Results at a glance
Every cell below is one library driving one screen-shaped task. It was verified
against a plain-Map reference model before timing, then timed in its own processes
beside a hand-written anchor. Each row is one scenario, named as in its own table below, and
the fastest library's cell carries a 🏆 and its time; every other cell says how many
times slower that library is than the fastest one: 1.3×, 12×. The two baseline-* columns are plain Map code with no
library; they are shown for scale and never win a row.
Most libraries here ship no collections, group indexes or sorted windows. In those rows the collections are written by hand on top of the library's primitives, without OIMDB, so the row measures that code, not a feature the library ships. Such a column is marked "+ hand-written collections".
| scenario | baseline-array | baseline-indexed | oimdb | oimdb-sorted-index | oimdb-3.12.1 | tinybase | signaldb | tanstack-db | tanstack-db-eager | tanstack-db-indexed | lokijs | mobx+ hand-written collections | jotai+ hand-written collections | legend-state+ hand-written collections | effector+ hand-written collections | nanostores+ hand-written collections | nanostores-per-row+ hand-written collections | redux-toolkit+ hand-written collections | preact-signals+ hand-written collections | alien-signals+ hand-written collections | zustand+ hand-written collections | valtio+ hand-written collections | xstate-store+ hand-written collections | vue-reactivity+ hand-written collections | solid-js+ hand-written collections | tanstack-store+ hand-written collections |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| batch insert | 0.74× | 3.2× | 3.9× | 1.7× | 17× | 68× | 72× | 19× | 40× | 41× | 6.0× | 25× | 14× | 51× | 2.0× | 2.3× | 3.2× | 16× | 3.5× | 4.0× | 1.1× | 44× | 🏆 998.13 µs | 3.4× | 5.1× | 1.8× |
| insert into a populated store | 3.1× | 3.2× | 2.7× | 2.1× | 7.1× | 14× | 66× | 23× | 25× | 20× | 🏆 113.67 µs | 7.9× | 5.1× | 18× | 7.4× | 67× | 2.9× | 45× | 2.7× | 2.8× | 20× | 10× | 21× | 2.6× | 3.7× | 2.6× |
| the population keeps growing | 1.3× | 0.77× | 1.2× | 🏆 1.082 ms | 7.7× | 28× | 102× | 16× | 25× | 24× | 1.4× | 12× | 8.9× | 28× | 5.7× | 22× | 2.8× | 48× | 1.6× | 2.3× | 20× | 33× | 20× | 1.8× | 3.4× | 1.1× |
| cold start to first render | 0.71× | 1.8× | 2.1× | 1.6× | 8.7× | 68× | 68× | 24× | 38× | 38× | 5.7× | 20× | 14× | 50× | 2.0× | 2.1× | 3.0× | 16× | 3.6× | 3.7× | 1.1× | 44× | 🏆 1.036 ms | 4.1× | 4.7× | 1.7× |
| tear the screen down | 1.4× | 2.0× | 2.4× | 1.9× | 18× | 5517× | 101× | 26× | 39× | 35× | 16× | 72× | 110× | 21× | 4.3× | 5.0× | 7.9× | 2.6× | 3.6× | 3.0× | 1.1× | 65× | 🏆 2.75 µs | 3.7× | 5.3× | 4.0× |
| patch one column | 28× | 2.7× | 1.1× | 🏆 133.52 µs | 1.1× | 4.2× | 1397× | 11× | 11× | 11× | 2.7× | 1.7× | 40× | 111× | 4.3× | 2.8× | 2.2× | 784× | 1.8× | 1.2× | 524× | 21× | 524× | 1.9× | 2.6× | 3.3× |
