diff --git a/patches/clang/0003-OpenCL-Support-cl_ext_float_atomics.patch b/patches/clang/0003-OpenCL-Support-cl_ext_float_atomics.patch
index a2e70b73..4eb95445 100644
--- a/patches/clang/0003-OpenCL-Support-cl_ext_float_atomics.patch
+++ b/patches/clang/0003-OpenCL-Support-cl_ext_float_atomics.patch
@@ -1,14 +1,17 @@
-From e601cdf68ecb1a1ce4111a515b22a05845bb07c7 Mon Sep 17 00:00:00 2001
+From d1b4758460cfd04d35ae332033a8d18614e6a7ff Mon Sep 17 00:00:00 2001
 From: haonanya <haonan.yang@intel.com>
 Date: Wed, 28 Jul 2021 14:20:08 +0800
-Subject: [PATCH] [OpenCL] Support cl_ext_float_atomics
+Subject: [PATCH] Support cl_ext_float_atomics
+
+This backports https://reviews.llvm.org/D106343 and https://reviews.llvm.org/D109740
 
 Signed-off-by: haonanya <haonan.yang@intel.com>
 ---
- clang/lib/Headers/opencl-c-base.h     |  19 +++
- clang/lib/Headers/opencl-c.h          | 232 ++++++++++++++++++++++++++
- clang/test/Headers/opencl-c-header.cl |  84 ++++++++++
- 3 files changed, 335 insertions(+)
+ clang/lib/Headers/opencl-c-base.h     |  19 ++
+ clang/lib/Headers/opencl-c.h          | 372 ++++++++++++++++++++++++++
+ clang/lib/Sema/Sema.cpp               |   3 +
+ clang/test/Headers/opencl-c-header.cl |  84 ++++++
+ 4 files changed, 478 insertions(+)
 
 diff --git a/clang/lib/Headers/opencl-c-base.h b/clang/lib/Headers/opencl-c-base.h
 index e8dcd70377e5..d94d64372dbb 100644
@@ -41,15 +44,128 @@ index e8dcd70377e5..d94d64372dbb 100644
  #endif // (defined(__OPENCL_CPP_VERSION__) || __OPENCL_C_VERSION__ >= 200)
  
 diff --git a/clang/lib/Headers/opencl-c.h b/clang/lib/Headers/opencl-c.h
-index ab665628c8e1..501a04f6e82b 100644
+index ab665628c8e1..63a5155fe221 100644
 --- a/clang/lib/Headers/opencl-c.h
 +++ b/clang/lib/Headers/opencl-c.h
-@@ -13531,6 +13531,238 @@ intptr_t __ovld atomic_fetch_max_explicit(volatile atomic_intptr_t *object, uint
+@@ -13531,6 +13531,378 @@ intptr_t __ovld atomic_fetch_max_explicit(volatile atomic_intptr_t *object, uint
  intptr_t __ovld atomic_fetch_max_explicit(volatile atomic_intptr_t *object, uintptr_t opermax, memory_order minder, memory_scope scope);
  #endif
  
++// The functionality added by cl_ext_float_atomics extension
 +#if defined(cl_ext_float_atomics)
 +
++#if defined(__opencl_c_ext_fp16_global_atomic_load_store)
++void __ovld atomic_store(volatile __global atomic_half *object, half operand);
++void __ovld atomic_store_explicit(volatile __global atomic_half *object,
++                                  half operand, memory_order order);
++void __ovld atomic_store_explicit(volatile __global atomic_half *object,
++                                  half operand, memory_order order,
++                                  memory_scope scope);
++half __ovld atomic_load(volatile __global atomic_half *object);
++half __ovld atomic_load_explicit(volatile __global atomic_half *object,
++                                 memory_order order);
++half __ovld atomic_load_explicit(volatile __global atomic_half *object,
++                                 memory_order order, memory_scope scope);
++half __ovld atomic_exchange(volatile __global atomic_half *object,
++                            half operand);
++half __ovld atomic_exchange_explicit(volatile __global atomic_half *object,
++                                     half operand, memory_order order);
++half __ovld atomic_exchange_explicit(volatile __global atomic_half *object,
++                                     half operand, memory_order order,
++                                     memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_global_atomic_load_store)
++
++#if defined(__opencl_c_ext_fp16_local_atomic_load_store)
++void __ovld atomic_store(volatile __local atomic_half *object, half operand);
++void __ovld atomic_store_explicit(volatile __local atomic_half *object,
++                                  half operand, memory_order order);
++void __ovld atomic_store_explicit(volatile __local atomic_half *object,
++                                  half operand, memory_order order,
++                                  memory_scope scope);
++half __ovld atomic_load(volatile __local atomic_half *object);
++half __ovld atomic_load_explicit(volatile __local atomic_half *object,
++                                 memory_order order);
++half __ovld atomic_load_explicit(volatile __local atomic_half *object,
++                                 memory_order order, memory_scope scope);
++half __ovld atomic_exchange(volatile __local atomic_half *object, half operand);
++half __ovld atomic_exchange_explicit(volatile __local atomic_half *object,
++                                     half operand, memory_order order);
