simdjson: MSVC simdjson twice as slow as ClangCL
On Windows, simdjson compiled with MSVC parses twitter.json at almost half the speed of ClangCL-compiled simdjson, on the same machine, in the same command prompt.
| Platform | Overall | Stage 1 | Stage 2 |
|---|---|---|---|
| MSVC 19.25.28614 | 1.3051 | 2.3777 | 3.3898 |
| ClangCL 9.0.0 | 2.2221 | 5.4161 | 4.6401 |
Methodology:
- MSVC:
git clean -ffdx build && cmake -B build && cmake --build build --target parse --config Release && build\benchmark\Release\parse jsonexamples\twitter.json - ClangCL:
git clean -ffdx build && cmake -B build -T ClangCL && cmake --build build --target parse --config Release && build\benchmark\Release\parse jsonexamples\twitter.json
I validated that MSVC simdjson is using the haswell implementation, both by running json2json to print out the implementation, and by doing set SIMDJSON_FORCE_IMPLEMENTATION=haswell.
About this issue
- Original URL
- State: closed
- Created 4 years ago
- Comments: 64 (56 by maintainers)
So, we looked into it, and it turns out the issue is that having an intrinsic in a loop actually causes our loop unroller to bail out. So that’s not great. We’ll get it fixed.
Compiling simdjson with GNU GCC (MinGW 64-bit) under Windows seems to give good performance that is on par with Linux. I am achieving 2.8 GB/s on twitter.json.
Here is a screenshot proof…
So it is not Linux holding us back. There is really something bad for us in the Visual Studio backend.
Hmmm…
cc @jkeiser @eddelbuettel