Add std/adler32 hasher.up_arm_neon

On a Raspberry Pi 4 (32-bit armv7l) with -march=native and -mfpu=neon
("native" means "armv8-a+crc+simd"):

wuffs_adler32_10k/clang9                                  1.21GB/s ± 0%   6.58GB/s ± 1%  +442.43%  (p=0.008 n=5+5)
wuffs_adler32_100k/clang9                                 1.20GB/s ± 0%   6.58GB/s ± 0%  +449.37%  (p=0.008 n=5+5)

wuffs_adler32_10k/gcc8                                    1.24GB/s ± 0%   4.31GB/s ± 0%  +248.87%  (p=0.008 n=5+5)
wuffs_adler32_100k/gcc8                                   1.22GB/s ± 0%   4.29GB/s ± 0%  +251.09%  (p=0.008 n=5+5)

mimic_adler32_10k                                         1.40GB/s ± 0%   1.40GB/s ± 0%    -0.04%  (p=0.008 n=5+5)
mimic_adler32_100k                                        1.35GB/s ± 0%   1.38GB/s ± 0%    +2.77%  (p=0.008 n=5+5)

wuffs_zlib_decode_10k/clang9                              85.3MB/s ± 0%   91.1MB/s ± 0%    +6.83%  (p=0.008 n=5+5)
wuffs_zlib_decode_100k/clang9                              107MB/s ± 0%    113MB/s ± 0%    +5.20%  (p=0.008 n=5+5)

wuffs_zlib_decode_10k/gcc8                                86.1MB/s ± 0%   91.0MB/s ± 0%    +5.69%  (p=0.008 n=5+5)
wuffs_zlib_decode_100k/gcc8                                104MB/s ± 0%    111MB/s ± 0%    +7.08%  (p=0.016 n=5+4)

mimic_zlib_decode_10k                                      108MB/s ± 0%    108MB/s ± 0%      ~     (p=0.548 n=5+5)
mimic_zlib_decode_100k                                     142MB/s ± 0%    141MB/s ± 1%      ~     (p=0.310 n=5+5)

wuffs_png_decode_image_19k_8bpp/clang9                    69.9MB/s ± 0%   74.3MB/s ± 0%    +6.37%  (p=0.016 n=5+4)
wuffs_png_decode_image_40k_24bpp/clang9                   99.4MB/s ± 0%  104.3MB/s ± 0%    +4.92%  (p=0.008 n=5+5)
wuffs_png_decode_image_77k_8bpp/clang9                     266MB/s ± 0%    280MB/s ± 0%    +5.21%  (p=0.008 n=5+5)
wuffs_png_decode_image_552k_32bpp_ignore_checksum/clang9   165MB/s ± 0%    169MB/s ± 0%    +2.45%  (p=0.008 n=5+5)
wuffs_png_decode_image_552k_32bpp_verify_checksum/clang9   146MB/s ± 0%    163MB/s ± 0%   +11.86%  (p=0.008 n=5+5)
wuffs_png_decode_image_4002k_24bpp/clang9                 93.2MB/s ± 0%   97.7MB/s ± 0%    +4.78%  (p=0.008 n=5+5)

wuffs_png_decode_image_19k_8bpp/gcc8                      70.3MB/s ± 0%   74.3MB/s ± 0%    +5.70%  (p=0.008 n=5+5)
wuffs_png_decode_image_40k_24bpp/gcc8                      101MB/s ± 0%    107MB/s ± 0%    +5.67%  (p=0.008 n=5+5)
wuffs_png_decode_image_77k_8bpp/gcc8                       247MB/s ± 0%    259MB/s ± 0%    +4.71%  (p=0.008 n=5+5)
wuffs_png_decode_image_552k_32bpp_ignore_checksum/gcc8     162MB/s ± 0%    162MB/s ± 0%    +0.09%  (p=0.032 n=5+4)
wuffs_png_decode_image_552k_32bpp_verify_checksum/gcc8     140MB/s ± 0%    155MB/s ± 0%   +10.67%  (p=0.008 n=5+5)
wuffs_png_decode_image_4002k_24bpp/gcc8                   94.1MB/s ± 0%   99.4MB/s ± 0%    +5.57%  (p=0.008 n=5+5)

mimic_png_decode_image_19k_8bpp                           43.6MB/s ± 0%   43.3MB/s ± 1%      ~     (p=0.167 n=5+5)
mimic_png_decode_image_40k_24bpp                          54.6MB/s ± 0%   54.6MB/s ± 0%      ~     (p=1.000 n=5+5)
mimic_png_decode_image_77k_8bpp                            125MB/s ± 0%    123MB/s ± 0%    -1.03%  (p=0.016 n=5+4)
mimic_png_decode_image_552k_32bpp_ignore_checksum               skipped
mimic_png_decode_image_552k_32bpp_verify_checksum          102MB/s ± 0%    101MB/s ± 0%    -0.43%  (p=0.008 n=5+5)
mimic_png_decode_image_4002k_24bpp                        81.3MB/s ± 0%   81.8MB/s ± 0%    +0.62%  (p=0.008 n=5+5)
This commit is contained in:
Nigel Tao committed 2021-02-16 22:17:46 +11:00
1 parent d13039d96e
commit 010b34ed38
7 files changed
+629 -50

No files matched your search

+114
View File
@@ -17665,6 +17665,13 @@ wuffs_adler32__hasher__up_sse42(
wuffs_base__slice_u8 a_x);
#endif // defined(WUFFS_BASE__CPU_ARCH__X86_64)
#if defined(WUFFS_BASE__CPU_ARCH__ARM_NEON)
static wuffs_base__empty_struct
wuffs_adler32__hasher__up_arm_neon(
wuffs_adler32__hasher* self,
wuffs_base__slice_u8 a_x);
#endif // defined(WUFFS_BASE__CPU_ARCH__ARM_NEON)
// ---------------- VTables
const wuffs_base__hasher_u32__func_ptrs
@@ -17776,6 +17783,9 @@ wuffs_adler32__hasher__update_u32(
self->private_impl.f_started = true;
self->private_impl.f_state = 1;
self->private_impl.choosy_up = (
#if defined(WUFFS_BASE__CPU_ARCH__ARM_NEON)
wuffs_base__cpu_arch__have_arm_neon() ? &wuffs_adler32__hasher__up_arm_neon :
#endif
#if defined(WUFFS_BASE__CPU_ARCH__X86_64)
wuffs_base__cpu_arch__have_sse42() ? &wuffs_adler32__hasher__up_sse42 :
#endif
@@ -17951,6 +17961,110 @@ wuffs_adler32__hasher__up_sse42(
}
#endif // defined(WUFFS_BASE__CPU_ARCH__X86_64)
// -------- func adler32.hasher.up_arm_neon
#if defined(WUFFS_BASE__CPU_ARCH__ARM_NEON)
static wuffs_base__empty_struct
wuffs_adler32__hasher__up_arm_neon(
wuffs_adler32__hasher* self,
wuffs_base__slice_u8 a_x) {
uint32_t v_s1 = 0;
uint32_t v_s2 = 0;
wuffs_base__slice_u8 v_remaining = {0};
wuffs_base__slice_u8 v_p = {0};
uint8x16_t v_p__left = {0};
uint8x16_t v_p_right = {0};
uint8x16_t v_v1 = {0};
uint8x16_t v_v2 = {0};
uint8x16_t v_col0 = {0};
uint8x16_t v_col1 = {0};
uint8x16_t v_col2 = {0};
uint8x16_t v_col3 = {0};
uint8x8_t v_sum1 = {0};
uint8x8_t v_sum2 = {0};
uint8x8_t v_sum12 = {0};
uint32_t v_num_iterate_bytes = 0;
uint64_t v_tail_index = 0;
v_s1 = ((self->private_impl.f_state) & 0xFFFF);
v_s2 = ((self->private_impl.f_state) >> (32 - (16)));
while ((((uint64_t)(a_x.len)) > 0) && ((15 & ((uint32_t)(0xFFF & (uintptr_t)(a_x.ptr)))) != 0)) {
v_s1 += ((uint32_t)(a_x.ptr[0]));
v_s2 += v_s1;
a_x = wuffs_base__slice_u8__subslice_i(a_x, 1);
}
v_s1 %= 65521;
v_s2 %= 65521;
while (((uint64_t)(a_x.len)) > 0) {
v_remaining = wuffs_base__slice_u8__subslice_j(a_x, 0);
if (((uint64_t)(a_x.len)) > 5536) {
v_remaining = wuffs_base__slice_u8__subslice_i(a_x, 5536);
a_x = wuffs_base__slice_u8__subslice_j(a_x, 5536);
}
v_num_iterate_bytes = ((uint32_t)((((uint64_t)(a_x.len)) & 4294967264)));
v_s2 += ((uint32_t)(v_s1 * v_num_iterate_bytes));
v_v1 = vreinterpretq_u8_u32(vdupq_n_u32(0));
v_v2 = vreinterpretq_u8_u32(vdupq_n_u32(0));
v_col0 = vreinterpretq_u8_u16(vdupq_n_u16(0));
v_col1 = vreinterpretq_u8_u16(vdupq_n_u16(0));
v_col2 = vreinterpretq_u8_u16(vdupq_n_u16(0));
v_col3 = vreinterpretq_u8_u16(vdupq_n_u16(0));
{
wuffs_base__slice_u8 i_slice_p = a_x;
v_p.ptr = i_slice_p.ptr;
v_p.len = 32;
uint8_t* i_end0_p = v_p.ptr + (((i_slice_p.len - (size_t)(v_p.ptr - i_slice_p.ptr)) / 32) * 32);
while (v_p.ptr < i_end0_p) {
v_p__left = vld1q_u8(wuffs_base__slice_u8__subslice_j(v_p, 16).ptr);
v_p_right = vld1q_u8(wuffs_base__slice_u8__subslice_ij(v_p, 16, 32).ptr);
v_v2 = vreinterpretq_u8_u32(vaddq_u32(vreinterpretq_u32_u8(v_v2), vreinterpretq_u32_u8(v_v1)));
v_v1 = vreinterpretq_u8_u32(vpadalq_u16(vreinterpretq_u32_u8(v_v1), vreinterpretq_u16_u8(vreinterpretq_u8_u16(vpadalq_u8(vreinterpretq_u16_u8(vreinterpretq_u8_u16(vpaddlq_u8(v_p__left))), v_p_right)))));
v_col0 = vreinterpretq_u8_u16(vaddw_u8(vreinterpretq_u16_u8(v_col0), vget_low_u8(v_p__left)));
v_col1 = vreinterpretq_u8_u16(vaddw_u8(vreinterpretq_u16_u8(v_col1), vget_high_u8(v_p__left)));
v_col2 = vreinterpretq_u8_u16(vaddw_u8(vreinterpretq_u16_u8(v_col2), vget_low_u8(v_p_right)));
v_col3 = vreinterpretq_u8_u16(vaddw_u8(vreinterpretq_u16_u8(v_col3), vget_high_u8(v_p_right)));
v_p.ptr += 32;
}
v_p.len = 0;
}
v_v2 = vreinterpretq_u8_u32(vshlq_n_u32(vreinterpretq_u32_u8(v_v2), 5));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_low_u16(vreinterpretq_u16_u8(v_col0)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){32, 31, 30, 29}))));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_high_u16(vreinterpretq_u16_u8(v_col0)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){28, 27, 26, 25}))));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_low_u16(vreinterpretq_u16_u8(v_col1)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){24, 23, 22, 21}))));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_high_u16(vreinterpretq_u16_u8(v_col1)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){20, 19, 18, 17}))));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_low_u16(vreinterpretq_u16_u8(v_col2)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){16, 15, 14, 13}))));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_high_u16(vreinterpretq_u16_u8(v_col2)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){12, 11, 10, 9}))));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_low_u16(vreinterpretq_u16_u8(v_col3)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){8, 7, 6, 5}))));
v_v2 = vreinterpretq_u8_u32(vmlal_u16(vreinterpretq_u32_u8(v_v2), vreinterpret_u16_u8(vreinterpret_u8_u16(vget_high_u16(vreinterpretq_u16_u8(v_col3)))), vreinterpret_u16_u8(vreinterpret_u8_u16((uint16x4_t){4, 3, 2, 1}))));
v_sum1 = vreinterpret_u8_u32(vpadd_u32(vreinterpret_u32_u8(vreinterpret_u8_u32(vget_low_u32(vreinterpretq_u32_u8(v_v1)))), vreinterpret_u32_u8(vreinterpret_u8_u32(vget_high_u32(vreinterpretq_u32_u8(v_v1))))));
v_sum2 = vreinterpret_u8_u32(vpadd_u32(vreinterpret_u32_u8(vreinterpret_u8_u32(vget_low_u32(vreinterpretq_u32_u8(v_v2)))), vreinterpret_u32_u8(vreinterpret_u8_u32(vget_high_u32(vreinterpretq_u32_u8(v_v2))))));
v_sum12 = vreinterpret_u8_u32(vpadd_u32(vreinterpret_u32_u8(v_sum1), vreinterpret_u32_u8(v_sum2)));
v_s1 += vget_lane_u32(vreinterpret_u32_u8(v_sum12), 0);
v_s2 += vget_lane_u32(vreinterpret_u32_u8(v_sum12), 1);
v_tail_index = (((uint64_t)(a_x.len)) & 18446744073709551584u);
if (v_tail_index < ((uint64_t)(a_x.len))) {
{
wuffs_base__slice_u8 i_slice_p = wuffs_base__slice_u8__subslice_i(a_x, v_tail_index);
v_p.ptr = i_slice_p.ptr;
v_p.len = 1;
uint8_t* i_end0_p = i_slice_p.ptr + i_slice_p.len;
while (v_p.ptr < i_end0_p) {
v_s1 += ((uint32_t)(v_p.ptr[0]));
v_s2 += v_s1;
v_p.ptr += 1;
}
v_p.len = 0;
}
}
v_s1 %= 65521;
v_s2 %= 65521;
a_x = v_remaining;
}
self->private_impl.f_state = (((v_s2 & 65535) << 16) | (v_s1 & 65535));
return wuffs_base__make_empty_struct();
}
#endif // defined(WUFFS_BASE__CPU_ARCH__ARM_NEON)
#endif // !defined(WUFFS_CONFIG__MODULES) || defined(WUFFS_CONFIG__MODULE__ADLER32)
#if !defined(WUFFS_CONFIG__MODULES) || defined(WUFFS_CONFIG__MODULE__BMP)