mirror of
https://github.com/GerbilSoft/zlib-ng.git
synced 2025-06-18 11:35:35 -04:00
50e9ca06e2
So a lot of alterations had to be done to make this not worse and so far, it's not really better, either. I had to force inlining for the adler routine, I had to remove the x4 load instruction otherwise pipelining stalled, and I had to use restrict pointers with a copy idiom for GCC to inline a copy routine for the tail. Still, we see a small benefit in benchmarks, particularly when done with size of our window or larger. There's also an added benefit that this will fix #1824.
70 lines
2.0 KiB
C
70 lines
2.0 KiB
C
#ifndef ARM_NEON_INTRINS_H
|
|
#define ARM_NEON_INTRINS_H
|
|
|
|
#if defined(_MSC_VER) && (defined(_M_ARM64) || defined(_M_ARM64EC))
|
|
/* arm64_neon.h is MSVC specific */
|
|
# include <arm64_neon.h>
|
|
#else
|
|
# include <arm_neon.h>
|
|
#endif
|
|
|
|
#if defined(ARM_NEON) && !defined(__aarch64__) && !defined(_M_ARM64) && !defined(_M_ARM64EC)
|
|
/* Compatibility shim for the _high family of functions */
|
|
#define vmull_high_u8(a, b) vmull_u8(vget_high_u8(a), vget_high_u8(b))
|
|
#define vmlal_high_u8(a, b, c) vmlal_u8(a, vget_high_u8(b), vget_high_u8(c))
|
|
#define vmlal_high_u16(a, b, c) vmlal_u16(a, vget_high_u16(b), vget_high_u16(c))
|
|
#define vaddw_high_u8(a, b) vaddw_u8(a, vget_high_u8(b))
|
|
#endif
|
|
|
|
#ifdef ARM_NEON
|
|
|
|
#define vqsubq_u16_x4_x1(out, a, b) do { \
|
|
out.val[0] = vqsubq_u16(a.val[0], b); \
|
|
out.val[1] = vqsubq_u16(a.val[1], b); \
|
|
out.val[2] = vqsubq_u16(a.val[2], b); \
|
|
out.val[3] = vqsubq_u16(a.val[3], b); \
|
|
} while (0)
|
|
|
|
#if defined(TARGET_OS_IPHONE) && TARGET_OS_IPHONE == 1
|
|
# define IS_IOS
|
|
#endif
|
|
|
|
# if defined(__arm__) && (defined(__ANDROID__) || defined(IS_IOS))
|
|
/* Android & iOS have too strict alignment requirement (:256) for x4 NEON intrinsics */
|
|
# undef ARM_NEON_HASLD4
|
|
# undef vld1q_u16_x4
|
|
# undef vld1q_u8_x4
|
|
# undef vst1q_u16_x4
|
|
# endif
|
|
|
|
# ifndef ARM_NEON_HASLD4
|
|
|
|
static inline uint16x8x4_t vld1q_u16_x4(uint16_t const *a) {
|
|
uint16x8x4_t ret;
|
|
ret.val[0] = vld1q_u16(a);
|
|
ret.val[1] = vld1q_u16(a+8);
|
|
ret.val[2] = vld1q_u16(a+16);
|
|
ret.val[3] = vld1q_u16(a+24);
|
|
return ret;
|
|
}
|
|
|
|
static inline uint8x16x4_t vld1q_u8_x4(uint8_t const *a) {
|
|
uint8x16x4_t ret;
|
|
ret.val[0] = vld1q_u8(a);
|
|
ret.val[1] = vld1q_u8(a+16);
|
|
ret.val[2] = vld1q_u8(a+32);
|
|
ret.val[3] = vld1q_u8(a+48);
|
|
return ret;
|
|
}
|
|
|
|
static inline void vst1q_u16_x4(uint16_t *p, uint16x8x4_t a) {
|
|
vst1q_u16(p, a.val[0]);
|
|
vst1q_u16(p + 8, a.val[1]);
|
|
vst1q_u16(p + 16, a.val[2]);
|
|
vst1q_u16(p + 24, a.val[3]);
|
|
}
|
|
# endif // HASLD4 check
|
|
#endif
|
|
|
|
#endif // include guard ARM_NEON_INTRINS_H
|