| patch a row off screen | 3216× | 1.9× | 2.4× | 2.6× | 5.5× | 258× | 150113× | 632× | 663× | 678× | 309× | 5.6× | 205× | 13227× | 492× | 319× | 257× | 91450× | 🏆 1.14 µs | 1.1× | 61887× | 2385× | 61757× | 1.1× | 1.2× | 372× |
| patch an unread column | 3166× | 0.85× | 3.3× | 1.9× | 112× | 193× | 160005× | 921× | 925× | 960× | 298× | 5.4× | 231× | 73× | 481× | 312× | 254× | 88726× | 🏆 1.20 µs | 1.1× | 60108× | 2308× | 58957× | 1.1× | 1.2× | 360× |
| batch patch | 2.2× | 1.0× | 1.4× | 1.4× | 2.9× | 30× | 1409× | 42× | 51× | 48× | 6.5× | 5.0× | 18× | 43× | 9.6× | 225× | 4.7× | 62× | 1.4× | 1.0× | 33× | 7.4× | 32050× | 🏆 69.54 µs | 1.7× | 2.2× |
| batch patch an unread column | 2.7× | 0.67× | 1.2× | 1.1× | 3.5× | 35× | 1720× | 53× | 60× | 60× | 7.9× | 4.6× | 15× | 32× | 12× | 279× | 5.9× | 77× | 1.2× | 🏆 56.26 µs | 42× | 9.4× | 40231× | 1.0× | 1.2× | 2.7× |
| most rows change every frame | 0.70× | 0.70× | 1.4× | 1.5× | 2.7× | 27× | 1226× | 50× | 56× | 55× | 4.8× | 4.0× | 12× | 26× | 3.6× | 184× | 3.8× | 21× | 1.3× | 🏆 901.46 µs | 6.3× | 6.1× | 25019× | 1.2× | 1.4× | 1.5× |
| one frame, every kind of change | 2.0× | 1.00× | 1.5× | 1.2× | 4.9× | 9.7× | 110× | 11× | 13× | 13× | 23× | 5.2× | 7.2× | 18× | 1.2× | 7.6× | 1.6× | 57× | 1.1× | 1.0× | 25× | 4.2× | 304× | 🏆 94.79 µs | 1.3× | 1.2× |
| row moves between groups | 9.2× | 2.8× | 1.7× | 🏆 101.76 µs | 2.8× | 9.2× | 522× | 6.4× | 6.3× | 6.4× | 1.8× | 18× | 23× | 76× | 3.1× | 5.8× | 1.9× | 270× | 1.2× | 1.3× | 177× | 13× | 178× | 1.3× | 1.5× | 2.1× |
| a row arrives mid-order | 6.0× | 1.9× | 1.6× | 1.1× | 3.4× | 18× | 328× | 11× | 11× | 12× | 2.1× | 13× | 20× | 60× | 2.3× | 7.9× | 1.9× | 179× | 🏆 80.79 µs | 1.1× | 111× | 10× | 220× | 1.1× | 1.5× | 1.6× |
| sort key moves | 21× | 2.0× | 1.8× | 1.1× | 4.4× | 10× | 1146× | 13× | 19× | 19× | 2.8× | 15× | 26× | 86× | 3.3× | 2.1× | 1.6× | 615× | 🏆 43.82 µs | 1.1× | 410× | 16× | 407× | 1.5× | 1.8× | 2.4× |
| the whole window re-sorted | 6.9× | 0.71× | 1.3× | 1.7× | 28× | 14× | 614× | 37× | 41× | 41× | 1.8× | 6.2× | 17× | 33× | 1.3× | 1.7× | 1.2× | 204× | 1.1× | 1.7× | 136× | 6.2× | 4376× | 1.5× | 2.2× | 🏆 132.34 µs |
| a consumer per group | 385× | 1.5× | 2.1× | 1.0× | 3.8× | 18× | 3474× | 8.2× | 9.3× | 9.3× | 4.3× | 17× | 31× | 71× | 2.6× | 4.7× | 2.0× | 440× | 🏆 454.18 µs | 1.2× | 325× | 14× | 313× | 1.2× | 1.5× | 1.9× |
| targeted column delivery | 1.8× | 0.94× | 1.0× | 1.0× | 2.1× | 3.4× | 3832× | 28× | 28× | 29× | 1.9× | 1.5× | 3.5× | 7.0× | 13× | 9.0× | 6.4× | 2883× | 🏆 18.69 µs | 1.0× | 1939× | 1.3× | 1930× | 1.1× | 1.4× | 1.1× |
| delete the visible window | 4.5× | 0.69× | 1.7× | 1.5× | 15× | 23× | 342× | 49× | 42× | 45× | 114× | 8.3× | 14× | 32× | 1.0× | 1.7× | 1.5× | 148× | 🏆 34.25 µs | 1.3× | 68× | 4.0× | 67× | 1.0× | 2.0× | 1.2× |
| delete one visible row | 19× | 1.9× | 1.9× | 1.4× | 5.4× | 22× | 810× | 13× | 14× | 15× | 190× | 18× | 22× | 85× | 2.9× | 9.2× | 2.0× | 693× | 🏆 57.46 µs | 1.2× | 315× | 12× | 315× | 1.2× | 1.8× | 2.5× |
| the group empties under the view | 3.5× | 1.0× | 1.9× | 1.8× | 16× | 52× | 474× | 59× | 68× | 69× | 17× | 8.5× | 22× | 46× | 🏆 78.35 µs | 1.3× | 1.5× | 109× | 1.5× | 1.5× | 57× | 14× | 57× | 1.2× | 1.9× | 1.5× |