++half __ovld atomic_exchange_explicit(volatile __local atomic_half *object,
++                                     half operand, memory_order order,
++                                     memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_local_atomic_load_store)
++
++#if defined(__opencl_c_ext_fp16_global_atomic_load_store) &&                   \
++    defined(__opencl_c_ext_fp16_local_atomic_load_store)
++void __ovld atomic_store(volatile atomic_half *object, half operand);
++void __ovld atomic_store_explicit(volatile atomic_half *object, half operand,
++                                  memory_order order);
++void __ovld atomic_store_explicit(volatile atomic_half *object, half operand,
++                                  memory_order order, memory_scope scope);
++half __ovld atomic_load(volatile atomic_half *object);
++half __ovld atomic_load_explicit(volatile atomic_half *object,
++                                 memory_order order);
++half __ovld atomic_load_explicit(volatile atomic_half *object,
++                                 memory_order order, memory_scope scope);
++half __ovld atomic_exchange(volatile atomic_half *object, half operand);
++half __ovld atomic_exchange_explicit(volatile atomic_half *object, half operand,
++                                     memory_order order);
++half __ovld atomic_exchange_explicit(volatile atomic_half *object, half operand,
++                                     memory_order order, memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_global_atomic_load_store) &&
++       // defined(__opencl_c_ext_fp16_local_atomic_load_store)
++
++#if defined(__opencl_c_ext_fp16_global_atomic_min_max)
++half __ovld atomic_fetch_min(volatile __global atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_max(volatile __global atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_min_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_max_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_min_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++half __ovld atomic_fetch_max_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_global_atomic_min_max)
++
++#if defined(__opencl_c_ext_fp16_local_atomic_min_max)
++half __ovld atomic_fetch_min(volatile __local atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_max(volatile __local atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_min_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_max_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_min_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++half __ovld atomic_fetch_max_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_local_atomic_min_max)
++
++#if defined(__opencl_c_ext_fp16_global_atomic_min_max) &&                      \
++    defined(__opencl_c_ext_fp16_local_atomic_min_max)
++half __ovld atomic_fetch_min(volatile atomic_half *object, half operand);
++half __ovld atomic_fetch_max(volatile atomic_half *object, half operand);
++half __ovld atomic_fetch_min_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_max_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_min_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++half __ovld atomic_fetch_max_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_global_atomic_min_max) &&                \
++    defined(__opencl_c_ext_fp16_local_atomic_min_max)
++
 +#if defined(__opencl_c_ext_fp32_global_atomic_min_max)
 +float __ovld atomic_fetch_min(volatile __global atomic_float *object,
 +                              float operand);
@@ -98,12 +214,10 @@ index ab665628c8e1..501a04f6e82b 100644
 +float __ovld atomic_fetch_max_explicit(volatile atomic_float *object,
 +                                       float operand, memory_order order,
 +                                       memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp32_global_atomic_min_max) &&                      \
++#endif // defined(__opencl_c_ext_fp32_global_atomic_min_max) &&                \
 +    defined(__opencl_c_ext_fp32_local_atomic_min_max)
 +
-+#if defined(__opencl_c_ext_fp64_global_atomic_min_max) &&                      \
-+    defined(cl_khr_int64_base_atomics) &&                                      \
-+    defined(cl_khr_int64_extended_atomics)
++#if defined(__opencl_c_ext_fp64_global_atomic_min_max)
 +double __ovld atomic_fetch_min(volatile __global atomic_double *object,
 +                               double operand);
 +double __ovld atomic_fetch_max(volatile __global atomic_double *object,
@@ -118,13 +232,9 @@ index ab665628c8e1..501a04f6e82b 100644
 +double __ovld atomic_fetch_max_explicit(volatile __global atomic_double *object,
 +                                        double operand, memory_order order,
