grit://cooking/bench/switch/sparse@sparse-120k

grit switch · sparse @sparse-120k

Sparse switch out and back, 300 visible files and 400 excluded directories

repo: 120,000-file repository with a 300-file visible sparse workset; sparse reset/switch scaling · 120,000 files · 2 commits · 3 refs

$ grit-bench daily --only switch/sparse@sparse-120k2026-10-27 · grit 8dbc58e77 · git version 2.43.0

tool	median	min	max	runs	peak memory
git	42 ms	41 ms	44 ms	20	7.8 MB
grit	23 ms	22 ms	24 ms	20	20 MB

grit is 1.8× fastermemory: 2.6× more

every timed run (log scale)

git

grit

18 ms55 ms

How it's measured

git runs

$ git switch --detach sparse-base && git switch --detach sparse-edited

grit runs

$ grit switch --detach sparse-base && grit switch --detach sparse-edited

checked first

before timing, git and grit were each run once and their results compared

Where the time goesone extra traced run per tool, outside the timed runs

git · median run 42 ms

unpack_trees: unpack_trees ×4

36 ms

unpack_trees: traverse_trees ×4

23 ms

index: refresh ×2

1.7 ms

index: do_write_index ×2

0.4 ms

cache_tree: update ×2

0.4 ms

index: clear_skip_worktree_from_present_files ×2

0.4 ms

index: do_read_index ×2

0.3 ms

cache_tree: read ×2

0.1 ms

unpack_trees: clear_ce_flags(0x00000000,0x02000000) ×2

0.1 ms

cache_tree: write ×2

0.1 ms

unpack_trees: clear_ce_flags(0x00080000,0x42000000) ×2

0.0 ms

grit · median run 23 ms

grit.index.write ×2

0.6 ms

grit.repository.open ×2

0.4 ms

grit.index.read ×2

0.2 ms

grit.config.load ×6

0.2 ms

grit.pack.index.open ×2

0.0 ms

Every day8 days

date	grit commit	git	grit	grit vs git	grit peak
2026-10-27	8dbc58e77	42 ms	23 ms	1.8× faster	20 MB
2026-10-26	c7cf2c443	—	not run	—	—
2026-10-25	2824ad391	—	not run	—	—
2026-10-24	157a05dde	—	not run	—	—
2026-10-23	bddf92f86	—	not run	—	—
2026-10-22	4c131b942	—	not run	—	—
2026-10-21	d4844e92a	—	not run	—	—
2026-10-20	4754ad0b9	—	not run	—	—