Rename SPV_INTEL_subgroup_buffer_prefetch extension

The extension should be renamed from `subgroups` to `subgroup`
to conform to current naming standards.

(cherry picked from commit face9bf)

Author: web-flow

IGC/BiFModule/CMakeLists.txt

@@ -480,7 +480,7 @@ set(KHR_DEFINES "cl_khr_f16" "cl_khr_fp64" "cl_khr_gl_msaa_sharing" "cl_khr_mipm
                 "cl_intel_subgroups_char" "cl_intel_subgroups_long" "cl_intel_subgroup_local_block_io" "cl_intel_64bit_global_atomics_placeholder"
                 "cl_khr_subgroup_extended_types" "cl_khr_subgroup_non_uniform_vote" "cl_khr_subgroup_ballot" "cl_khr_subgroup_shuffle"
                 "cl_khr_subgroup_shuffle_relative" "cl_khr_subgroup_non_uniform_arithmetic" "cl_khr_subgroup_clustered_reduce"
-                "cl_khr_extended_bit_ops" "cl_intel_bit_instructions" "cl_intel_global_float_atomics" "cl_intel_subgroups_buffer_prefetch")
+                "cl_khr_extended_bit_ops" "cl_intel_bit_instructions" "cl_intel_global_float_atomics" "cl_intel_subgroup_buffer_prefetch")
 set(KHR_DEFINES ${KHR_DEFINES} "cl_intel_subgroup_matrix_multiply_accumulate" "cl_intel_subgroup_split_matrix_multiply_accumulate")
 set(KHR_DEFINES ${KHR_DEFINES} "cl_intel_rt_production")
 set(KHR_DEFINES ${KHR_DEFINES} "cl_intel_subgroup_matrix_multiply_accumulate_tf32")

IGC/BiFModule/Headers/spirv.h

@@ -4830,12 +4830,12 @@ long4  SPIRV_OVERLOADABLE SPIRV_BUILTIN(SubgroupImageBlockReadINTEL, _v4i64_img2
 long8  SPIRV_OVERLOADABLE SPIRV_BUILTIN(SubgroupImageBlockReadINTEL, _v8i64_img2d_ro_v2i32, _Rlong8)(global Img2d_ro* image, int2 coord);
 #endif // cl_intel_subgroups_long
 
-#ifdef cl_intel_subgroups_buffer_prefetch
+#ifdef cl_intel_subgroup_buffer_prefetch
 void SPIRV_OVERLOADABLE SPIRV_BUILTIN(SubgroupBlockPrefetchINTEL, _p1i8, )(const global uchar* ptr, uint num_bytes);
 void SPIRV_OVERLOADABLE SPIRV_BUILTIN(SubgroupBlockPrefetchINTEL, _p1i16, )(const global ushort* ptr, uint num_bytes);
 void SPIRV_OVERLOADABLE SPIRV_BUILTIN(SubgroupBlockPrefetchINTEL, _p1i32, )(const global uint* ptr, uint num_bytes);
 void SPIRV_OVERLOADABLE SPIRV_BUILTIN(SubgroupBlockPrefetchINTEL, _p1i64, )(const global ulong* ptr, uint num_bytes);
-#endif // cl_intel_subgroups_buffer_prefetch
+#endif // cl_intel_subgroup_buffer_prefetch
 
 #define DECL_SUB_GROUP_BROADCAST(TYPE, TYPE_ABBR)       \
 DECL_SUB_GROUP_BROADCAST_BASE(TYPE, TYPE_ABBR)          \

IGC/BiFModule/Implementation/IGCBiF_Intrinsics_Lsc.cl

@@ -405,7 +405,7 @@ ulong  __builtin_IB_lsc_atomic_cmpxchg_global_ulong(volatile __global ulong *bas
 ulong  __builtin_IB_lsc_atomic_cmpxchg_local_ulong(volatile __local ulong *base, int immElemOff, ulong cmp, ulong val);
 #endif // cl_intel_pvc_lsc_validation
 
-#ifdef cl_intel_subgroups_buffer_prefetch
+#ifdef cl_intel_subgroup_buffer_prefetch
 // 1D Block prefetches
 void __builtin_IB_lsc_simd_block_prefetch_uchar(const __global uchar*, enum LSC_LDCC cacheOpt);
 void __builtin_IB_lsc_simd_block_prefetch_uchar2(const __global uchar*, enum LSC_LDCC cacheOpt);
@@ -428,7 +428,7 @@ void __builtin_IB_lsc_simd_block_prefetch_ulong(const __global ulong*, enum LSC_
 void __builtin_IB_lsc_simd_block_prefetch_ulong2(const __global ulong*, enum LSC_LDCC cacheOpt);
 void __builtin_IB_lsc_simd_block_prefetch_ulong4(const __global ulong*, enum LSC_LDCC cacheOpt);
 void __builtin_IB_lsc_simd_block_prefetch_ulong8(const __global ulong*, enum LSC_LDCC cacheOpt);
-#endif // cl_intel_subgroups_buffer_prefetch
+#endif // cl_intel_subgroup_buffer_prefetch
 
