gcc/ChangeLog:

        * config/i386/avx10v2auxintrin.h (_mm_cvtbf8_bf4s): New intrins.
        (_mm256_cvtbf8_bf4s): Ditto.
        (_mm512_cvtbf8_bf4s): Ditto.
        (_mm_cvtbf8_bf4s_storeu): Ditto.
        (_mm256_cvtbf8_bf4s_storeu): Ditto.
        (_mm512_cvtbf8_bf4s_storeu): Ditto.
        (_mm_cvthf8_bf4s): Ditto.
        (_mm256_cvthf8_bf4s): Ditto.
        (_mm512_cvthf8_bf4s): Ditto.
        (_mm_cvthf8_bf4s_storeu): Ditto.
        (_mm256_cvthf8_bf4s_storeu): Ditto.
        (_mm512_cvthf8_bf4s_storeu): Ditto.
        (_mm_cvtbf4_hf8): Ditto.
        (_mm_mask_cvtbf4_hf8): Ditto.
        (_mm_maskz_cvtbf4_hf8): Ditto.
        (_mm256_cvtbf4_hf8): Ditto.
        (_mm256_mask_cvtbf4_hf8): Ditto.
        (_mm256_maskz_cvtbf4_hf8): Ditto.
        (_mm512_cvtbf4_hf8): Ditto.
        (_mm512_mask_cvtbf4_hf8): Ditto.
        (_mm512_maskz_cvtbf4_hf8): Ditto.
        * config/i386/i386-builtin-types.def : New function types.
        * config/i386/i386-builtin.def (BDESC): Add builtins.
        * config/i386/i386-expand.cc (ix86_expand_args_builtin): Handle new 
function types.
        (ix86_expand_special_args_builtin): Ditto.
        * config/i386/sse.md (vcvt<convertfp82bf4>v16qi): New.
        (*vcvt<convertfp82bf4>v16qi): Ditto.
        (vcvt<convertfp82bf4><mode>): Ditto.
        (vcvt<convertfp82bf4><mode>_store): Ditto.
        (vcvtbf42hf8<mode><mask_name>): Ditto.

gcc/testsuite/ChangeLog:

        * gcc.target/i386/avx10v2aux-convert-1e.c: New test.
        * gcc.target/i386/avx10v2aux-convert-1f.c: New test.

Co-authored-by: Venkataramanan Kumar <[email protected]>
Co-authored-by: Haochen Jiang <[email protected]>

---
 gcc/config/i386/avx10v2auxintrin.h            | 181 ++++++++++++++++++
 gcc/config/i386/i386-builtin-types.def        |   7 +
 gcc/config/i386/i386-builtin.def              |  17 ++
 gcc/config/i386/i386-expand.cc                |   6 +
 gcc/config/i386/sse.md                        |  86 +++++++++
 .../gcc.target/i386/avx10v2aux-convert-1e.c   |  52 +++++
 .../gcc.target/i386/avx10v2aux-convert-1f.c   |  36 ++++
 7 files changed, 385 insertions(+)
 create mode 100644 gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1e.c
 create mode 100644 gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1f.c

diff --git a/gcc/config/i386/avx10v2auxintrin.h 
b/gcc/config/i386/avx10v2auxintrin.h
index a862f3173eb..bc004a24dfb 100644
--- a/gcc/config/i386/avx10v2auxintrin.h
+++ b/gcc/config/i386/avx10v2auxintrin.h
@@ -1195,6 +1195,187 @@ _mm512_maskz_cvthf8_ps (__mmask16 __U, __m128i __A)
                                                     (__mmask16) __U);
 }

