Most CITM/Twitter realloc cost was due to growing from 1024 in 7-9 doublings
to reach the actual output size. Bumping the initial capacity to 256KB means
Twitter (82KB) fits in one allocation and CITM (496KB) only needs 1 growth.
Measured (TRUE A/B, 7 alternating rounds in single docker, with all prior
follow-ups in this branch + this cap bump):
CITM: 4358 -> 4924 MB/s (+13.0%) - now ~2.3% AHEAD of Glaze
Twitter: 6338 -> 8034 MB/s (+26.8%) - ~50% AHEAD of Glaze
Trade-off: 256KB upfront per string_builder instance. Reasonable for
high-perf JSON serialization but wasteful for tiny one-off messages.
Users serializing small payloads should pass a smaller initial_capacity
to the constructor.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add append_raw_n<N>(const char*) which propagates the key length as a
template parameter, letting the compiler emit a fully inlined memcpy
with a constant size (direct loads/stores) instead of the size-passed-
as-runtime-arg variant which sometimes fell back to a libc memcpy
dispatch.
Use it in the reflection struct atom for both first_key and rest_key
(both are compile-time constants from define_static_string).
Measured (TRUE A/B, 7 alternating rounds in single docker invocation,
two independent runs combined):
CITM: baseline ~4395 -> patched ~4791 MB/s (+9%)
Glaze ~4810 MB/s -> simdjson now at PARITY with Glaze
(within +/- 3% noise across runs)
Twitter: still well ahead of Glaze, no change from baseline
Output is byte-identical, all reflection comprehensive tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three small follow-ups on top of #2707, all aimed at closing the
remaining CITM gap:
1. simdjson_really_inline on atom<map>, atom<smart_pointer>,
atom<enum>. The previous PR missed these three. atom<map> in
particular was visible at ~11% of CITM profile time before this
change (CITM has events: std::map<string, CITMEvent>).
2. struct atom passes the key size explicitly to append_raw(c, len)
instead of going through append_raw(const char*) which calls
std::strlen on every key (8 fields × 184 events on CITM).
3. append_raw(const char*) uses std::char_traits<char>::length
(constexpr) instead of std::strlen so the compiler can fold the
length when the pointer is to a compile-time string.
Measured (TRUE A/B, 7 alternating rounds in single docker invocation):
CITM: baseline 4437 -> patched 4686 MB/s (+5.6%)
Twitter: within noise
Output is byte-identical to baseline. All static_reflection_comprehensive_tests
pass.
WIP — pushing for safekeeping while continuing to investigate the
remaining ~5% gap to Glaze on CITM. Not ready for PR yet.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* add compile option 'SIMDJSON_ENABLE_NAN_INF' but disable by default
* extend parser to support NaN/Infinity when SIMDJSON_ENABLE_NAN_INF=1
* update tests to check parsing of NaN/Infinity, when enabled
* update minefield tests: mark nan/inf tests as passing when nan/inf is ON
* update CI/CD to run tests with extensions for NaN/Infinity enabled
* builder: force-inline atom templates and replace integer writer
Two complementary changes that together speed up reflection-driven JSON
serialization by ~28% on integer-heavy structs (CITM) and ~12% on
string-heavy structs (Twitter).
(1) Add simdjson_really_inline (always_inline) to the hot atom<T>
template overloads (arithmetic, struct, container, optional,
string-like). The constexpr-only declaration was only a hint; the
compiler routinely chose to leave atom<unsigned long> as a real
out-of-line function.
(2) Replace string_builder::append<UInt> body with a forward
cascade-on-magnitude integer writer. The old code computed
digit_count(v) upfront and wrote backward in a loop; the new code is
a straight-line if/else cascade that writes digits forward, no loop,
no helper call.
Either change alone gives only modest gains. Together they unlock the
compiler's cross-call optimization: with all atoms inlined and the
integer writer reduced to straight-line code, the compiler can hoist
b.position into a register across the whole struct serialization, fold
redundant capacity_check calls, and eliminate the strict-aliasing
penalty that otherwise forces b.position/b.capacity reloads after every
char* write.
Output is byte-identical to master on CITM (496682 bytes) and Twitter
(81927 bytes). All static_reflection_comprehensive_tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* builder: de-recurse write_uint_jeaiii so always_inline applies on g++/MSVC
g++ ('inlining failed in call to always_inline ...: function not
considered for inlining') and MSVC ('warning C4714: __forceinline not
inlined' under warnings-as-errors) both refuse to inline recursive
functions marked simdjson_really_inline. The original write_uint_jeaiii
called itself in the >=10^4 branches.
Refactor into a non-recursive DAG of helpers: write_lt100, write_lt10000,
write_4_digits, write_lt1e8, write_uint_jeaiii. Each calls strictly
smaller-domain helpers, no cycles. Same straight-line cascade behavior,
same byte output, but every node is now a candidate for always_inline on
all compilers.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* builder: port signed-int append to jeaiii writer and drop digit_count helpers
The signed-integer branch in string_builder::append was structurally
identical to the OLD unsigned branch — same digit_count() upfront +
backward 4-digit batched loop. Port it to use the same forward
write_uint_jeaiii() helper as the unsigned branch (write '-'
unconditionally and advance position only if negative — branchless).
