binary_sfpu_init.h"
+#include "llk_math_eltwise_binary_sfpu_params.h"
+#include "ckernel_sfpu_shift.h"
+
+namespace ckernel {
+
+// New LLK SFPU APIs
+
+template <bool APPROXIMATE>
+inline void llk_math_eltwise_binary_sfpu_shift_init() {
+    llk_math_eltwise_binary_sfpu_init<SfpuType::unused, APPROXIMATE>();
+}
+
+template <bool APPROXIMATE>
+inline void llk_math_eltwise_binary_sfpu_left_shift(
+    uint dst_index0, uint32_t dst_index1, int vector_mode = VectorMode::RC) {
+    llk_math_eltwise_binary_sfpu_params<APPROXIMATE>(
+        ckernel::sfpu::calculate_binary_left_shift<APPROXIMATE>, dst_index0, dst_index1, vector_mode);
+}
+
+template <bool APPROXIMATE>
+inline void llk_math_eltwise_binary_sfpu_right_shift(
+    uint dst_index0, uint32_t dst_index1, int vector_mode = VectorMode::RC) {
+    llk_math_eltwise_binary_sfpu_params<APPROXIMATE>(
+        ckernel::sfpu::calculate_binary_right_shift<APPROXIMATE>, dst_index0, dst_index1, vector_mode);
+}
+
+}  // namespace ckernel
diff --git a/tt_metal/include/compute_kernel_api/add_int32_sfpu.h b/tt_metal/include/compute_kernel_api/add_int32_sfpu.h
index 4de5ee5b55a..f566c7e34da 100644
--- a/tt_metal/include/compute_kernel_api/add_int32_sfpu.h
+++ b/tt_metal/include/compute_kernel_api/add_int32_sfpu.h
@@ -19,6 +19,8 @@ namespace ckernel {
  * Performs an elementwise add operation with the two integer inputs: y = add(x0,x1)
  * Output overwrites first operand in DST.
  *
+ * The DST register buffer must be in acquired state via *acquire_dst* call. This call is blocking and is only available
+ * on the compute engine.
  * A maximum of 4 tiles from each operand can be loaded into DST at once, for a total of 8 tiles,
  * when using 16 bit formats. This gets reduced to 2 tiles from each operand for 32 bit formats.
  *
diff --git a/tt_metal/include/compute_kernel_api/binary_bitwise_sfpu.h b/tt_metal/include/compute_kernel_api/binary_bitwise_sfpu.h
index cf2a20d0090..1ec6d40ceca 100644
--- a/tt_metal/include/compute_kernel_api/binary_bitwise_sfpu.h
+++ b/tt_metal/include/compute_kernel_api/binary_bitwise_sfpu.h
@@ -19,6 +19,8 @@ namespace ckernel {
  * Performs an elementwise binary bitwise operation with the two inputs: y = bitwise(x0,x1)
  * Output overwrites first operand in DST.
  *
+ * The DST register buffer must be in acquired state via *acquire_dst* call. This call is blocking and is only available
+ * on the compute engine.
  * A maximum of 4 tiles from each operand can be loaded into DST at once, for a total of 8 tiles,
  * when using 16 bit formats. This gets reduced to 2 tiles from each operand for 32 bit formats.
  *
diff --git a/tt_metal/include/compute_kernel_api/binary_shift.h b/tt_metal/include/compute_kernel_api/binary_shift.h
new file mode 100644
index 00000000000..3bd2ddb9a59
--- /dev/null
+++ b/tt_metal/include/compute_kernel_api/binary_shift.h
@@ -0,0 +1,68 @@
+// SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include "compute_kernel_api/common_globals.h"
+#ifdef TRISC_MATH
+#include "llk_math_eltwise_binary_sfpu_shift.h"
+#define MAIN math_main()
+#define MATH(x) x
+#else
+#define MATH(x)
+#endif
+
+namespace ckernel {
+
+/**
+ * Performs an elementwise shift operation to the left on the input at idst0, by input at idst1: y = x0 << x1
+ * Both inputs must be of Int32 data type only. Output overwrites first operand in DST.
+ *
+ * The DST register buffer must be in acquired state via *acquire_dst* call. This call is blocking and is only available
+ * on the compute engine.
+ * A maximum of 4 tiles from each operand can be loaded into DST at once, for a total of 8 tiles,
+ * when using 16 bit formats. This gets reduced to 2 tiles from each operand for 32 bit formats.
