| Commit message (Collapse) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | adler32: Add benchmark |  Runxi Yu | 2025-11-20 | 1 | -0/+22 |
| | | |||||
| * | Add the right go:build's | Runxi Yu | 2025-11-19 | 2 | -0/+4 |
| | | |||||
| * | Replace adler32 with github.com/mhr3/adler32-simd | Runxi Yu | 2025-11-19 | 18 | -892/+930 |
| | | | | | Unfortunately no VSX support yet | ||||
| * | Ignore some errors in the tests (actually from stdlib but linter complains) | Runxi Yu | 2025-11-19 | 2 | -11/+11 |
| | | |||||
| * | Remove the broken compression benchmarking utility | Runxi Yu | 2025-11-19 | 6 | -224/+0 |
| | | |||||
| * | SIMD with AVX2 on supported AMD64 machines | Runxi Yu | 2025-11-19 | 5 | -92/+423 |
| | | | | | | | | | | Some help from gpt-5.1-thinking taken: used wrong register size for the weighted sum at first, so it was truncating the second half of our block; also there was an overflow from the modulus and stuff. Unfortunately the AVX2 adler32 is only about 20% faster than the generic version which doesn't make for much. | ||||
| * | adler32: Add tests from the standard library | Runxi Yu | 2025-11-19 | 2 | -0/+388 |
| | | |||||
| * | Switch back to internal zlib for loose objects too | Runxi Yu | 2025-11-19 | 4 | -2/+424 |
| | | | | | Still marginally less overhead due to pooling | ||||
| * | Probably should name the custom packages specially | Runxi Yu | 2025-11-19 | 10 | -17/+17 |
| | | |||||
| * | flate: Remove unused code | Runxi Yu | 2025-11-19 | 3 | -447/+4 |
| | | |||||
| * | SliceInflater is probably a better name for that | Runxi Yu | 2025-11-19 | 2 | -52/+496 |
| | | | | | And let's just basically not use io.Reader at all | ||||
| * | Initial attempt to make a compressor with less overhead | Runxi Yu | 2025-11-19 | 5 | -0/+330 |
| | | | | | io.Reader actually has massive overhead... | ||||
| * | adler32: Unroll update loop | Runxi Yu | 2025-11-19 | 1 | -7/+70 |
| | | |||||
| * | byteorder: Import | Runxi Yu | 2025-11-19 | 2 | -0/+176 |
| | | |||||
| * | Add adler32 and LICENSE | Runxi Yu | 2025-11-19 | 2 | -0/+156 |
| | | |||||
| * | zlib: Use internal adler32 | Runxi Yu | 2025-11-19 | 1 | -1/+1 |
| | | |||||
| * | Our zlib ought to be read-only for now | Runxi Yu | 2025-11-19 | 5 | -1941/+8 |
| | | |||||
| * | Add zlib test data | Runxi Yu | 2025-11-19 | 6 | -0/+222 |
| | | |||||
| * | Import flate | Runxi Yu | 2025-11-19 | 10 | -2/+3234 |
| | | |||||
| * | zlib: Add missing error check for flate Reset | Runxi Yu | 2025-11-18 | 1 | -1/+4 |
| | | |||||
| * | Use a pooled zlib implementation | Runxi Yu | 2025-11-18 | 3 | -0/+414 |
| | | | | | | It's trivially forked from the standard library and improves git ls-tree -r --long speed by approximately 25%. | ||||
| * | Extract bufpool into its own package | Runxi Yu | 2025-11-16 | 2 | -0/+180 |
 |
index : ~runxiyu/furgit.git | |
| Unnamed repository; edit this file 'description' to name the repository. |
| aboutsummaryrefslogtreecommitdiff |