 +                                        memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp64_global_atomic_min_max) &&
-+       // defined(cl_khr_int64_base_atomics) &&
-+       // defined(cl_khr_int64_extended_atomics)
++#endif // defined(__opencl_c_ext_fp64_global_atomic_min_max)
 +
-+#if defined(__opencl_c_ext_fp64_local_atomic_min_max) &&                       \
-+    defined(cl_khr_int64_base_atomics) &&                                      \
-+    defined(cl_khr_int64_extended_atomics)
++#if defined(__opencl_c_ext_fp64_local_atomic_min_max)
 +double __ovld atomic_fetch_min(volatile __local atomic_double *object,
 +                               double operand);
 +double __ovld atomic_fetch_max(volatile __local atomic_double *object,
@@ -139,14 +249,10 @@ index ab665628c8e1..501a04f6e82b 100644
 +double __ovld atomic_fetch_max_explicit(volatile __local atomic_double *object,
 +                                        double operand, memory_order order,
 +                                        memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp64_local_atomic_min_max) &&
-+       // defined(cl_khr_int64_base_atomics) &&
-+       // defined(cl_khr_int64_extended_atomics)
++#endif // defined(__opencl_c_ext_fp64_local_atomic_min_max)
 +
 +#if defined(__opencl_c_ext_fp64_global_atomic_min_max) &&                      \
-+    defined(__opencl_c_ext_fp64_local_atomic_min_max) &&                       \
-+    defined(cl_khr_int64_base_atomics) &&                                      \
-+    defined(cl_khr_int64_extended_atomics)
++    defined(__opencl_c_ext_fp64_local_atomic_min_max)
 +double __ovld atomic_fetch_min(volatile atomic_double *object, double operand);
 +double __ovld atomic_fetch_max(volatile atomic_double *object, double operand);
 +double __ovld atomic_fetch_min_explicit(volatile atomic_double *object,
@@ -159,10 +265,59 @@ index ab665628c8e1..501a04f6e82b 100644
 +double __ovld atomic_fetch_max_explicit(volatile atomic_double *object,
 +                                        double operand, memory_order order,
 +                                        memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp64_global_atomic_min_max) &&
-+       // defined(__opencl_c_ext_fp64_local_atomic_min_max) &&
-+       // defined(cl_khr_int64_base_atomics) &&
-+       // defined(cl_khr_int64_extended_atomics)
++#endif // defined(__opencl_c_ext_fp64_global_atomic_min_max) &&                \
++    defined(__opencl_c_ext_fp64_local_atomic_min_max)
++
++#if defined(__opencl_c_ext_fp16_global_atomic_add)
++half __ovld atomic_fetch_add(volatile __global atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_sub(volatile __global atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_add_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_sub_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_add_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++half __ovld atomic_fetch_sub_explicit(volatile __global atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_global_atomic_add)
++
++#if defined(__opencl_c_ext_fp16_local_atomic_add)
++half __ovld atomic_fetch_add(volatile __local atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_sub(volatile __local atomic_half *object,
++                             half operand);
++half __ovld atomic_fetch_add_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_sub_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_add_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++half __ovld atomic_fetch_sub_explicit(volatile __local atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_local_atomic_add)
++
++#if defined(__opencl_c_ext_fp16_global_atomic_add) &&                          \
++    defined(__opencl_c_ext_fp16_local_atomic_add)
++half __ovld atomic_fetch_add(volatile atomic_half *object, half operand);
++half __ovld atomic_fetch_sub(volatile atomic_half *object, half operand);
++half __ovld atomic_fetch_add_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_sub_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order);
++half __ovld atomic_fetch_add_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++half __ovld atomic_fetch_sub_explicit(volatile atomic_half *object,
++                                      half operand, memory_order order,
++                                      memory_scope scope);
++#endif // defined(__opencl_c_ext_fp16_global_atomic_add) &&                    \
++    defined(__opencl_c_ext_fp16_local_atomic_add)
 +
 +#if defined(__opencl_c_ext_fp32_global_atomic_add)
 +float __ovld atomic_fetch_add(volatile __global atomic_float *object,
@@ -212,12 +367,10 @@ index ab665628c8e1..501a04f6e82b 100644