+// VCVTBF82BF4S
+
+extern __inline __m128i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cvtbf8_bf4s (__m128i __A)
+{
+  return (__m128i) __builtin_ia32_vcvtbf82bf4s128 ((__v16qi) __A);
+}
+
+extern __inline __m128i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cvtbf8_bf4s (__m256i __A)
+{
+  return (__m128i) __builtin_ia32_vcvtbf82bf4s256 ((__v32qi) __A);
+}
+
+extern __inline __m256i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cvtbf8_bf4s (__m512i __A)
+{
+  return (__m256i) __builtin_ia32_vcvtbf82bf4s512 ((__v64qi) __A);
+}
+
+extern __inline void
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cvtbf8_bf4s_storeu (void * __P, __m128i __A)
+{
+  __builtin_ia32_vcvtbf82bf4s128mem ((__v8qi *) __P, (__v16qi) __A);
+}
+
+extern __inline void
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cvtbf8_bf4s_storeu (void * __P, __m256i __A)
+{
+  __builtin_ia32_vcvtbf82bf4s256mem ((__v16qi *) __P, (__v32qi) __A);
+}
+
+extern __inline void
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cvtbf8_bf4s_storeu (void * __P, __m512i __A)
+{
+  __builtin_ia32_vcvtbf82bf4s512mem ((__v32qi *) __P, (__v64qi) __A);
+}
+
+// VCVTHF82BF4S
+
+extern __inline __m128i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cvthf8_bf4s (__m128i __A)
+{
+  return (__m128i) __builtin_ia32_vcvthf82bf4s128 ((__v16qi) __A);
+}
+
+extern __inline __m128i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cvthf8_bf4s (__m256i __A)
+{
+  return (__m128i) __builtin_ia32_vcvthf82bf4s256 ((__v32qi) __A);
+}
+
+extern __inline __m256i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cvthf8_bf4s (__m512i __A)
+{
+  return (__m256i) __builtin_ia32_vcvthf82bf4s512 ((__v64qi) __A);
+}
+
+extern __inline void
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cvthf8_bf4s_storeu (void * __P, __m128i __A)
+{
+  __builtin_ia32_vcvthf82bf4s128mem ((__v8qi *) __P, (__v16qi) __A);
+}
+
+extern __inline void
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cvthf8_bf4s_storeu (void * __P, __m256i __A)
+{
+  __builtin_ia32_vcvthf82bf4s256mem ((__v16qi *) __P, (__v32qi) __A);
+}
+
+extern __inline void
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cvthf8_bf4s_storeu (void * __P, __m512i __A)
+{
+  __builtin_ia32_vcvthf82bf4s512mem ((__v32qi *) __P, (__v64qi) __A);
+}
+
+// VCVTBF42HF8 - 128-bit
+
+extern __inline __m128i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cvtbf4_hf8 (__m128i __A)
+{
+  return (__m128i) __builtin_ia32_vcvtbf42hf8128_mask ((__v16qi) __A,
+                                                      (__v16qi)
+                                                      _mm_undefined_si128 (),
+                                                      (__mmask16) -1);
+}
+
+extern __inline __m128i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cvtbf4_hf8 (__m128i __W, __mmask16 __U, __m128i __A)
+{
+  return (__m128i) __builtin_ia32_vcvtbf42hf8128_mask ((__v16qi) __A,
+                                                      (__v16qi) __W,
+                                                      (__mmask16) __U);
+}
+
+extern __inline __m128i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_maskz_cvtbf4_hf8 (__mmask16 __U, __m128i __A)
+{
+  return (__m128i) __builtin_ia32_vcvtbf42hf8128_mask ((__v16qi) __A,
+                                                      (__v16qi)
+                                                      _mm_setzero_si128 (),
+                                                      (__mmask16) __U);
+}
+
+// VCVTBF42HF8 - 256-bit
+
+extern __inline __m256i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cvtbf4_hf8 (__m128i __A)
+{
+  return (__m256i) __builtin_ia32_vcvtbf42hf8256_mask ((__v16qi) __A,
+                                                      (__v32qi)
+                                                      _mm256_undefined_si256 
(),
+                                                      (__mmask32) -1);
+}
+
+extern __inline __m256i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cvtbf4_hf8 (__m256i __W, __mmask32 __U, __m128i __A)
+{
+  return (__m256i) __builtin_ia32_vcvtbf42hf8256_mask ((__v16qi) __A,
+                                                      (__v32qi) __W,
+                                                      (__mmask32) __U);
+}
+
+extern __inline __m256i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_maskz_cvtbf4_hf8 (__mmask32 __U, __m128i __A)
+{
+  return (__m256i) __builtin_ia32_vcvtbf42hf8256_mask ((__v16qi) __A,
+                                                      (__v32qi)
+                                                      _mm256_setzero_si256 (),
+                                                      (__mmask32) __U);
+}
+
+// VCVTBF42HF8 - 512-bit
+
+extern __inline __m512i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cvtbf4_hf8 (__m256i __A)
+{
+  return (__m512i) __builtin_ia32_vcvtbf42hf8512_mask ((__v32qi) __A,
+                                                      (__v64qi)
+                                                      _mm512_undefined_si512 
(),
+                                                      (__mmask64) -1);
+}
+
+extern __inline __m512i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cvtbf4_hf8 (__m512i __W, __mmask64 __U, __m256i __A)
+{
+  return (__m512i) __builtin_ia32_vcvtbf42hf8512_mask ((__v32qi) __A,
+                                                      (__v64qi) __W,
+                                                      (__mmask64) __U);
+}
+
+extern __inline __m512i
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_maskz_cvtbf4_hf8 (__mmask64 __U, __m256i __A)
+{
+  return (__m512i) __builtin_ia32_vcvtbf42hf8512_mask ((__v32qi) __A,
+                                                      (__v64qi)
+                                                      _mm512_setzero_si512 (),
+                                                      (__mmask64) __U);
+}
+
 #ifdef __DISABLE_AVX10V2AUX__
 #undef __DISABLE_AVX10V2AUX__
 #pragma GCC pop_options
diff --git a/gcc/config/i386/i386-builtin-types.def 
b/gcc/config/i386/i386-builtin-types.def
index 20078ba295e..bc72594a124 100644
--- a/gcc/config/i386/i386-builtin-types.def
+++ b/gcc/config/i386/i386-builtin-types.def
@@ -1484,6 +1484,13 @@ DEF_FUNCTION_TYPE (V16QI, V16SI, V16SF, V16QI, UHI)
 DEF_FUNCTION_TYPE (V4SF, V16QI, V4SF, UQI)
 DEF_FUNCTION_TYPE (V8SF, V16QI, V8SF, UQI)
 DEF_FUNCTION_TYPE (V16SF, V16QI, V16SF, UHI)
+DEF_FUNCTION_TYPE (V16QI, V32QI)
+DEF_FUNCTION_TYPE (V32QI, V64QI)
+DEF_FUNCTION_TYPE (V64QI, V32QI, V64QI, UDI)
+DEF_FUNCTION_TYPE (VOID, PV8QI, V16QI)
+DEF_FUNCTION_TYPE (VOID, PV16QI, V32QI)
+DEF_FUNCTION_TYPE (VOID, PV32QI, V64QI)
+

