Skip to content

S
simde
Commit 848fb777 authored by LI Qingwu's avatar LI Qingwu Committed by Michael R. Crusoe
Browse files
Options

arm neon: fix arm64 gcc11 build excess elements in vector failure 

Resolved a build failure on ARM64 with GCC 11 caused by excess
elements in a vector initializer. The issue stemmed from the
`__builtin_shuffle` function, where the number of elements in
the argument vector(s) and the mask vector must match.

For more details, refer to issue #1211.
Signed-off-by: default avatarLI Qingwu <Qing-wu.Li@leica-geosystems.com.cn>
parent 0aaf7829
Hide whitespace changes
Inline Side-by-side
Showing with 40 additions and 40 deletions
arm/neon/cadd_rot270.h
View file @ 848fb777
...@@ -54,7 +54,7 @@ simde_float16x4_t simde_vcadd_rot270_f16(simde_float16x4_t a, simde_float16x4_t ...@@ -54,7 +54,7 @@ simde_float16x4_t simde_vcadd_rot270_f16(simde_float16x4_t a, simde_float16x4_t
r_.sv64 = __riscv_vfadd_vv_f16m1(op1, a_.sv64, 4); r_.sv64 = __riscv_vfadd_vv_f16m1(op1, a_.sv64, 4);
#elif defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #elif defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 5, 0, 7, 2); b_.values = SIMDE_SHUFFLE_VECTOR_(16, 8, -b_.values, b_.values, 5, 0, 7, 2);
r_.values = b_.values + a_.values; r_.values = b_.values + a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -90,7 +90,7 @@ simde_float16x8_t simde_vcaddq_rot270_f16(simde_float16x8_t a, simde_float16x8_t ...@@ -90,7 +90,7 @@ simde_float16x8_t simde_vcaddq_rot270_f16(simde_float16x8_t a, simde_float16x8_t
r_.sv128 = __riscv_vfadd_vv_f16m1(op1, a_.sv128, 8); r_.sv128 = __riscv_vfadd_vv_f16m1(op1, a_.sv128, 8);
#elif defined(SIMDE_SHUFFLE_VECTOR_) && \ #elif defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 8, -b_.values, b_.values, 9, 0, 11, 2, 13, 4, 15, 6); b_.values = SIMDE_SHUFFLE_VECTOR_(16, 16, -b_.values, b_.values, 9, 0, 11, 2, 13, 4, 15, 6);
r_.values = b_.values + a_.values; r_.values = b_.values + a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
......
arm/neon/cadd_rot90.h
View file @ 848fb777
...@@ -54,7 +54,7 @@ simde_float16x4_t simde_vcadd_rot90_f16(simde_float16x4_t a, simde_float16x4_t b ...@@ -54,7 +54,7 @@ simde_float16x4_t simde_vcadd_rot90_f16(simde_float16x4_t a, simde_float16x4_t b
r_.sv64 = __riscv_vfadd_vv_f16m1(op1, a_.sv64, 4); r_.sv64 = __riscv_vfadd_vv_f16m1(op1, a_.sv64, 4);
#elif defined(SIMDE_SHUFFLE_VECTOR_) && \ #elif defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 1, 4, 3, 6); b_.values = SIMDE_SHUFFLE_VECTOR_(16, 8, -b_.values, b_.values, 1, 4, 3, 6);
r_.values = b_.values + a_.values; r_.values = b_.values + a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -90,7 +90,7 @@ simde_float16x8_t simde_vcaddq_rot90_f16(simde_float16x8_t a, simde_float16x8_t ...@@ -90,7 +90,7 @@ simde_float16x8_t simde_vcaddq_rot90_f16(simde_float16x8_t a, simde_float16x8_t
r_.sv128 = __riscv_vfadd_vv_f16m1(op1, a_.sv128, 8); r_.sv128 = __riscv_vfadd_vv_f16m1(op1, a_.sv128, 8);
#elif defined(SIMDE_SHUFFLE_VECTOR_) && \ #elif defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 8, -b_.values, b_.values, 1, 8, 3, 10, 5, 12, 7, 14); b_.values = SIMDE_SHUFFLE_VECTOR_(16, 16, -b_.values, b_.values, 1, 8, 3, 10, 5, 12, 7, 14);
r_.values = b_.values + a_.values; r_.values = b_.values + a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
......