 #ifdef cl_intel_subgroup_extended_block_read
 // 2d block read

IGC/BiFModule/Implementation/group.cl

@@ -1738,7 +1738,7 @@ DEF_INTEL_SUB_GROUP_BLOCK_READ_GLOBAL(long4, v4i64, long, i64, simd_block_read_4
 DEF_INTEL_SUB_GROUP_BLOCK_READ_GLOBAL(long8, v8i64, long, i64, simd_block_read_8_global_l)
 #endif // cl_intel_subgroups_long
 
-#ifdef cl_intel_subgroups_buffer_prefetch
+#ifdef cl_intel_subgroup_buffer_prefetch
 
 void __internal_SubgroupBlockPrefetchINTEL_char_cache_controls(const global uchar* ptr, uint num_bytes, enum LSC_LDCC cacheOpt)
 {
@@ -1859,7 +1859,7 @@ void SPIRV_OVERLOADABLE SPIRV_BUILTIN(SubgroupBlockPrefetchINTEL, _p1i64, )(cons
 {
     __internal_SubgroupBlockPrefetchINTEL_long_cache_controls(ptr, num_bytes, LSC_LDCC_DEFAULT);
 }
-#endif // cl_intel_subgroups_buffer_prefetch
+#endif // cl_intel_subgroup_buffer_prefetch
 
 #ifdef cl_intel_subgroup_local_block_io
 

IGC/BiFModule/Languages/OpenCL/IBiF_Sub_Groups.cl

@@ -790,7 +790,7 @@ DEFN_INTEL_SUB_GROUP_BLOCK_WRITE_GLOBAL(intel_sub_group_block_write_ul4, ulong4,
 DEFN_INTEL_SUB_GROUP_BLOCK_WRITE_GLOBAL(intel_sub_group_block_write_ul8, ulong8, ulong, i64, v8i64, __builtin_IB_simd_block_write_8_global_l)
 #endif // cl_intel_subgroups_long
 
-#ifdef cl_intel_subgroups_buffer_prefetch
+#ifdef cl_intel_subgroup_buffer_prefetch
 #define DEFN_INTEL_SUB_GROUP_BLOCK_PREFETCH(FUNC_POSTFIX, PTR_TYPE, PTR_TYPE_MANGLING, NUM_BYTES) \
 void OVERLOADABLE intel_sub_group_block_prefetch_##FUNC_POSTFIX(const global PTR_TYPE* p)         \
 {                                                                                                 \
@@ -818,7 +818,7 @@ DEFN_INTEL_SUB_GROUP_BLOCK_PREFETCH(ul,   ulong,  i64, 8)
 DEFN_INTEL_SUB_GROUP_BLOCK_PREFETCH(ul2,  ulong,  i64, 16)
 DEFN_INTEL_SUB_GROUP_BLOCK_PREFETCH(ul4,  ulong,  i64, 32)
 DEFN_INTEL_SUB_GROUP_BLOCK_PREFETCH(ul8,  ulong,  i64, 64)
-#endif // cl_intel_subgroups_buffer_prefetch
+#endif // cl_intel_subgroup_buffer_prefetch
 
 #ifdef cl_intel_subgroup_local_block_io
 //

IGC/BiFModule/Languages/OpenCL/opencl_cth_released.h

@@ -6501,7 +6501,7 @@ void    __attribute__((overloadable)) intel_sub_group_block_write_ul8(__local ul
 
 #endif // cl_intel_subgroup_local_block_io
 
-#ifdef cl_intel_subgroups_buffer_prefetch
+#ifdef cl_intel_subgroup_buffer_prefetch
 void __attribute__((overloadable)) intel_sub_group_block_prefetch_uc(const __global uchar* p);
 void __attribute__((overloadable)) intel_sub_group_block_prefetch_uc2(const __global uchar* p);
 void __attribute__((overloadable)) intel_sub_group_block_prefetch_uc4(const __global uchar* p);
@@ -6523,7 +6523,7 @@ void __attribute__((overloadable)) intel_sub_group_block_prefetch_ul(const __glo
 void __attribute__((overloadable)) intel_sub_group_block_prefetch_ul2(const __global ulong* p);
 void __attribute__((overloadable)) intel_sub_group_block_prefetch_ul4(const __global ulong* p);
 void __attribute__((overloadable)) intel_sub_group_block_prefetch_ul8(const __global ulong* p);
-#endif // cl_intel_subgroups_buffer_prefetch
+#endif // cl_intel_subgroup_buffer_prefetch
 
 #if defined(cl_khr_subgroup_non_uniform_arithmetic) || defined(cl_khr_subgroup_clustered_reduce)
 #define DECL_SUB_GROUP_NON_UNIFORM_OPERATION(TYPE, GROUP_TYPE, OPERATION)

IGC/ocloc_tests/Builtins/cl_intel_subgroup_buffer_prefetch/PVC/block_prefetch.cl

@@ -11,7 +11,7 @@ SPDX-License-Identifier: MIT
 // UNSUPPORTED: sys32
 
 // RUN: ocloc compile -file %s -device pvc -options "-igc_opts 'DumpVISAASMToConsole=1'" \
-// RUN: -internal_options "-cl-ext=-all,+cl_intel_subgroups_buffer_prefetch" | FileCheck %s
+// RUN: -internal_options "-cl-ext=-all,+cl_intel_subgroup_buffer_prefetch" | FileCheck %s
 
 __attribute__((intel_reqd_sub_group_size(16)))
 kernel void test_char(const global uchar* buffer) {