 # SM4 builtins
 DEF_FUNCTION_TYPE (V16SI, V16SI, V16SI)
diff --git a/gcc/config/i386/i386-builtin.def b/gcc/config/i386/i386-builtin.def
index 07a80e28598..077565d3139 100644
--- a/gcc/config/i386/i386-builtin.def
+++ b/gcc/config/i386/i386-builtin.def
@@ -523,6 +523,14 @@ BDESC (OPTION_MASK_ISA_64BIT, OPTION_MASK_ISA2_MOVRS | 
OPTION_MASK_ISA2_AVX10_2,
 BDESC (OPTION_MASK_ISA_64BIT, OPTION_MASK_ISA2_MOVRS | 
OPTION_MASK_ISA2_AVX10_2, CODE_FOR_avx10_2_vmovrsqv2di_mask, 
"__builtin_ia32_vmovrsq128_mask", IX86_BUILTIN_VMOVRSQ_128, UNKNOWN, (int) 
V2DI_FTYPE_PCV2DI_V2DI_UQI)
 BDESC (OPTION_MASK_ISA_64BIT, OPTION_MASK_ISA2_MOVRS | 
OPTION_MASK_ISA2_AVX10_2, CODE_FOR_avx10_2_vmovrswv8hi_mask, 
"__builtin_ia32_vmovrsw128_mask", IX86_BUILTIN_VMOVRSW_128, UNKNOWN, (int) 
V8HI_FTYPE_PCV8HI_V8HI_UQI)