arm/neon/cmla_lane.h
View file @ 848fb777
...@@ -59,7 +59,7 @@ simde_float16x4_t simde_vcmla_lane_f16(simde_float16x4_t r, simde_float16x4_t a, ...@@ -59,7 +59,7 @@ simde_float16x4_t simde_vcmla_lane_f16(simde_float16x4_t r, simde_float16x4_t a,
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 0, 0, 2, 2); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 0, 0, 2, 2);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -142,7 +142,7 @@ simde_float16x4_t simde_vcmla_laneq_f16(simde_float16x4_t r, simde_float16x4_t a ...@@ -142,7 +142,7 @@ simde_float16x4_t simde_vcmla_laneq_f16(simde_float16x4_t r, simde_float16x4_t a
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 0, 0, 2, 2); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 0, 0, 2, 2);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -228,8 +228,8 @@ simde_float16x8_t simde_vcmlaq_lane_f16(simde_float16x8_t r, simde_float16x8_t a ...@@ -228,8 +228,8 @@ simde_float16x8_t simde_vcmlaq_lane_f16(simde_float16x8_t r, simde_float16x8_t a
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 0, 0, 2, 2); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 0, 0, 2, 2);
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 0, 0, 2, 2); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 0, 0, 2, 2);
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
#else #else
...@@ -320,9 +320,9 @@ simde_float16x8_t simde_vcmlaq_laneq_f16(simde_float16x8_t r, simde_float16x8_t ...@@ -320,9 +320,9 @@ simde_float16x8_t simde_vcmlaq_laneq_f16(simde_float16x8_t r, simde_float16x8_t
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 0, 0, 2, 2); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 0, 0, 2, 2);
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 0, 0, 2, 2); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 0, 0, 2, 2);
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
......
arm/neon/cmla_rot180_lane.h
View file @ 848fb777
...@@ -62,8 +62,8 @@ simde_float16x4_t simde_vcmla_rot180_lane_f16(simde_float16x4_t r, simde_float16 ...@@ -62,8 +62,8 @@ simde_float16x4_t simde_vcmla_rot180_lane_f16(simde_float16x4_t r, simde_float16
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 0, 0, 2, 2); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 0, 0, 2, 2);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 0, 1, 2, 3); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 0, 1, 2, 3);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -151,9 +151,9 @@ simde_float16x8_t simde_vcmlaq_rot180_lane_f16(simde_float16x8_t r, simde_float1 ...@@ -151,9 +151,9 @@ simde_float16x8_t simde_vcmlaq_rot180_lane_f16(simde_float16x8_t r, simde_float1
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 0, 0, 2, 2); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 0, 0, 2, 2);
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 0, 0, 2, 2); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 0, 0, 2, 2);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 0, 1, 2, 3); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 0, 1, 2, 3);
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
#else #else
...@@ -243,8 +243,8 @@ simde_float16x4_t simde_vcmla_rot180_laneq_f16(simde_float16x4_t r, simde_float1 ...@@ -243,8 +243,8 @@ simde_float16x4_t simde_vcmla_rot180_laneq_f16(simde_float16x4_t r, simde_float1
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 0, 0, 2, 2); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 0, 0, 2, 2);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 0, 1, 2, 3); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 0, 1, 2, 3);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -332,9 +332,9 @@ simde_float16x8_t simde_vcmlaq_rot180_laneq_f16(simde_float16x8_t r, simde_float ...@@ -332,9 +332,9 @@ simde_float16x8_t simde_vcmlaq_rot180_laneq_f16(simde_float16x8_t r, simde_float
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 0, 0, 2, 2); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 0, 0, 2, 2);
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 0, 0, 2, 2); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 0, 0, 2, 2);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 0, 1, 2, 3); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 0, 1, 2, 3);
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
#else #else
......
