-
## Why
[simdjson](https://github.com/simdjson/simdjson) is a high performance json library, which is faster than rapidjson.
We can use it to replace the rapidjson. This will enhance
1. speed of i…
-
Current we fuzz simdjson for crashes and internal checks, but not for correctness.
We have a framework for testing for correctness by generating random JSON and testing against different JSON parse…
-
The simdjson library has a new UTF8 validator called lookup 4 which is simpler and faster than most alternatives.
See
https://github.com/simdjson/simdjson/blob/master/src/generic/stage1/utf8_lo…
-
Hi,
My understanding that the key benefit of SIMD is that we can "progressively" parse a stream of JSON hence the tape reading like implementation.
Although in benchmarks with varying stop poin…
-
Most of your code for x86_64 under skylake ifdef
Not all servers have very modern CPU, haswell and westmere overloads exists for simdjson and simdutf for an example
Examples under Apache 2.0:
h…
MBkkt updated
4 weeks ago
-
TinyUSDZ now support UTF-8 Identifier by default in `dev` branch.
https://github.com/PixarAnimationStudios/USD/pull/2120
https://openusd.org/release/api/_usd__page__u_t_f_8.html
- [x] Basic U…
syoyo updated
2 months ago
-
I just had a showerthought idea regarding two possible performance improvements in ccls.
I didn't do any benchmarking neither know whether these components are the performance bottleneck of ccls. S…
-
The simdjson framework can be almost trivially extended to CSV files. Not 1.0 feature, but it would be a natural extension.
-
Now when oss-fuzz provides pretty code coverage, it is easy to see which parts of the code are not fuzzed properly.
Identify what can be done, and add fuzzers targeting the uncovered code.
To trac…
-
https://github.com/simdjson/simdjson/pull/1849/files