This makes int_log2 / fast_digit_count_32 / fast_digit_count_64 /
digit_count fully unused (verified via grep across include/ and src/);
remove them, ~80 lines of dead code.
The signed write path now benefits from the same compiler-level
optimization (full inlining, capacity-check fusion, no opaque-loop
boundary) as the unsigned path. Signed-int microbench (200 × 100k
values, mixed magnitude and sign): 2272 → 2685 MB/s (+18.2%).
CITM and Twitter benchmarks unchanged in shape and remain byte-identical
to baseline output.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix tests
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: Daniel Lemire <daniel@lemire.me>
Each non-first field in a reflection-driven struct serialization was
emitting three separate string_builder::append calls (',', "\"key\"", ':'),
each going through capacity_check + a small write. Combine them into a
single compile-time string per field so each field does one capacity_check
and one memcpy. Same pattern fixed in atom(), append(), and extract_from().
Measured under clang-p2996 -O3 -DNDEBUG -freflection -std=c++26
(median of 5 contemporaneous runs, output byte-identical to baseline):
CITM serialization (496682 bytes):
simdjson_reuse_buffer: 3192 -> 3673 MB/s (+15.1%)
simdjson_static_reflection: 3014 -> 3443 MB/s (+14.2%)
simdjson_to: 2717 -> 3234 MB/s (+19.0%)
simdjson_to_reuse: 2836 -> 3226 MB/s (+13.7%)
Twitter serialization (81927 bytes):
simdjson_reuse_buffer: 6637 -> 7162 MB/s (+7.9%)
simdjson_static_reflection: 5527 -> 5915 MB/s (+7.0%)
simdjson_to: 5070 -> 5350 MB/s (+5.5%)
simdjson_to_reuse: 4976 -> 5303 MB/s (+6.6%)
extract_from (out-of-tree micro-bench, 100 records per call):
User 4-of-9 fields: 1833 -> 1888 MB/s (+3.0%)
Status 3-of-6 fields: 4286 -> 4334 MB/s (+1.1%)
Parsing benchmarks unchanged. tests/builder/static_reflection_comprehensive_tests
still passes (round-trip through extract_from / extract_into is verified).
* add std::ranges support for On-Demand API (#2382)
Add zero-cost range wrappers (array_range, object_range) that satisfy
std::ranges::input_range, enabling std::views::transform and other
C++20 range adaptors with the On-Demand parser.
Uses direct forwarding via simdjson_inline with no value buffering,
avoiding the per-element overhead (~20%) of the previous approach.
Guarded by SIMDJSON_SUPPORTS_RANGES.
* fix: replace non-ASCII em dash in test comment
The just_ascii CI check flags any non-ASCII characters in source files.
* let us see what we get with this...
* minor tweak
* minor update
* update doc
---------
Co-authored-by: Justin Li <justin53@bu.edu>
* adding memory-file mapping to Windows.
* making memory-file mapping optional under windows, as it is fragile
* removing non-ascii
* saving.
* bumping up
* take 2
* adding a guard in document::allocate.
Co-authored-by: jmestwa-coder jmestwa@gmail.com
* adding a max depth
Co-authored-by: jmestwa-coder jmestwa@gmail.com
* add get_int32() and get_uint32() to the On Demand API
Add convenience methods that call get_int64()/get_uint64() and
range-check the result, returning NUMBER_OUT_OF_RANGE on overflow.
Added to value, document, document_reference, and their
simdjson_result wrappers.
Closes#1890
* fix document_reference streaming for get_int32/get_uint32
The document_reference getters delegated to document::get_int32/get_uint32
which call get_root_int64/get_root_uint64 with allow_trailing_content=true,
rejecting the next document in a stream as trailing content.
Call get_root_value_iterator().get_root_int64/get_root_uint64(false) directly
to allow trailing content, matching get_int64/get_uint64.
Add streaming tests for both getters.
* add get_int32/get_uint32 to basics.md
* add get_int32/get_uint32 to wrong-type error tests
* add get<uint32_t>/get<int32_t> template specializations
Wire the 32-bit getters into the generic get<T>() and get(T&)
dispatch so that doc.get<uint32_t>() compiles on C++11/14/17
without requiring the C++20 concepts path in std_deserialize.h.
When parser.number_as_string(true) is set and a number exceeds uint64
range, write the raw digits to the string buffer and emit a BIGINT ('Z')
tape tag instead of returning BIGINT_ERROR. Default behavior unchanged.
Changes:
- tape_type.h: add BIGINT = 'Z'
- dom_parser_implementation.h: add _number_as_string flag
- parser.h: add number_as_string() getter/setter
- parser-inl.h: propagate flag before parse
- tape_builder.h: visit_number checks flag on BIGINT_ERROR
- tape_writer.h: add append_bigint helper
- serialization-inl.h: handle BIGINT in serialization (raw digits)
- tape.md: document big integer tape format
Builds on PR #2139 which added BIGINT_ERROR detection.
All 119 existing tests pass — default behavior unchanged.
Closes#167
Co-authored-by: harshagd <harshagd@amazon.com>
* Added benchmarks for on-demand get_int64 and get_double
* Remove local IDE settings
* Fix PCH build failure with <bit> on non-C++20 compilers
* Update bench_dom_api.cpp
Francisco Geiman Thiesen