| re-send unchanged list | 68× | 7.0× | 1.9× | 1.9× | 1.5× | 48× | 5240× | 20× | 21× | 20× | 14× | 8.1× | 83× | 24× | 12× | 12× | 10× | 1325× | 1.1× | 2.4× | 1331× | 12× | 1342× | 🏆 1.69 µs | 1.8× | 10× |
| mount and unmount one subscriber | 1.1× | · | 1.0× | 🏆 2.58 µs | 2.6× | 14× | 11967× | 250× | 264× | 255× | 1.7× | 6.1× | 5.2× | 19× | 5.8× | 1.1× | 1.1× | 20829× | 1.7× | 1.2× | 13860× | 3.2× | 13956× | 2.7× | 3.2× | 1.2× |
| N subscribers on one root, oldest out first | 0.96× | · | 1.0× | 1.0× | 1.9× | 8.4× | 2296× | 61× | 66× | 61× | 2.2× | 13× | 4.9× | 24× | 12× | 🏆 19.47 µs | 1.0× | 2725× | 3.2× | 2.1× | 1834× | 4.6× | 1838× | 9.1× | 6.7× | 1.7× |
| N subscribers on one root, newest out first | 1.0× | · | 🏆 20.12 µs | 1.0× | 1.8× | 8.0× | 2276× | 54× | 60× | 52× | 1.6× | 13× | 4.7× | 23× | 12× | 1.1× | 1.1× | 2650× | 5.2× | 2.0× | 1772× | 4.4× | 1777× | 6.9× | 6.0× | 1.6× |
| N×8 subscribers on one root, oldest out first | 1.1× | · | 1.0× | 🏆 147.19 µs | 1.9× | 7.5× | ✗ | 50× | 54× | 49× | 6.9× | 12× | 4.6× | 23× | 12× | 1.2× | 1.2× | 367× | 5.7× | 2.2× | 243× | 4.6× | 245× | 7.7× | 6.7× | 1.9× |
| N×8 subscribers on one root, newest out first | 1.4× | · | 🏆 146.71 µs | 1.0× | 1.7× | 7.5× | ✗ | 46× | 49× | 47× | 2.3× | 12× | 4.5× | 23× | 12× | 1.5× | 1.4× | 367× | 5.4× | 2.2× | 244× | 4.6× | 247× | 7.8× | 8.8× | 1.9× |
| one subscriber over many sources | 1.9× | · | 1.0× | 🏆 13.41 µs | 5.1× | 20× | 2422× | 38× | 39× | 37× | 1.9× | 25× | 28× | 27× | 50× | 3190× | 3197× | 3968× | 7.6× | 4.8× | 2656× | 4.6× | 2652× | 9.5× | 20× | 1.3× |
| N subscribers on one column | 0.93× | · | 1.1× | 1.1× | 2.0× | 8.3× | 2260× | ✗ | ✗ | ✗ | · | 8.3× | 36× | 28× | 73× | 1.5× | 1.5× | 2629× | 1.5× | 🏆 20.32 µs | 1754× | 2.5× | 1762× | 4.8× | 3.6× | 2.7× |
| N subscribers on N rows, a list | 1.1× | · | 1.0× | 🏆 26.88 µs | 2.5× | 16× | 1.9× | 1.1× | ||||||||||||||||||
| N subscribers over a derivation chain | 1.7× | · | 2.0× | 2.0× | ✗ | ✗ | · | 13× | 19× | 348× | 5.7× | 🏆 153.48 µs | · | 2.7× | 2.5× | |||||||||||
| a write through standing chains | 0.54× | · | 🏆 51.71 µs | 1.0× | 2.2× | ✗ | · | 7.1× | 22× | 1055× | 1.2× | 1.1× | · | 1.4× | 2.5× | |||||||||||
| N subscribers over a diamond | 1.2× | · | 1.8× | 1.9× | 11× | · | 2.6× | 🏆 110.04 µs | · | 2.8× | 3.7× | |||||||||||||||
| a write through standing diamonds | 0.27× | · | 1.0× | 1.0× | 4.2× | · | 1.2× | 🏆 78.31 µs | · | 1.8× | 3.6× | |||||||||||||||
| churn against a standing population | 0.71× | · | 1.0× | 🏆 93.90 µs | 1.6× | 1.1× | ||||||||||||||||||||
| an application screen fills | 1.1× | · | 🏆 22.08 µs | 1.0× | 8.8× | 1.1× | ||||||||||||||||||||
| an application screen in use | 0.65× | · | 1.0× | 🏆 341.73 µs | 1.7× | 7.7× | 2.5× | 1.9× | 5.2× | 5.0× | ||||||||||||||||
| one setting moves | 0.86× | · | 🏆 4.68 µs | 1.0× | 1.4× | 1.2× | 1.9× | 2.2× | ||||||||||||||||||
