Skip to content

S
simde
Commit 2ec1f51f authored by Michael R. Crusoe's avatar Michael R. Crusoe Committed by Michael R. Crusoe
Browse files
Options

pow: consistently use simde_math_pow

parent 80f65573
Hide whitespace changes
Inline Side-by-side
Showing with 55 additions and 55 deletions
arm/neon/cvt_n.h
View file @ 2ec1f51f
......@@ -40,7 +40,7 @@ simde_vcvth_n_u16_f16(simde_float16_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 16) {
return simde_vcvth_u16_f16(
simde_float16_from_float32(
simde_float16_to_float32(a) * HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n))));
simde_float16_to_float32(a) * HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n))));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARM_NEON_FP16)
#define simde_vcvth_n_u16_f16(a, n) vcvth_n_u16_f16(a, n)
......@@ -56,7 +56,7 @@ simde_vcvth_n_f16_s16(int16_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 16) {
return simde_float16_from_float32(
HEDLEY_STATIC_CAST(simde_float32_t,
HEDLEY_STATIC_CAST(simde_float64_t, a) / pow(2, n)));
HEDLEY_STATIC_CAST(simde_float64_t, a) / simde_math_pow(2, n)));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARM_NEON_FP16)
#define simde_vcvth_n_f16_s16(a, n) vcvth_n_f16_s16(a, n)
......@@ -72,7 +72,7 @@ simde_vcvth_n_f16_u16(uint16_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 16) {
return simde_float16_from_float32(
HEDLEY_STATIC_CAST(simde_float32_t,
HEDLEY_STATIC_CAST(simde_float64_t, a) / pow(2, n)));
HEDLEY_STATIC_CAST(simde_float64_t, a) / simde_math_pow(2, n)));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARM_NEON_FP16)
#define simde_vcvth_n_f16_u16(a, n) vcvth_n_f16_u16(a, n)
......@@ -86,7 +86,7 @@ SIMDE_FUNCTION_ATTRIBUTES
int32_t
simde_vcvts_n_s32_f32(simde_float32_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 32) {
return simde_vcvts_s32_f32(a * HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n)));
return simde_vcvts_s32_f32(a * HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n)));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvts_n_s32_f32(a, n) vcvts_n_s32_f32(a, n)
......@@ -100,7 +100,7 @@ SIMDE_FUNCTION_ATTRIBUTES
uint32_t
simde_vcvts_n_u32_f32(simde_float32_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 32) {
return simde_vcvts_u32_f32(a * HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n)));
return simde_vcvts_u32_f32(a * HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n)));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvts_n_u32_f32(a, n) vcvts_n_u32_f32(a, n)
......@@ -115,7 +115,7 @@ simde_float32_t
simde_vcvts_n_f32_s32(int32_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 32) {
return HEDLEY_STATIC_CAST(simde_float32_t,
HEDLEY_STATIC_CAST(simde_float64_t, a) / pow(2, n));
HEDLEY_STATIC_CAST(simde_float64_t, a) / simde_math_pow(2, n));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvts_n_f32_s32(a, n) vcvts_n_f32_s32(a, n)
......@@ -130,7 +130,7 @@ simde_float32_t
simde_vcvts_n_f32_u32(uint32_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 32) {
return HEDLEY_STATIC_CAST(simde_float32_t,
HEDLEY_STATIC_CAST(simde_float64_t, a) / pow(2, n));
HEDLEY_STATIC_CAST(simde_float64_t, a) / simde_math_pow(2, n));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvts_n_f32_u32(a, n) vcvts_n_f32_u32(a, n)
......@@ -144,7 +144,7 @@ SIMDE_FUNCTION_ATTRIBUTES
int64_t
simde_vcvtd_n_s64_f64(simde_float64_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 64) {
return simde_vcvtd_s64_f64(a * pow(2, n));
return simde_vcvtd_s64_f64(a * simde_math_pow(2, n));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvtd_n_s64_f64(a, n) vcvtd_n_s64_f64(a, n)
......@@ -158,7 +158,7 @@ SIMDE_FUNCTION_ATTRIBUTES
uint64_t
simde_vcvtd_n_u64_f64(simde_float64_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 64) {
return simde_vcvtd_u64_f64(a * pow(2, n));
return simde_vcvtd_u64_f64(a * simde_math_pow(2, n));
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvtd_n_u64_f64(a, n) vcvtd_n_u64_f64(a, n)
......@@ -172,7 +172,7 @@ SIMDE_FUNCTION_ATTRIBUTES
simde_float64_t
