grit://cooking/bench/reset/hard-delete-heavy@wide-48k
← cookingall benchmarksraw data

grit reset · hard-delete-heavy @wide-48k

Hard reset to a commit deleting 2000 files in 20 dirs of a 48k-file tree and back: grit reset --hard vs git reset --hard

repo: 48,000-file baseline tree; other branch deletes 2,000 files · 48,000 files · 2 commits · 3 refs

$ grit-bench daily --only reset/hard-delete-heavy@wide-48k2026-10-27 · grit 8dbc58e77 · git version 2.43.0
toolmedianminmaxrunspeak memory
git225 ms222 ms229 ms721 MB
grit236 ms234 ms250 ms752 MB
grit is parmemory: 2.5× more
every timed run (log scale)
git
grit
176 ms315 ms
How it's measured
before each run
grit-bench prepare-switch --git git
git runs
$ git reset -q --hard bench-b && git reset -q --hard 438e0171ffc3b3bd1228cdd85901840f19dc6595
grit runs
$ grit reset --hard bench-b && grit reset --hard 438e0171ffc3b3bd1228cdd85901840f19dc6595
Where the time goesone extra traced run per tool, outside the timed runs
git · median run 225 ms
unpack_trees: unpack_trees ×2
171 ms
unpack_trees: traverse_trees ×2
118 ms
index: do_write_index ×2
19 ms
cache-tree: prime_cache_tree ×2
17 ms
index: do_read_index ×2
5.2 ms
cache_tree: read ×2
0.2 ms
cache_tree: write ×2
0.1 ms
grit · median run 236 ms
grit.index.write ×2
20 ms
grit.index.read ×2
8.0 ms
grit.diff.tree ×2
1.7 ms
grit.repository.open ×2
0.4 ms
grit.config.load ×4
0.1 ms
grit.pack.index.open ×2
0.0 ms
Every day8 days
dategrit commitgitgritgrit vs gitgrit peak
2026-10-278dbc58e77225 ms236 mspar52 MB
2026-10-26c7cf2c443375 ms418 ms1.1× slower56 MB
2026-10-252824ad391227 ms225 mspar50 MB
2026-10-24157a05dde239 ms238 mspar50 MB
2026-10-23bddf92f86—not run——
2026-10-224c131b942—not run——
2026-10-21d4844e92a—not run——
2026-10-204754ad0b9—not run——