| read a page by key | 0.98× | 0.98× | 1.0× | 1.0× | 🏆 36.11 µs | 1.2× | 1.3× | |||||||||||||||||||
| scroll the window | 21× | 14× | 1.2× | 1.3× | 2.0× | 🏆 42.36 µs | 1.0× | 1.2× | 1.5× | |||||||||||||||||
| scroll deep into the group | 21× | 4.8× | 1.3× | 2.3× | 1.8× | 1.1× | 🏆 48.67 µs | 1.3× | 1.6× | |||||||||||||||||
| the view switches group | 19× | 1.6× | 2.3× | 2.3× | 1.5× | 1.0× | 🏆 51.30 µs | 1.2× | 1.6× | |||||||||||||||||
| a view filtered on two axes | 13× | 3.3× | 2.2× | 1.1× | 5.1× | 🏆 34.15 µs | 1.1× | 1.1× | 1.7× | |||||||||||||||||
| a write to a related entity | 2.0× | · | 1.0× | 🏆 7.78 µs | 1.5× | 1.5× | 1.5× | 1.6× | 1.8× | |||||||||||||||||
| a record keyed by two identities | 1.9× | · | 1.0× | 🏆 8.94 µs | 1.6× | 1.7× | 1.6× | 1.8× | 2.1× | |||||||||||||||||
| write one value | 0.85× | · | 1.1× | 1.1× | 1.8× | 1.1× | 🏆 5.87 µs | |||||||||||||||||||
| re-write one value | 21× | · | 5.1× | 5.1× | 1.1× | 🏆 80 ns |
How to read a cell
- Colour runs from green to red: dark green is the fastest library, light green is
a little slower, yellow is about twice as slow, and red is slower still, darkest at
an order of magnitude and beyond. The thresholds are in
BenchMatrix.tsx. The value is always printed; colour only helps the eye find the pattern. - Hatched cells are stale (the code, a library version or the sampling plan changed since they were measured) or flagged (the batch's control pair could not resolve its own known difference, the machine was busy, a process was lost). Tick only fresh, unflagged cells to hide them.
·means the library declined that wiring, with a written reason in the scenario tables.✗means it ran and either disagreed with the reference model or crashed; it gets no number. An empty dashed cell has not been measured yet.- Hover a cell for the absolute time, drift, render counts and date.
- The six floors (scripts with no store, measured once to bound the harness's own share) are instruments, not rows to compare, so they appear only in the scenario tables.
The matrix ranks nothing on its own: a place is given only in the scenario tables, and only where a row stands clear of every row above it by more than both rows' measured drift.
Published 2026-10-31 19:36 UTC from arm64 8× Apple M1, Node v24.16.0.
How much code each task took
Time is one cost; the code a developer writes to get it is another. Each cell is the number
of code lines (comments and blanks excluded) reachable from the participant's wiring for that
task's shape: its builder, every helper it calls and the constants it reads, but not the
methods only another shape needs. The source is first re-printed in one canonical layout
(the TypeScript printer, every block and object literal one member per line, comments
dropped), so joining calls onto one long line or splitting them changes nothing. The count
is computed from the adapter source by npm run lines, needs no measurement, and is current
on every publish. Each row is one wiring, shared by the tasks it names; hover it for the list.