simde_vcvtd_n_f64_s64(int64_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 64) {
return HEDLEY_STATIC_CAST(simde_float64_t, a) / pow(2, n);
return HEDLEY_STATIC_CAST(simde_float64_t, a) / simde_math_pow(2, n);
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvtd_n_f64_s64(a, n) vcvtd_n_f64_s64(a, n)
......@@ -186,7 +186,7 @@ SIMDE_FUNCTION_ATTRIBUTES
simde_float64_t
simde_vcvtd_n_f64_u64(uint64_t a, const int n)
SIMDE_REQUIRE_CONSTANT_RANGE(n, 1, 64) {
return HEDLEY_STATIC_CAST(simde_float64_t, a) / pow(2, n);
return HEDLEY_STATIC_CAST(simde_float64_t, a) / simde_math_pow(2, n);
}
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
#define simde_vcvtd_n_f64_u64(a, n) vcvtd_n_f64_u64(a, n)
......@@ -205,7 +205,7 @@ simde_vcvt_n_s32_f32(simde_float32x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvts_s32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n)));
r_.values[i] = simde_vcvts_s32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n)));
}
return simde_int32x2_from_private(r_);
......@@ -227,7 +227,7 @@ simde_vcvt_n_s64_f64(simde_float64x1_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvtd_s64_f64(a_.values[i] * pow(2, n));
r_.values[i] = simde_vcvtd_s64_f64(a_.values[i] * simde_math_pow(2, n));
}
return simde_int64x1_from_private(r_);
......@@ -251,7 +251,7 @@ simde_vcvt_n_u16_f16(simde_float16x4_t a, const int n)
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvth_u16_f16(simde_float16_from_float32(
simde_float16_to_float32(a_.values[i]) *
HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n))));
HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n))));
}
return simde_uint16x4_from_private(r_);
......@@ -273,7 +273,7 @@ simde_vcvt_n_u32_f32(simde_float32x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvts_u32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n)));
r_.values[i] = simde_vcvts_u32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n)));
}
return simde_uint32x2_from_private(r_);
......@@ -295,7 +295,7 @@ simde_vcvt_n_u64_f64(simde_float64x1_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvtd_u64_f64(a_.values[i] * pow(2, n));
r_.values[i] = simde_vcvtd_u64_f64(a_.values[i] * simde_math_pow(2, n));
}
return simde_uint64x1_from_private(r_);
......@@ -317,7 +317,7 @@ simde_vcvtq_n_s32_f32(simde_float32x4_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvts_s32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n)));
r_.values[i] = simde_vcvts_s32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n)));
}
return simde_int32x4_from_private(r_);
......@@ -339,7 +339,7 @@ simde_vcvtq_n_s64_f64(simde_float64x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvtd_s64_f64(a_.values[i] * pow(2, n));
r_.values[i] = simde_vcvtd_s64_f64(a_.values[i] * simde_math_pow(2, n));
}
return simde_int64x2_from_private(r_);
......@@ -363,7 +363,7 @@ simde_vcvtq_n_u16_f16(simde_float16x8_t a, const int n)
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvth_u16_f16(simde_float16_from_float32(
simde_float16_to_float32(a_.values[i]) *
HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n))));
HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n))));
}
return simde_uint16x8_from_private(r_);
......@@ -385,7 +385,7 @@ simde_vcvtq_n_u32_f32(simde_float32x4_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvts_u32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, pow(2, n)));
r_.values[i] = simde_vcvts_u32_f32(a_.values[i] * HEDLEY_STATIC_CAST(simde_float32_t, simde_math_pow(2, n)));
}
return simde_uint32x4_from_private(r_);
......@@ -407,7 +407,7 @@ simde_vcvtq_n_u64_f64(simde_float64x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_vcvtd_u64_f64(a_.values[i] * pow(2, n));
r_.values[i] = simde_vcvtd_u64_f64(a_.values[i] * simde_math_pow(2, n));
}
return simde_uint64x2_from_private(r_);
......@@ -429,7 +429,7 @@ simde_vcvt_n_f16_u16(simde_uint16x4_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n)));
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n)));
}
return simde_float16x4_from_private(r_);
......@@ -451,7 +451,7 @@ simde_vcvt_n_f16_s16(simde_int16x4_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n)));
