FazBrowse GitHub Viewer | Trending | | Home

Tools: [Download Repo ZIP] [Original HTTPS Page]

draft: Port `GetPointerToFirstInvalidByte` from SimdUnicode by BenjaminBucher · Pull Request #974 · simdutf/simdutf · GitHub

draft: Port GetPointerToFirstInvalidByte from SimdUnicode - #974

Draft

BenjaminBucher wants to merge 50 commits into

simdutf:masterfrom

BenjaminBucher:simdunicode-port

Commits on Aug 7, 2026

signature file
Benjamin Bucher
committed
8905e63
generate function stubs
Benjamin Bucher
committed
9484e81
initial scalar implementation
Benjamin Bucher
committed
ad86aaa
use scalar implementation everywhere for now
Benjamin Bucher
committed
d52723d
add brute force test
Benjamin Bucher
committed
517286d
fix for scalar
Benjamin Bucher
committed
cac418d
rework: validate_utf8_with_counts and utf8_result
Benjamin Bucher
committed
db14864
Revert "use scalar implementation everywhere for now"
Benjamin Bucher
committed
20571c3
Revert "generate function stubs"
Benjamin Bucher
committed
919d586
update signature file
Benjamin Bucher
committed
be218f9
Add function stubs and use scalar implementation
Benjamin Bucher
committed
aaed67d
fix test and remove last reference to former sig
Benjamin Bucher
committed
4137f82
wip generic implementation
Benjamin Bucher
committed
e901b35
Fix test and couple compile errors
Benjamin Bucher
committed
85179c4
fix compile error
Benjamin Bucher
committed
77938fa
Revert "Add function stubs and use scalar implementation"
Benjamin Bucher
committed
b4a83ee
Add stubs with scalar implementation again
Benjamin Bucher
committed
c5ac565
Use generic implementation in haswell, westmere
Benjamin Bucher
committed
9b25081
properly update counts on rewind
Benjamin Bucher
committed
02db66f
Fix rewind function
Benjamin Bucher
committed
8f239df
fix buffer underflow if rewind was called on first chunk
Benjamin Bucher
committed
3a38c72
explicitly make char signed in negative comparison
Benjamin Bucher
committed
76c0e40
use int8_t instead of signed char
Benjamin Bucher
committed
e1521b6
non-ascii counter (untested)
Benjamin Bucher
committed
4ed8444
utf16_length method for utf8_result
Benjamin Bucher
committed
a631262
proposed fix for PR 974.

lemire
authored and
Benjamin Bucher
committed
30cb780
extend validate_utf8_with_counts test
Benjamin Bucher
committed
4842286
fix rewind_and_validate_with_counts with non_ascii val
Benjamin Bucher
committed
4ba35d5
validate_utf8_with_counts icelake implementation
Benjamin Bucher
committed
8de3c86
benchmarks: add validate_utf8_with_counts
Benjamin Bucher
committed
adb424e
remove sig file
Benjamin Bucher
committed
c8b69c3
remove outdated comment
Benjamin Bucher
committed
1247d1a
use size_t for counters instead of uint64_t
Benjamin Bucher
committed
39617f1
fix undefined behavior in icelake implementation
Benjamin Bucher
committed
c38672b
document and export validate_utf8_with_counts

lemire
authored and
Benjamin Bucher
committed
15fe159
lsx, lasx, ppc64: use the generic SIMD validate_utf8_with_counts

lemire
authored and
Benjamin Bucher
committed
a1f920d
tests: sweep validate_utf8_with_counts errors across block boundaries

lemire
authored and
Benjamin Bucher
committed
bc831ee
icelake+scalar+generic: ASCII fast-skip counting, drop unused non_ascii_count

lemire
authored and
Benjamin Bucher
committed
444451b
arm64: use generic NEON validate_utf8_with_counts instead of scalar

lemire
authored and
Benjamin Bucher
committed
e9af504
generic/arm64: count via horizontal byte-sum on NEON instead of movemask+popcount

lemire
authored and
Benjamin Bucher
committed
cdf1efd
arm64/NEON: count with deferred cross-lane reduction (SimdUnicode trick)

lemire
authored and
Benjamin Bucher
committed
735fdde
generic: return ASCII bool from check_next_input; use it in NEON counter

lemire
authored and
Benjamin Bucher
committed
8bbfb52
clang-format
Benjamin Bucher
committed
59686e6
run scripts/reflow_readme.py
Benjamin Bucher
committed
54c7439

Commits on Aug 13, 2026

constexpr branches
Benjamin Bucher
committed
940af9b
uint64 mask(s)
Benjamin Bucher
committed
866eee3
ascii skip path
Benjamin Bucher
committed
d5205cc
try to add span
Benjamin Bucher
committed
1a0d822
rename scalar::utf8::validate{_utf8,}_with_counts
Benjamin Bucher
committed
1b789d5
fix last renames
Benjamin Bucher
committed
5a52d91

Back | FazBrowse Home | New Git URL