mirror of
https://github.com/kovidgoyal/kitty
synced 2026-07-11 18:32:12 +02:00
More work on SIMD utf8 decode
This commit is contained in:
Kovid Goyal
@@ -28,6 +28,7 @@ _Pragma("clang diagnostic pop")
|
||||
|
||||
#if BITS == 128
|
||||
#define set1_epi8(x) simde_mm_set1_epi8((char)(x))
|
||||
#define set_epi8 simde_mm_set_epi8
|
||||
#define add_epi8 simde_mm_add_epi8
|
||||
#define load_unaligned simde_mm_loadu_si128
|
||||
#define store_aligned simde_mm_store_si128
|
||||
@@ -40,6 +41,14 @@ _Pragma("clang diagnostic pop")
|
||||
#define extract_lower_quarter_as_chars simde_mm_cvtepu8_epi32
|
||||
#define shift_right_by_one_byte(x) simde_mm_slli_si128(x, 1)
|
||||
#define shift_right_by_two_bytes(x) simde_mm_slli_si128(x, 2)
|
||||
#define shift_right_by_four_bytes(x) simde_mm_slli_si128(x, 4)
|
||||
#define shift_right_by_eight_bytes(x) simde_mm_slli_si128(x, 8)
|
||||
#define shift_right_by_sixteen_bytes(x) simde_mm_slli_si128(x, 16)
|
||||
#define shift_left_by_one_byte(x) simde_mm_srli_si128(x, 1)
|
||||
#define shift_left_by_two_bytes(x) simde_mm_srli_si128(x, 2)
|
||||
#define shift_left_by_four_bytes(x) simde_mm_srli_si128(x, 4)
|
||||
#define shift_left_by_eight_bytes(x) simde_mm_srli_si128(x, 8)
|
||||
#define shift_left_by_sixteen_bytes(x) simde_mm_srli_si128(x, 16)
|
||||
#define blendv_epi8 simde_mm_blendv_epi8
|
||||
#define shift_left_by_bits16 _mm_slli_epi16
|
||||
#define shift_right_by_bits32 _mm_srli_epi32
|
||||
@@ -48,6 +57,7 @@ _Pragma("clang diagnostic pop")
|
||||
#define create_zero_integer _mm_setzero_si128
|
||||
#else
|
||||
#define set1_epi8(x) simde_mm256_set1_epi8((char)(x))
|
||||
#define set_epi8 simde_mm256_set_epi8
|
||||
#define add_epi8 simde_mm256_add_epi8
|
||||
#define load_unaligned simde_mm256_loadu_si256
|
||||
#define store_aligned simde_mm256_store_si256
|
||||
@@ -63,8 +73,16 @@ _Pragma("clang diagnostic pop")
|
||||
#define shift_left_by_bits16 _mm256_slli_epi16
|
||||
#define shift_right_by_bits32 _mm256_srli_epi32
|
||||
#define create_zero_integer _mm256_setzero_si256
|
||||
#define shift_right_by_one_byte(x) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(0, 0, 2, 0)), 16 - 1)
|
||||
#define shift_right_by_two_bytes(x) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(0, 0, 2, 0)), 16 - 2)
|
||||
#define shift_right_by_one_byte(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(0, 0, 2, 0)), 16 - 1)
|
||||
#define shift_right_by_two_bytes(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(0, 0, 2, 0)), 16 - 2)
|
||||
#define shift_right_by_four_bytes(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(0, 0, 2, 0)), 16 - 4)
|
||||
#define shift_right_by_eight_bytes(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(0, 0, 2, 0)), 16 - 8)
|
||||
#define shift_right_by_sixteen_bytes(vec) _mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(0, 0, 2, 0))
|
||||
#define shift_left_by_one_byte(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(2, 0, 0, 1)), 1)
|
||||
#define shift_left_by_two_bytes(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(2, 0, 0, 1)), 2)
|
||||
#define shift_left_by_four_bytes(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(2, 0, 0, 1)), 4)
|
||||
#define shift_left_by_eight_bytes(vec) simde_mm256_alignr_epi8(vec, simde_mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(2, 0, 0, 1)), 8)
|
||||
#define shift_left_by_sixteen_bytes(vec) _mm256_permute2x128_si256(vec, vec, _MM_SHUFFLE(2, 0, 0, 1))
|
||||
|
||||
#endif
|
||||
|
||||
@@ -135,6 +153,7 @@ FUNC(output_plain_ascii)(UTF8Decoder *d, integer_t vec, size_t src_sz) {
|
||||
|
||||
static inline bool
|
||||
FUNC(utf8_decode_to_esc)(UTF8Decoder *d, const uint8_t *src, size_t src_sz) {
|