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n)));
}
return simde_float16x4_from_private(r_);
......@@ -473,7 +473,7 @@ simde_vcvtq_n_f16_u16(simde_uint16x8_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n)));
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n)));
}
return simde_float16x8_from_private(r_);
......@@ -495,7 +495,7 @@ simde_vcvtq_n_f16_s16(simde_int16x8_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32_t, (a_.values[i] / pow(2, n))));
r_.values[i] = simde_float16_from_float32(HEDLEY_STATIC_CAST(simde_float32_t, (a_.values[i] / simde_math_pow(2, n))));
}
return simde_float16x8_from_private(r_);
......@@ -517,7 +517,7 @@ simde_vcvt_n_f32_u32(simde_uint32x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float32x2_from_private(r_);
......@@ -539,7 +539,7 @@ simde_vcvt_n_f32_s32(simde_int32x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float32x2_from_private(r_);
......@@ -561,7 +561,7 @@ simde_vcvt_n_f64_u64(simde_uint64x1_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float64x1_from_private(r_);
......@@ -583,7 +583,7 @@ simde_vcvtq_n_f64_u64(simde_uint64x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float64x2_from_private(r_);
......@@ -605,7 +605,7 @@ simde_vcvt_n_f64_s64(simde_int64x1_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float64x1_from_private(r_);
......@@ -627,7 +627,7 @@ simde_vcvtq_n_f64_s64(simde_int64x2_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float64_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float64x2_from_private(r_);
......@@ -649,7 +649,7 @@ simde_vcvtq_n_f32_s32(simde_int32x4_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float32x4_from_private(r_);
......@@ -671,7 +671,7 @@ simde_vcvtq_n_f32_u32(simde_uint32x4_t a, const int n)
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / pow(2, n));
r_.values[i] = HEDLEY_STATIC_CAST(simde_float32_t, HEDLEY_STATIC_CAST(simde_float64_t, a_.values[i]) / simde_math_pow(2, n));
}
return simde_float32x4_from_private(r_);
......
simde-math.h
View file @ 2ec1f51f
......@@ -952,16 +952,36 @@ simde_math_fpclass(double v, const int imm8) {
#endif
#endif
#if !defined(simde_math_pow)
#if SIMDE_MATH_BUILTIN_LIBM(pow)
#define simde_math_pow(y, x) __builtin_pow(y, x)
#elif defined(SIMDE_MATH_HAVE_CMATH)
#define simde_math_pow(y, x) std::pow(y, x)
#elif defined(SIMDE_MATH_HAVE_MATH_H)
#define simde_math_pow(y, x) pow(y, x)
#endif
#endif
#if !defined(simde_math_powf)
#if SIMDE_MATH_BUILTIN_LIBM(powf)
#define simde_math_powf(y, x) __builtin_powf(y, x)
#elif defined(SIMDE_MATH_HAVE_CMATH)
#define simde_math_powf(y, x) std::pow(y, x)
#elif defined(SIMDE_MATH_HAVE_MATH_H)
#define simde_math_powf(y, x) powf(y, x)
#endif
#endif
#if HEDLEY_HAS_BUILTIN(__builtin_exp10) || HEDLEY_GCC_VERSION_CHECK(3,4,0)
# define simde_math_exp10(v) __builtin_exp10(v)
#else
# define simde_math_exp10(v) pow(10.0, (v))
# define simde_math_exp10(v) simde_math_pow(10.0, (v))
#endif
#if HEDLEY_HAS_BUILTIN(__builtin_exp10f) || HEDLEY_GCC_VERSION_CHECK(3,4,0)
# define simde_math_exp10f(v) __builtin_exp10f(v)
#else
# define simde_math_exp10f(v) powf(10.0f, (v))
# define simde_math_exp10f(v) simde_math_powf(10.0f, (v))
#endif
#if !defined(simde_math_fabs)
......@@ -1204,26 +1224,6 @@ simde_math_fpclass(double v, const int imm8) {
#endif
#endif
#if !defined(simde_math_pow)
#if SIMDE_MATH_BUILTIN_LIBM(pow)
#define simde_math_pow(y, x) __builtin_pow(y, x)
#elif defined(SIMDE_MATH_HAVE_CMATH)
#define simde_math_pow(y, x) std::pow(y, x)
#elif defined(SIMDE_MATH_HAVE_MATH_H)
#define simde_math_pow(y, x) pow(y, x)
#endif
#endif
#if !defined(simde_math_powf)
#if SIMDE_MATH_BUILTIN_LIBM(powf)
#define simde_math_powf(y, x) __builtin_powf(y, x)
#elif defined(SIMDE_MATH_HAVE_CMATH)
#define simde_math_powf(y, x) std::pow(y, x)
#elif defined(SIMDE_MATH_HAVE_MATH_H)
#define simde_math_powf(y, x) powf(y, x)
#endif
#endif
#if !defined(simde_math_rint)
#if SIMDE_MATH_BUILTIN_LIBM(rint)
#define simde_math_rint(v) __builtin_rint(v)
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment