andrew
0 Followers · 0 Following
  • Joined on 2023-03-22
Repositories Projects Packages Public Activity Starred Repositories
andrew pushed to main at weaselab/conflict-set 2024-08-22 23:52:48 +00:00
8a04e57353 Add to corpus
andrew pushed to main at weaselab/conflict-set 2024-08-21 21:00:17 +00:00
7f86fdee66 Test 64 bit versions
442755d0a6 Update implementation notes
Compare 2 commits »
andrew pushed to main at weaselab/conflict-set 2024-08-20 23:59:08 +00:00
e15b3bb137 Bump version
andrew released v0.0.12 at weaselab/conflict-set 2024-08-20 23:58:04 +00:00
andrew pushed tag v0.0.12 to weaselab/conflict-set 2024-08-20 23:58:03 +00:00
andrew pushed to main at weaselab/conflict-set 2024-08-20 23:34:40 +00:00
311794c37e Update GCOVR annotations now that jenkins agent has avx512
andrew pushed to main at weaselab/conflict-set 2024-08-20 20:36:29 +00:00
dfa178ba19 Add to corpus (from fuzzing on macos)
andrew commented on issue weaselab/conflict-set#5 2024-08-20 20:35:54 +00:00
Try interleaving queries to improve ILP

Instructions per cycle is already pretty high, so I'm not optimistic this will help. We can still try at some point though

andrew closed issue weaselab/conflict-set#18 2024-08-20 20:35:13 +00:00
Performance with USE_SIMD_FALLBACK=ON is surprisingly close to the SIMD performance. Investigate
andrew commented on issue weaselab/conflict-set#18 2024-08-20 20:35:13 +00:00
Performance with USE_SIMD_FALLBACK=ON is surprisingly close to the SIMD performance. Investigate

Probably stale

andrew closed issue weaselab/conflict-set#6 2024-08-20 20:34:57 +00:00
Look for micro-optimizations
andrew commented on issue weaselab/conflict-set#6 2024-08-20 20:34:57 +00:00
Look for micro-optimizations

Many instructions have been shaved. We'll still keep doing this going forward.

andrew closed issue weaselab/conflict-set#11 2024-08-20 20:34:20 +00:00
Rewrite addWriteRange to erase more efficiently
andrew commented on issue weaselab/conflict-set#11 2024-08-20 20:34:20 +00:00
Rewrite addWriteRange to erase more efficiently

Abandoned: too much complexity for too little value. The existing code is already pretty good it seems.

andrew pushed to main at weaselab/conflict-set 2024-08-20 19:06:41 +00:00
a16d18edfe Update README.md
andrew pushed to main at weaselab/conflict-set 2024-08-20 18:38:08 +00:00
2b60287448 Partition valgrind into tests of size at most 100
andrew pushed to main at weaselab/conflict-set 2024-08-20 17:36:42 +00:00
0a9ac59676 Commit to non-simd Node3 implementations
e3a77ed773 Remove unnecessary casts
cdf9a8a7b0 Save 8 bytes in Node3
305dfdd52f Change whitespace in node structs for consistency
7261c91492 Remove Node48::nextFree, and improve padding to save 8 bytes
Compare 5 commits »
andrew pushed to main at weaselab/conflict-set 2024-08-20 04:55:11 +00:00
f11720f5ae Add to corpus
e2b7298af5 Instrument setOldestVersion for callgrind too
Compare 2 commits »
andrew pushed to main at weaselab/conflict-set 2024-08-17 23:40:50 +00:00
8e1e344f4b Fix clang-18 warning about std::basic_string<uint8_t>
andrew pushed to main at weaselab/conflict-set 2024-08-17 21:45:54 +00:00
3634b6a59b Simplify slightly in checkMaxBetweenExclusive
a3cc14c807 Fix double counting of remove node in showMemory mode
Compare 2 commits »