+/* AVX10V2AUX.  */
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf82bf4sv16qi_store, 
"__builtin_ia32_vcvtbf82bf4s128mem", IX86_BUILTIN_VCVTBF82BF4S128_MEM, UNKNOWN, 
(int) VOID_FTYPE_PV8QI_V16QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf82bf4sv32qi_store, 
"__builtin_ia32_vcvtbf82bf4s256mem", IX86_BUILTIN_VCVTBF82BF4S256_MEM, UNKNOWN, 
(int) VOID_FTYPE_PV16QI_V32QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf82bf4sv64qi_store, 
"__builtin_ia32_vcvtbf82bf4s512mem", IX86_BUILTIN_VCVTBF82BF4S512_MEM, UNKNOWN, 
(int) VOID_FTYPE_PV32QI_V64QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82bf4sv16qi_store, 
"__builtin_ia32_vcvthf82bf4s128mem", IX86_BUILTIN_VCVTHF82BF4S128_MEM, UNKNOWN, 
(int) VOID_FTYPE_PV8QI_V16QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82bf4sv32qi_store, 
"__builtin_ia32_vcvthf82bf4s256mem", IX86_BUILTIN_VCVTHF82BF4S256_MEM, UNKNOWN, 
(int) VOID_FTYPE_PV16QI_V32QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82bf4sv64qi_store, 
"__builtin_ia32_vcvthf82bf4s512mem", IX86_BUILTIN_VCVTHF82BF4S512_MEM, UNKNOWN, 
(int) VOID_FTYPE_PV32QI_V64QI)
+
 BDESC_END (SPECIAL_ARGS, PURE_ARGS)

 /* AVX */
@@ -3406,6 +3414,15 @@ BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, 
CODE_FOR_vcvtbf82psv16sf_mask, "__builtin
 BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82psv4sf_mask, 
"__builtin_ia32_vcvthf82ps128_mask", IX86_BUILTIN_VCVTHF82PS128_MASK, UNKNOWN, 
(int) V4SF_FTYPE_V16QI_V4SF_UQI)
 BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82psv8sf_mask, 
"__builtin_ia32_vcvthf82ps256_mask", IX86_BUILTIN_VCVTHF82PS256_MASK, UNKNOWN, 
(int) V8SF_FTYPE_V16QI_V8SF_UQI)
 BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82psv16sf_mask, 
"__builtin_ia32_vcvthf82ps512_mask", IX86_BUILTIN_VCVTHF82PS512_MASK, UNKNOWN, 
(int) V16SF_FTYPE_V16QI_V16SF_UHI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf82bf4sv16qi, 
"__builtin_ia32_vcvtbf82bf4s128", IX86_BUILTIN_VCVTBF82BF4S128, UNKNOWN, (int) 
V16QI_FTYPE_V16QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf82bf4sv32qi, 
"__builtin_ia32_vcvtbf82bf4s256", IX86_BUILTIN_VCVTBF82BF4S256, UNKNOWN, (int) 
V16QI_FTYPE_V32QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf82bf4sv64qi, 
"__builtin_ia32_vcvtbf82bf4s512", IX86_BUILTIN_VCVTBF82BF4S512, UNKNOWN, (int) 
V32QI_FTYPE_V64QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82bf4sv16qi, 
"__builtin_ia32_vcvthf82bf4s128", IX86_BUILTIN_VCVTHF82BF4S128, UNKNOWN, (int) 
V16QI_FTYPE_V16QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82bf4sv32qi, 
"__builtin_ia32_vcvthf82bf4s256", IX86_BUILTIN_VCVTHF82BF4S256, UNKNOWN, (int) 
V16QI_FTYPE_V32QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvthf82bf4sv64qi, 
"__builtin_ia32_vcvthf82bf4s512", IX86_BUILTIN_VCVTHF82BF4S512, UNKNOWN, (int) 
V32QI_FTYPE_V64QI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf42hf8v16qi_mask, 
"__builtin_ia32_vcvtbf42hf8128_mask", IX86_BUILTIN_VCVTBF42HF8128_MASK, 
UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI_UHI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf42hf8v32qi_mask, 
"__builtin_ia32_vcvtbf42hf8256_mask", IX86_BUILTIN_VCVTBF42HF8256_MASK, 
UNKNOWN, (int) V32QI_FTYPE_V16QI_V32QI_USI)
+BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX, CODE_FOR_vcvtbf42hf8v64qi_mask, 
"__builtin_ia32_vcvtbf42hf8512_mask", IX86_BUILTIN_VCVTBF42HF8512_MASK, 
UNKNOWN, (int) V64QI_FTYPE_V32QI_V64QI_UDI)