arm/neon/cmla_rot270_lane.h
View file @ 848fb777
...@@ -62,8 +62,8 @@ simde_float16x4_t simde_vcmla_rot270_lane_f16(simde_float16x4_t r, simde_float16 ...@@ -62,8 +62,8 @@ simde_float16x4_t simde_vcmla_rot270_lane_f16(simde_float16x4_t r, simde_float16
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 1, 1, 3, 3); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 5, 0, 7, 2); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 5, 0, 7, 2);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -151,9 +151,9 @@ simde_float16x8_t simde_vcmlaq_rot270_lane_f16(simde_float16x8_t r, simde_float1 ...@@ -151,9 +151,9 @@ simde_float16x8_t simde_vcmlaq_rot270_lane_f16(simde_float16x8_t r, simde_float1
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 1, 1, 3, 3); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 1, 1, 3, 3);
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 1, 1, 3, 3); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 5, 0, 7, 2); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 5, 0, 7, 2);
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
#else #else
...@@ -243,8 +243,8 @@ simde_float16x4_t simde_vcmla_rot270_laneq_f16(simde_float16x4_t r, simde_float1 ...@@ -243,8 +243,8 @@ simde_float16x4_t simde_vcmla_rot270_laneq_f16(simde_float16x4_t r, simde_float1
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 1, 1, 3, 3); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 5, 0, 7, 2); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 5, 0, 7, 2);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -332,9 +332,9 @@ simde_float16x8_t simde_vcmlaq_rot270_laneq_f16(simde_float16x8_t r, simde_float ...@@ -332,9 +332,9 @@ simde_float16x8_t simde_vcmlaq_rot270_laneq_f16(simde_float16x8_t r, simde_float
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 1, 1, 3, 3); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 1, 1, 3, 3);
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 1, 1, 3, 3); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 5, 0, 7, 2); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 5, 0, 7, 2);
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
#else #else
......
arm/neon/cmla_rot90_lane.h
View file @ 848fb777
...@@ -62,8 +62,8 @@ simde_float16x4_t simde_vcmla_rot90_lane_f16(simde_float16x4_t r, simde_float16x ...@@ -62,8 +62,8 @@ simde_float16x4_t simde_vcmla_rot90_lane_f16(simde_float16x4_t r, simde_float16x
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 1, 1, 3, 3); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 1, 4, 3, 6); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 1, 4, 3, 6);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -147,8 +147,8 @@ simde_float16x4_t simde_vcmla_rot90_laneq_f16(simde_float16x4_t r, simde_float16 ...@@ -147,8 +147,8 @@ simde_float16x4_t simde_vcmla_rot90_laneq_f16(simde_float16x4_t r, simde_float16
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_.values, a_.values, 1, 1, 3, 3); a_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.values, a_.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 1, 4, 3, 6); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 1, 4, 3, 6);
r_.values += b_.values * a_.values; r_.values += b_.values * a_.values;
#else #else
SIMDE_VECTORIZE SIMDE_VECTORIZE
...@@ -237,9 +237,9 @@ simde_float16x8_t simde_vcmlaq_rot90_lane_f16(simde_float16x8_t r, simde_float16 ...@@ -237,9 +237,9 @@ simde_float16x8_t simde_vcmlaq_rot90_lane_f16(simde_float16x8_t r, simde_float16
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x4_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_BUG_GCC_100760) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 1, 1, 3, 3); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 1, 1, 3, 3);
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 1, 1, 3, 3); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 1, 4, 3, 6); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 1, 4, 3, 6);
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
#else #else
...@@ -333,9 +333,9 @@ simde_float16x8_t simde_vcmlaq_rot90_laneq_f16(simde_float16x8_t r, simde_float1 ...@@ -333,9 +333,9 @@ simde_float16x8_t simde_vcmlaq_rot90_laneq_f16(simde_float16x8_t r, simde_float1
simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane]))); simde_vcvt_f32_f16(simde_vdup_n_f16(simde_float16x8_to_private(b).values[lane])));
#if defined(SIMDE_SHUFFLE_VECTOR_) && \ #if defined(SIMDE_SHUFFLE_VECTOR_) && \
((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16)) ((SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FP16) || (SIMDE_FLOAT16_API == SIMDE_FLOAT16_API_FLOAT16))
a_low.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_low.values, a_low.values, 1, 1, 3, 3); a_low.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_low.values, a_low.values, 1, 1, 3, 3);
a_high.values = SIMDE_SHUFFLE_VECTOR_(16, 4, a_high.values, a_high.values, 1, 1, 3, 3); a_high.values = SIMDE_SHUFFLE_VECTOR_(32, 16, a_high.values, a_high.values, 1, 1, 3, 3);
b_.values = SIMDE_SHUFFLE_VECTOR_(16, 4, -b_.values, b_.values, 1, 4, 3, 6); b_.values = SIMDE_SHUFFLE_VECTOR_(32, 16, -b_.values, b_.values, 1, 4, 3, 6);
r_low.values += b_.values * a_low.values; r_low.values += b_.values * a_low.values;
r_high.values += b_.values * a_high.values; r_high.values += b_.values * a_high.values;
#else #else
......
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