||||
// Based on the algorithm described in: https://woboq.com/blog/utf-8-processing-using-simd.html
|
||||
d->output_sz = 0; d->num_consumed = 0;
|
||||
src_sz = MIN(src_sz, sizeof(integer_t));
|
||||
integer_t vec = load_unaligned((integer_t*)src);
|
||||
@@ -222,6 +241,41 @@ FUNC(utf8_decode_to_esc)(UTF8Decoder *d, const uint8_t *src, size_t src_sz) {
|
||||
output3 = and_si(output3, count3_locations); // keep only count3 bytes
|
||||
print_register_as_bytes(output3);
|
||||
|
||||
// Shuffle bytes to remove continuation bytes
|
||||
integer_t shifts = subtract_saturate_epu8(count, one); // number of bytes we need to skip for each UTF-8 sequence
|
||||
// propagate the shifts to all subsequent bytes by shift and add
|
||||
shifts = add_epi8(shifts, shift_right_by_one_byte(shifts));
|
||||
shifts = add_epi8(shifts, shift_right_by_two_bytes(shifts));
|
||||
shifts = add_epi8(shifts, shift_right_by_four_bytes(shifts));
|
||||
shifts = add_epi8(shifts, shift_right_by_eight_bytes(shifts));
|
||||
#if BITS == 256
|
||||
shifts = add_epi8(shifts, shift_right_by_sixteen_bytes(shifts));
|
||||
#endif
|
||||
// zero the shifts for discarded continuation bytes
|
||||
shifts = and_si(shifts, cmplt_epi8(counts, two));
|
||||
// now we need to convert shifts into a mask for the shuffle. The mask has each byte of the
|
||||
// form 0000xxxx the lower four bits indicating the destination location for the byte. For 256 bit shuffle we use lower 5 bits.
|
||||
// First we move the numbers in shifts to discard the unwanted UTF-8 sequence bytes. We note that the numbers
|
||||
// are bounded by sizeof(integer_t) and so we need at most 4 (for 128 bit) or 5 (for 256 bit) moves. The numbers are
|
||||
// monotonic from left to right and change value only at the end of a UTF-8 sequence. We move them leftwards, accumulating the
|
||||
// moves bit-by-bit.
|
||||
//
|
||||
#define move(shifts, amt, which_bit) blendv_epi8(shifts, shift_left_by_##amt(shifts), shift_left_by_##amt(shift_left_by_bits16(shifts, 8 - which_bit)))
|
||||
shifts = move(shifts, one_byte, 1);
|
||||
shifts = move(shifts, two_bytes, 2);
|
||||
shifts = move(shifts, four_bytes, 3);
|
||||
shifts = move(shifts, eight_bytes, 4);
|
||||
#if BITS == 256
|
||||
shifts = move(shifts, sixteen_bytes, 5);
|
||||
#endif
|
||||
#undef move
|
||||
// convert the shifts into a suitable mask for shuffle by adding the byte number to each byte
|
||||
#if BITS == 128
|
||||
shifts = add_epi8(shifts, set_epi8(15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0));
|
||||
#else
|
||||
shifts = add_epi8(shifts, set_epi8(31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0));
|
||||
#endif
|
||||
print_register_as_bytes(shifts);
|
||||
return sentinel_found;
|
||||
}
|
||||
|
||||
@@ -229,6 +283,7 @@ FUNC(utf8_decode_to_esc)(UTF8Decoder *d, const uint8_t *src, size_t src_sz) {
|
||||
#undef FUNC
|
||||
#undef integer_t
|
||||
#undef set1_epi8
|
||||
#undef set_epi8
|
||||
#undef load_unaligned
|
||||
#undef store_aligned
|
||||
#undef cmpeq_epi8
|
||||
@@ -242,6 +297,14 @@ FUNC(utf8_decode_to_esc)(UTF8Decoder *d, const uint8_t *src, size_t src_sz) {
|
||||
#undef BITS
|
||||
#undef shift_right_by_one_byte
|
||||
#undef shift_right_by_two_bytes
|
||||
#undef shift_right_by_four_bytes
|
||||
#undef shift_right_by_eight_bytes
|
||||
#undef shift_right_by_sixteen_bytes
|
||||
#undef shift_left_by_one_byte
|
||||
#undef shift_left_by_two_bytes
|
||||
#undef shift_left_by_four_bytes
|
||||
#undef shift_left_by_eight_bytes
|
||||
#undef shift_left_by_sixteen_bytes
|
||||
#undef shift_left_by_bits16
|
||||
#undef shift_right_by_bits32
|
||||
#undef shift_right_by_bytes128
|
||||
|
||||
Reference in New Issue
Block a user