 /* Builtins with rounding support.  */
 BDESC_END (ARGS, ROUND_ARGS)
diff --git a/gcc/config/i386/i386-expand.cc b/gcc/config/i386/i386-expand.cc
index 511ab7e5733..aea2ac7b31a 100644
--- a/gcc/config/i386/i386-expand.cc
+++ b/gcc/config/i386/i386-expand.cc
@@ -12684,6 +12684,8 @@ ix86_expand_args_builtin (const struct 
builtin_description *d,
     case V16BF_FTYPE_V16SF:
     case V8BF_FTYPE_V8SF:
     case V8BF_FTYPE_V4SF:
+    case V16QI_FTYPE_V32QI:
+    case V32QI_FTYPE_V64QI:
       nargs = 1;
       break;
     case V4SF_FTYPE_V4SF_VEC_MERGE:
@@ -13064,6 +13066,7 @@ ix86_expand_args_builtin (const struct 
builtin_description *d,
     case V4SF_FTYPE_V16QI_V4SF_UQI:
     case V8SF_FTYPE_V16QI_V8SF_UQI:
     case V16SF_FTYPE_V16QI_V16SF_UHI:
+    case V64QI_FTYPE_V32QI_V64QI_UDI:
       nargs = 3;
       break;
     case V32QI_FTYPE_V32QI_V32QI_INT:
@@ -14324,6 +14327,9 @@ ix86_expand_special_args_builtin (const struct 
builtin_description *d,
     case VOID_FTYPE_PV8DI_V8DI:
     case VOID_FTYPE_PV4DI_V4DI:
     case VOID_FTYPE_PV2DI_V2DI:
+    case VOID_FTYPE_PV8QI_V16QI:
+    case VOID_FTYPE_PV16QI_V32QI:
+    case VOID_FTYPE_PV32QI_V64QI:
     case VOID_FTYPE_PCHAR_V32QI:
     case VOID_FTYPE_PCHAR_V16QI:
     case VOID_FTYPE_PFLOAT_V16SF:
diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index 1583c59db3b..dd6393cabf9 100644
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -272,6 +272,9 @@
   UNSPEC_VCVTBIASPS2HF8S
   UNSPEC_VCVTBF82PS
   UNSPEC_VCVTHF82PS
+  UNSPEC_VCVTBF82BF4S
+  UNSPEC_VCVTHF82BF4S
+  UNSPEC_VCVTBF42HF8
 ])