 +float __ovld atomic_fetch_sub_explicit(volatile atomic_float *object,
 +                                       float operand, memory_order order,
 +                                       memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp32_global_atomic_add) &&                          \
++#endif // defined(__opencl_c_ext_fp32_global_atomic_add) &&                    \
 +    defined(__opencl_c_ext_fp32_local_atomic_add)
 +
-+#if defined(__opencl_c_ext_fp64_global_atomic_add) &&                          \
-+    defined(cl_khr_int64_base_atomics) &&                                      \
-+    defined(cl_khr_int64_extended_atomics)
++#if defined(__opencl_c_ext_fp64_global_atomic_add)
 +double __ovld atomic_fetch_add(volatile __global atomic_double *object,
 +                               double operand);
 +double __ovld atomic_fetch_sub(volatile __global atomic_double *object,
@@ -232,13 +385,9 @@ index ab665628c8e1..501a04f6e82b 100644
 +double __ovld atomic_fetch_sub_explicit(volatile __global atomic_double *object,
 +                                        double operand, memory_order order,
 +                                        memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp64_global_atomic_add) &&
-+       // defined(cl_khr_int64_base_atomics) &&
-+       // defined(cl_khr_int64_extended_atomics)
++#endif // defined(__opencl_c_ext_fp64_global_atomic_add)
 +
-+#if defined(__opencl_c_ext_fp64_local_atomic_add) &&                           \
-+    defined(cl_khr_int64_base_atomics) &&                                      \
-+    defined(cl_khr_int64_extended_atomics)
++#if defined(__opencl_c_ext_fp64_local_atomic_add)
 +double __ovld atomic_fetch_add(volatile __local atomic_double *object,
 +                               double operand);
 +double __ovld atomic_fetch_sub(volatile __local atomic_double *object,
@@ -253,14 +402,10 @@ index ab665628c8e1..501a04f6e82b 100644
 +double __ovld atomic_fetch_sub_explicit(volatile __local atomic_double *object,
 +                                        double operand, memory_order order,
 +                                        memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp64_local_atomic_add) &&
-+       // defined(cl_khr_int64_base_atomics) &&
-+       // defined(cl_khr_int64_extended_atomics)
++#endif // defined(__opencl_c_ext_fp64_local_atomic_add)
 +
 +#if defined(__opencl_c_ext_fp64_global_atomic_add) &&                          \
-+    defined(__opencl_c_ext_fp64_local_atomic_add) &&                           \
-+    defined(cl_khr_int64_base_atomics) &&                                      \
-+    defined(cl_khr_int64_extended_atomics)
++    defined(__opencl_c_ext_fp64_local_atomic_add)
 +double __ovld atomic_fetch_add(volatile atomic_double *object, double operand);
 +double __ovld atomic_fetch_sub(volatile atomic_double *object, double operand);
 +double __ovld atomic_fetch_add_explicit(volatile atomic_double *object,
@@ -273,16 +418,28 @@ index ab665628c8e1..501a04f6e82b 100644
 +double __ovld atomic_fetch_sub_explicit(volatile atomic_double *object,
 +                                        double operand, memory_order order,
 +                                        memory_scope scope);
-+#endif // defined(__opencl_c_ext_fp64_global_atomic_add) &&
-+       // defined(__opencl_c_ext_fp64_local_atomic_add) &&
-+       // defined(cl_khr_int64_base_atomics) &&
-+       // defined(cl_khr_int64_extended_atomics)
++#endif // defined(__opencl_c_ext_fp64_global_atomic_add) &&                    \
++    defined(__opencl_c_ext_fp64_local_atomic_add)
 +
 +#endif // cl_ext_float_atomics
 +
  // atomic_store()
  
  void __ovld atomic_store(volatile atomic_int *object, int desired);
+diff --git a/clang/lib/Sema/Sema.cpp b/clang/lib/Sema/Sema.cpp
+index 450f9c020f7f..a91291c7af38 100644
+--- a/clang/lib/Sema/Sema.cpp
++++ b/clang/lib/Sema/Sema.cpp
+@@ -313,6 +313,9 @@ void Sema::Initialize() {
+       addImplicitTypedef("atomic_long", AtomicLongT);
+       auto AtomicULongT = Context.getAtomicType(Context.UnsignedLongTy);
+       addImplicitTypedef("atomic_ulong", AtomicULongT);
++      auto AtomicHalfT = Context.getAtomicType(Context.HalfTy);
++      addImplicitTypedef("atomic_half", AtomicHalfT);
++      setOpenCLExtensionForType(AtomicHalfT, "cl_khr_fp16");
+       addImplicitTypedef("atomic_float",
+                          Context.getAtomicType(Context.FloatTy));
+       auto AtomicDoubleT = Context.getAtomicType(Context.DoubleTy);
 diff --git a/clang/test/Headers/opencl-c-header.cl b/clang/test/Headers/opencl-c-header.cl
 index 13a3b62481ec..443f682c711a 100644
 --- a/clang/test/Headers/opencl-c-header.cl