+ *
+ * Return value: None
+ *
+ * | Argument       | Description                                                           | Type     | Valid Range |
+ * Required |
+ * |----------------|-----------------------------------------------------------------------|----------|-------------------------------------------------------|----------|
+ * | idst0          | The index of the tile in DST register buffer to use as first operand  | uint32_t | Must be less
+ * than the size of the DST register buffer | True     | | idst1          | The index of the tile in DST register buffer
+ * to use as second operand | uint32_t | Must be less than the size of the DST register buffer | True     |
+ */
+ALWI void binary_left_shift_tile(uint32_t idst0, uint32_t idst1) {
+    MATH((llk_math_eltwise_binary_sfpu_left_shift<APPROX>(idst0, idst1)));
+}
+
+/**
+ * Performs an elementwise shift operation to the right on the input at idst0, by input at idst1: y = x0 >> x1
+ * Both inputs must be of Int32 data type only. Output overwrites first operand in DST.
+ *
+ * The DST register buffer must be in acquired state via *acquire_dst* call. This call is blocking and is only available
+ * on the compute engine.
+ * A maximum of 4 tiles from each operand can be loaded into DST at once, for a total of 8 tiles,
+ * when using 16 bit formats. This gets reduced to 2 tiles from each operand for 32 bit formats.
+ *
+ * Return value: None
+ *
+ * | Argument       | Description                                                           | Type     | Valid Range |
+ * Required |
+ * |----------------|-----------------------------------------------------------------------|----------|-------------------------------------------------------|----------|
+ * | idst0          | The index of the tile in DST register buffer to use as first operand  | uint32_t | Must be less
+ * than the size of the DST register buffer | True     | | idst1          | The index of the tile in DST register buffer
+ * to use as second operand | uint32_t | Must be less than the size of the DST register buffer | True     |
+ */
+
+ALWI void binary_right_shift_tile(uint32_t idst0, uint32_t idst1) {
+    MATH((llk_math_eltwise_binary_sfpu_right_shift<APPROX>(idst0, idst1)));
+}
+
+/**
+ * Please refer to documentation for any_init.
+ */
+ALWI void binary_shift_tile_init() { MATH((llk_math_eltwise_binary_sfpu_shift_init<APPROX>())); }
+
+}  // namespace ckernel
diff --git a/tt_metal/include/compute_kernel_api/eltwise_binary_sfpu.h b/tt_metal/include/compute_kernel_api/eltwise_binary_sfpu.h
index 22fc4c13fcf..23995891940 100644
--- a/tt_metal/include/compute_kernel_api/eltwise_binary_sfpu.h
+++ b/tt_metal/include/compute_kernel_api/eltwise_binary_sfpu.h
@@ -19,6 +19,8 @@ namespace ckernel {
  * Performs an elementwise binop operation with the two floating point inputs: y = binop(x0,x1)
  * Output overwrites first operand in DST.
  *
+ * The DST register buffer must be in acquired state via *acquire_dst* call. This call is blocking and is only available
+ * on the compute engine.
  * A maximum of 4 tiles from each operand can be loaded into DST at once, for a total of 8 tiles,
  * when using 16 bit formats. This gets reduced to 2 tiles from each operand for 32 bit formats.
  *
diff --git a/tt_metal/third_party/tt_llk_blackhole b/tt_metal/third_party/tt_llk_blackhole
index 7536fbacd75..973288fb014 160000
--- a/tt_metal/third_party/tt_llk_blackhole
+++ b/tt_metal/third_party/tt_llk_blackhole
@@ -1 +1 @@
-Subproject commit 7536fbacd75a4ad62047c63c9c54176fae079e06
+Subproject commit 973288fb014a22ce72cdba1c38a9f41f48532d6d
diff --git a/tt_metal/third_party/tt_llk_wormhole_b0 b/tt_metal/third_party/tt_llk_wormhole_b0
index 0f57d4e9dec..33a7f6a0267 160000
--- a/tt_metal/third_party/tt_llk_wormhole_b0
+++ b/tt_metal/third_party/tt_llk_wormhole_b0
@@ -1 +1 @@
-Subproject commit 0f57d4e9dec602b68671be8891e7af876285f275
+Subproject commit 33a7f6a026719af509a119d8a4e8e36c7c31854c