 (define_c_enum "unspecv" [
@@ -34380,3 +34383,86 @@
   "TARGET_AVX10V2AUX"
   "vcvt<convertfp82ps>\t{%1, 
%0<mask_operand2>|%0<mask_operand2>,%<iptrssebvec_3>1}"
   [(set_attr "prefix" "evex")])
+
+;; FP8 to FP4 converts (VCVTBF82BF4S, VCVTHF82BF4S) - no masking
+
+(define_int_iterator UNSPEC_CONVERTFP82BF4
+  [UNSPEC_VCVTBF82BF4S UNSPEC_VCVTHF82BF4S])
+
+(define_int_attr convertfp82bf4
+  [(UNSPEC_VCVTBF82BF4S "bf82bf4s")
+   (UNSPEC_VCVTHF82BF4S "hf82bf4s")])
+
+(define_mode_iterator FP82FP4_PART [V32QI V64QI])
+
+(define_mode_attr fp82fp4_dest
+  [(V32QI "V16QI") (V64QI "V32QI")])
+
+(define_mode_iterator FP82FP4_MEM [V16QI V32QI V64QI])
+
+(define_mode_attr fp82fp4_mem
+  [(V16QI "V8QI") (V32QI "V16QI") (V64QI "V32QI")])
+
+(define_expand "vcvt<convertfp82bf4>v16qi"
+  [(set (match_operand:V16QI 0 "register_operand")
+       (vec_concat:V16QI
+         (unspec:V8QI
+           [(match_operand:V16QI 1 "register_operand")]
+           UNSPEC_CONVERTFP82BF4)
+         (match_dup 2)))]
+  "TARGET_AVX10V2AUX"
+  "operands[2] = CONST0_RTX (V8QImode);")
+
+(define_insn "*vcvt<convertfp82bf4>v16qi"
+  [(set (match_operand:V16QI 0 "register_operand" "=v")
+       (vec_concat:V16QI
+         (unspec:V8QI
+           [(match_operand:V16QI 1 "register_operand" "v")]
+           UNSPEC_CONVERTFP82BF4)
+         (match_operand:V8QI 2 "const0_operand")))]
+  "TARGET_AVX10V2AUX"
+  "vcvt<convertfp82bf4>\t{%1, %0|%0, %1}"
+  [(set_attr "prefix" "evex")
+   (set_attr "mode" "TI")])
+
+(define_insn "vcvt<convertfp82bf4><mode>"
+  [(set (match_operand:<fp82fp4_dest> 0 "register_operand" "=v")
+       (unspec:<fp82fp4_dest>
+         [(match_operand:FP82FP4_PART 1 "register_operand" "v")]
+         UNSPEC_CONVERTFP82BF4))]
+  "TARGET_AVX10V2AUX"
+  "vcvt<convertfp82bf4>\t{%1, %0|%0, %1}"
+  [(set_attr "prefix" "evex")
+   (set_attr "mode" "<sseinsnmode>")])
+
+(define_insn "vcvt<convertfp82bf4><mode>_store"
+  [(set (match_operand:<fp82fp4_mem> 0 "memory_operand" "=m")
+       (unspec:<fp82fp4_mem>
+         [(match_operand:FP82FP4_MEM 1 "register_operand" "v")]
+         UNSPEC_CONVERTFP82BF4))]
+  "TARGET_AVX10V2AUX"
+  "vcvt<convertfp82bf4>\t{%1, %0|%0, %1}"
+  [(set_attr "type" "ssecvt")
+   (set_attr "memory" "store")
+   (set_attr "prefix" "evex")
+   (set_attr "mode" "<sseinsnmode>")])
+
+;; FP4 to FP8 converts (VCVTBF42HF8) with masking
+
+(define_mode_iterator FP42FP8_MODE [V16QI V32QI V64QI])
+
+(define_mode_attr fp82fp4_src
+  [(V16QI "V16QI") (V32QI "V16QI") (V64QI "V32QI")])
+
+(define_mode_attr iptr_fp42fp8
+  [(V16QI "q") (V32QI "") (V64QI "")])
+
+(define_insn "vcvtbf42hf8<mode><mask_name>"
+  [(set (match_operand:FP42FP8_MODE 0 "register_operand" "=v")
+       (unspec:FP42FP8_MODE
+         [(match_operand:<fp82fp4_src> 1 "nonimmediate_operand" "vm")]
+         UNSPEC_VCVTBF42HF8))]
+  "TARGET_AVX10V2AUX"
+  "vcvtbf42hf8\t{%1, %0<mask_operand2>|%0<mask_operand2>, %<iptr_fp42fp8>1}"
+  [(set_attr "prefix" "evex")
+   (set_attr "mode" "<sseinsnmode>")])
diff --git a/gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1e.c 
b/gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1e.c
new file mode 100644
index 00000000000..d1faab9899f
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1e.c
@@ -0,0 +1,52 @@
+/* { dg-do compile } */
+/* { dg-options "-mavx10v2aux -O2 -fno-fuse-ops-with-volatile-access" } */
+/* { dg-final { scan-assembler-times "vcvtbf82bf4s\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf82bf4s\[ 
\\t\]*%ymm\[0-9\]+,\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf82bf4s\[ 
