Scan the strings of json_view with NEON and SSE2

Long runs of string bytes are scanned 16 at a time with NEON (AArch64, with
GCC and Clang) and SSE2 (x86-64): both belong to the baseline instruction
sets. A signed compare with 0x20 finds control characters and non-ASCII
bytes at once. Keys keep 16 table checks before the vector loop (their
lengths repeat from record to record, so the branches predict well);
string values have 8, as their lengths vary more.

Non-ASCII text is validated 16 bytes at a time with the "lookup4" check of
simdjson (J. Keiser and D. Lemire, "Validating UTF-8 In Less Than One
Instruction Per Byte", 2021): with NEON, and on x86-64 with SSSE3 if
JSON_VIEW_USE_SSSE3 is defined (SSSE3 is not part of x86-64, and the code
must not depend on the flags of a translation unit). JSON_VIEW_NO_SIMD
selects the portable code. The vector code sits in
detail/view/simd.hpp; the same input is accepted either way.

json_document::parse, best of 7 runs in separate processes (M1 Max):
poet.json (CJK text) -72%, random.json -25%, twitter.json -22%,
gsoc-2018.json -20%, semanticscholar -19%, github_events -11%,
apache_builds -9.5%, canada/citm -5/-6%; lottie +4%, tree-pretty +2.5%.

Tests: every two-byte sequence and three- and four-byte sequences with
continuation bytes at the edges of their ranges, at every offset around
the vector blocks of keys and values, cut short, and long runs of text
with a damaged byte, against json::accept and json::parse. CMake builds
the parser tests again with JSON_VIEW_NO_SIMD, and on x86-64 with
JSON_VIEW_USE_SSSE3 and -mssse3; the macros are documented.

Signed-off-by: Niels Lohmann <mail@nlohmann.me>
This commit is contained in:
Niels Lohmann
2026-09-29 14:03:06 +02:00
parent a491cc7663
commit ebfbbf29cb
15 changed files with 863 additions and 20 deletions
+29 -5
View File
@@ -16,6 +16,7 @@
#include <nlohmann/json.hpp>
#include <nlohmann/detail/view/macro_scope.hpp>
#include <nlohmann/detail/view/simd.hpp>
// Scanning primitives of the view's parser. The unrolled checks at fixed
// offsets follow yyjson (https://github.com/ibireme/yyjson, MIT license): the
@@ -62,9 +63,12 @@ NLOHMANN_VIEW_ALWAYS_INLINE std::uint16_t load16(const unsigned char* p) noexcep
/// Advance over plain string bytes and well-formed UTF-8. Stops at a quote,
/// a backslash, a control character, ill-formed UTF-8, or the end. The first
/// 16 bytes are checked one by one, so that the position advances by
/// constants in predicted branches (most strings are short); longer runs
/// continue eight bytes at a time.
/// bytes are checked one by one, so that the position advances by constants
/// in predicted branches: 16 for keys, whose lengths repeat from record to
/// record, and 8 for string values (Value) where a vector loop follows, as
/// their lengths vary more. Longer runs continue 16 bytes at a time with NEON
/// or SSE2, else eight bytes at a time.
template<bool Value = false>
NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* scan_string_run(const unsigned char* p, const unsigned char* e) noexcept
{
const std::uint8_t* plain = string_plain();
@@ -73,9 +77,23 @@ NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* scan_string_run(const unsigned
if (e - p >= 16)
{
#define NLOHMANN_VIEW_STEP(i) if (NLOHMANN_VIEW_LIKELY(plain[p[i]] != 0)) {} else { p += (i); goto stop; }
NLOHMANN_VIEW_REPEAT16(NLOHMANN_VIEW_STEP)
NLOHMANN_VIEW_STEP(0) NLOHMANN_VIEW_STEP(1) NLOHMANN_VIEW_STEP(2) NLOHMANN_VIEW_STEP(3)
NLOHMANN_VIEW_STEP(4) NLOHMANN_VIEW_STEP(5) NLOHMANN_VIEW_STEP(6) NLOHMANN_VIEW_STEP(7)
if (!Value || !NLOHMANN_VIEW_VECTOR)
{
NLOHMANN_VIEW_STEP(8) NLOHMANN_VIEW_STEP(9) NLOHMANN_VIEW_STEP(10) NLOHMANN_VIEW_STEP(11)
NLOHMANN_VIEW_STEP(12) NLOHMANN_VIEW_STEP(13) NLOHMANN_VIEW_STEP(14) NLOHMANN_VIEW_STEP(15)
p += 8;
}
#undef NLOHMANN_VIEW_STEP
p += 16;
p += 8;
#if NLOHMANN_VIEW_VECTOR
p = vector_plain_run(p, e);
if (p != e && plain[*p] == 0)
{
goto stop;
}
#else
while (e - p >= 8)
{
const std::uint64_t special = swar_string_special(read_eight_bytes(p));
@@ -86,6 +104,7 @@ NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* scan_string_run(const unsigned
}
p += 8;
}
#endif
continue;
}
while (p != e && plain[*p] != 0)
@@ -101,6 +120,10 @@ stop:
{
return p; // quote, backslash, or control character
}
#if NLOHMANN_VIEW_VECTOR_UTF8
// non-ASCII: the vector check, out of line
return scan_string_vector(p, e, plain);
#else
// non-ASCII: a run of well-formed sequences (the library's check, so
// that exactly what json::parse accepts is accepted)
do
@@ -113,6 +136,7 @@ stop:
p += n;
}
while (p != e && *p >= 0x80);
#endif
}
}