diff options
Diffstat (limited to 'src/core/NEON/wrapper/intrinsics/cvt.h')
| -rw-r--r-- | src/core/NEON/wrapper/intrinsics/cvt.h | 61 |
1 files changed, 36 insertions, 25 deletions
diff --git a/src/core/NEON/wrapper/intrinsics/cvt.h b/src/core/NEON/wrapper/intrinsics/cvt.h index 6e79a92bc2..381de2284a 100644 --- a/src/core/NEON/wrapper/intrinsics/cvt.h +++ b/src/core/NEON/wrapper/intrinsics/cvt.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2020 Arm Limited. + * Copyright (c) 2020, 2022-2023 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -30,12 +30,11 @@ namespace arm_compute { namespace wrapper { -#define VCVT_TO_F32_IMPL(ptype, vtype, prefix, postfix1, postfix2) \ - template <typename T> \ - inline typename std::enable_if<std::is_same<T, float>::value, float32x4_t>::type \ - vcvt(const vtype &a) \ - { \ - return prefix##_##postfix1##_##postfix2(a); \ +#define VCVT_TO_F32_IMPL(ptype, vtype, prefix, postfix1, postfix2) \ + template <typename T> \ + inline typename std::enable_if<std::is_same<T, float>::value, float32x4_t>::type vcvt(const vtype &a) \ + { \ + return prefix##_##postfix1##_##postfix2(a); \ } VCVT_TO_F32_IMPL(float32x4_t, uint32x4_t, vcvtq, f32, u32) @@ -46,12 +45,11 @@ VCVT_TO_F32_IMPL(float32x4_t, float16x4_t, vcvt, f32, f16) #undef VCVT_TO_F32_IMPL #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC -#define VCVT_TO_F16_IMPL(ptype, vtype, prefix, postfix1, postfix2) \ - template <typename T> \ - inline typename std::enable_if<std::is_same<T, float16_t>::value, float16x4_t>::type \ - vcvt(const vtype &a) \ - { \ - return prefix##_##postfix1##_##postfix2(a); \ +#define VCVT_TO_F16_IMPL(ptype, vtype, prefix, postfix1, postfix2) \ + template <typename T> \ + inline typename std::enable_if<std::is_same<T, float16_t>::value, float16x4_t>::type vcvt(const vtype &a) \ + { \ + return prefix##_##postfix1##_##postfix2(a); \ } VCVT_TO_F16_IMPL(float16x4_t, float32x4_t, vcvt, f16, f32) @@ -59,20 +57,34 @@ VCVT_TO_F16_IMPL(float16x4_t, float32x4_t, vcvt, f16, f32) #endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC template <typename T> -inline typename std::enable_if<std::is_same<T, uint8_t>::value, uint32x4_t>::type +inline typename std::enable_if<std::is_same<T, uint8_t>::value || std::is_same<T, uint32_t>::value, uint32x4_t>::type vcvt(const float32x4_t &a) { return vcvtq_u32_f32(a); } template <typename T> -inline typename std::enable_if<std::is_same<T, int8_t>::value, int32x4_t>::type +inline typename std::enable_if<std::is_same<T, int8_t>::value || std::is_same<T, int32_t>::value, int32x4_t>::type vcvt(const float32x4_t &a) { return vcvtq_s32_f32(a); } -#if defined(__ARM_FEATURE_BF16_VECTOR_ARITHMETIC) || defined(ARM_COMPUTE_FORCE_BF16) +#ifdef __aarch64__ +template <typename T> +inline typename std::enable_if<std::is_same<T, uint32_t>::value, uint32x4_t>::type vcvta(const float32x4_t &a) +{ + return vcvtaq_u32_f32(a); +} + +template <typename T> +inline typename std::enable_if<std::is_same<T, int32_t>::value, int32x4_t>::type vcvta(const float32x4_t &a) +{ + return vcvtaq_s32_f32(a); +} +#endif //__aarch64__ + +#if defined(ARM_COMPUTE_ENABLE_BF16) /** Convert 2x128-bit floating point vectors into 1x128-bit bfloat16 vector * * @param[in] inptr Pointer to the input memory to load values from @@ -80,16 +92,15 @@ vcvt(const float32x4_t &a) */ inline void vcvt_bf16_f32(const float *inptr, uint16_t *outptr) { - __asm __volatile( - "ldp q0, q1, [%[inptr]]\n" - ".inst 0xea16800\n" // BFCVTN v0, v0 - ".inst 0x4ea16820\n" // BFCVTN2 v0, v1 - "str q0, [%[outptr]]\n" - : [inptr] "+r"(inptr) - : [outptr] "r"(outptr) - : "v0", "v1", "memory"); + __asm __volatile("ldp q0, q1, [%[inptr]]\n" + ".inst 0xea16800\n" // BFCVTN v0, v0 + ".inst 0x4ea16820\n" // BFCVTN2 v0, v1 + "str q0, [%[outptr]]\n" + : [inptr] "+r"(inptr) + : [outptr] "r"(outptr) + : "v0", "v1", "memory"); } -#endif /* defined(__ARM_FEATURE_BF16_VECTOR_ARITHMETIC) || defined(ARM_COMPUTE_FORCE_BF16) */ +#endif /* defined(ARM_COMPUTE_ENABLE_BF16) */ } // namespace wrapper } // namespace arm_compute |