\\t\]*%zmm\[0-9\]+,\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvthf82bf4s\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvthf82bf4s\[ 
\\t\]*%ymm\[0-9\]+,\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvthf82bf4s\[ 
\\t\]*%zmm\[0-9\]+,\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf82bf4s\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*\\(%\[a-z0-9\]+\\)(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf82bf4s\[ 
\\t\]*%ymm\[0-9\]+,\[^\{\n\]*\\(%\[a-z0-9\]+\\)(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf82bf4s\[ 
\\t\]*%zmm\[0-9\]+,\[^\{\n\]*\\(%\[a-z0-9\]+\\)(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvthf82bf4s\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*\\(%\[a-z0-9\]+\\)(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvthf82bf4s\[ 
\\t\]*%ymm\[0-9\]+,\[^\{\n\]*\\(%\[a-z0-9\]+\\)(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvthf82bf4s\[ 
\\t\]*%zmm\[0-9\]+,\[^\{\n\]*\\(%\[a-z0-9\]+\\)(?:\n|\[ \\t\]+#)" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m128i x128i;
+volatile __m256i x256i;
+volatile __m512i x512i;
+
+void extern
+avx10v2aux_vcvtbf82bf4s_test (void)
+{
+  x128i = _mm_cvtbf8_bf4s (x128i);
+  x128i = _mm256_cvtbf8_bf4s (x256i);
+  x256i = _mm512_cvtbf8_bf4s (x512i);
+}
+
+void extern
+avx10v2aux_vcvthf82bf4s_test (void)
+{
+  x128i = _mm_cvthf8_bf4s (x128i);
+  x128i = _mm256_cvthf8_bf4s (x256i);
+  x256i = _mm512_cvthf8_bf4s (x512i);
+}
+
+void extern
+avx10v2aux_vcvtbf82bf4s_store_test (void *p)
+{
+  _mm_cvtbf8_bf4s_storeu (p, x128i);
+  _mm256_cvtbf8_bf4s_storeu (p, x256i);
+  _mm512_cvtbf8_bf4s_storeu (p, x512i);
+}
+
+void extern
+avx10v2aux_vcvthf82bf4s_store_test (void *p)
+{
+  _mm_cvthf8_bf4s_storeu (p, x128i);
+  _mm256_cvthf8_bf4s_storeu (p, x256i);
+  _mm512_cvthf8_bf4s_storeu (p, x512i);
+}
diff --git a/gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1f.c 
b/gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1f.c
new file mode 100644
index 00000000000..afed68f8ec6
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/avx10v2aux-convert-1f.c
@@ -0,0 +1,36 @@
+/* { dg-do compile } */
+/* { dg-options "-mavx10v2aux -O2 -fno-fuse-ops-with-volatile-access" } */
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } 
*/
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 
} } */
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%ymm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } 
*/
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%xmm\[0-9\]+,\[^\{\n\]*%ymm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 
} } */
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%ymm\[0-9\]+,\[^\{\n\]*%zmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%ymm\[0-9\]+,\[^\{\n\]*%zmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } 
*/
+/* { dg-final { scan-assembler-times "vcvtbf42hf8\[ 
\\t\]*%ymm\[0-9\]+,\[^\{\n\]*%zmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 
} } */
+
+#include <immintrin.h>
+
+volatile __m128i x128i;
+volatile __m256i x256i;
+volatile __m512i x512i;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx10v2aux_vcvtbf42hf8_test (void)
+{
+  x128i = _mm_cvtbf4_hf8 (x128i);
+  x128i = _mm_mask_cvtbf4_hf8 (x128i, m16, x128i);
+  x128i = _mm_maskz_cvtbf4_hf8 (m16, x128i);
+
+  x256i = _mm256_cvtbf4_hf8 (x128i);
+  x256i = _mm256_mask_cvtbf4_hf8 (x256i, m32, x128i);
+  x256i = _mm256_maskz_cvtbf4_hf8 (m32, x128i);
+
+  x512i = _mm512_cvtbf4_hf8 (x256i);
+  x512i = _mm512_mask_cvtbf4_hf8 (x512i, m64, x256i);
+  x512i = _mm512_maskz_cvtbf4_hf8 (m64, x256i);
+}
--
2.34.1

Reply via email to