The library with the fewest lines in a row gets a 🏆, and the colour runs from green (fewest)
to red (ten times as many or more). The two baseline-* columns are plain Map code, shown
grey for scale. Fewer lines is not better on its own: read it beside the time, never
instead of it.
| lines of code to wire | baseline-array | baseline-indexed | oimdb | oimdb-sorted-index | oimdb-3.12.1 | tinybase | signaldb | tanstack-db | tanstack-db-eager | tanstack-db-indexed | lokijs | mobx+ hand-written collections | jotai+ hand-written collections | legend-state+ hand-written collections | effector+ hand-written collections | nanostores+ hand-written collections | nanostores-per-row+ hand-written collections | redux-toolkit+ hand-written collections | preact-signals+ hand-written collections | alien-signals+ hand-written collections | zustand+ hand-written collections | valtio+ hand-written collections | xstate-store+ hand-written collections | vue-reactivity+ hand-written collections | solid-js+ hand-written collections | tanstack-store+ hand-written collections |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Board screen 21 tasks | 123 | 246 | 285 | 285 | 216 | 214 | 271 | 249 | 249 | 249 | 212 | 200 | 305 | 🏆 182 | 314 | 216 | 216 | 290 | 298 | 328 | 245 | 212 | 297 | 332 | 322 | 244 |
| A consumer per group 1 task | 117 | 240 | 279 | 279 | 210 | 208 | 265 | 243 | 243 | 243 | 206 | 194 | 299 | 🏆 176 | 308 | 210 | 210 | 284 | 292 | 322 | 239 | 206 | 291 | 326 | 316 | 238 |
| Subscribe and unsubscribe 10 tasks | 151 | · | 155 | 155 | 117 | 🏆 95 | 210 | 189 | 189 | 189 | 172 | 152 | 162 | 114 | 179 | 104 | 104 | 225 | 143 | 141 | 215 | 116 | 228 | 172 | 140 | 130 |
| One subscriber over many sources 1 task | 151 | · | 155 | 155 | 117 | 🏆 95 | 210 | 189 | 189 | 189 | 172 | 152 | 162 | 114 | 179 | 104 | 104 | 225 | 143 | 141 | 215 | 116 | 228 | 172 | 140 | 130 |
| Subscribers on one column 1 task | 148 | · | 157 | 157 | 101 | 🏆 89 | 208 | 199 | 199 | 199 | · | 143 | 151 | 105 | 169 | 91 | 91 | 220 | 133 | 131 | 210 | 103 | 215 | 172 | 140 | 127 |
| Subscribers over a derivation chain 2 tasks | 167 | · | 229 | 229 | 133 | 112 | 201 | 219 | 219 | 219 | · | 154 | 158 | 116 | 174 | 🏆 101 | 🏆 101 | 224 | 144 | 142 | 214 | · | 226 | 178 | 147 | 130 |
| Subscribers over a diamond 2 tasks | 161 | · | 229 | 229 | 123 | 110 | 193 | 212 | 212 | 212 | · | 146 | 150 | 108 | 172 | 🏆 94 | 🏆 94 | 224 | 137 | 135 | 214 | · | 219 | 176 | 145 | 129 |
| Read a page by key 1 task | 89 | 161 | 193 | 193 | 168 | 194 | 206 | 206 | 206 | 206 | 🏆 150 | 175 | 243 | 171 | 252 | 175 | 175 | 281 | 284 | 300 | 240 | 164 | 309 | 308 | 291 | 210 |
| Scroll a window 2 tasks | 83 | 155 | 189 | 189 | 162 | 177 | 202 | 210 | 210 | 210 | 🏆 144 | 168 | 236 | 164 | 242 | 170 | 170 | 271 | 276 | 293 | 237 | 156 | 302 | 301 | 284 | 203 |
| Switch the viewed group 1 task | 84 | 156 | 194 | 194 | 164 | 180 | 203 | 203 | 203 | 203 | 🏆 146 | 169 | 236 | 167 | 242 | 170 | 170 | 271 | 279 | 296 | 237 | 159 | 302 | 302 | 287 | 206 |
| Filter on two axes 1 task | 79 | 151 | 183 | 183 | 157 | 176 | 198 | 196 | 196 | 196 | 🏆 140 | 165 | 233 | 161 | 241 | 164 | 164 | 270 | 273 | 290 | 229 | 153 | 298 | 298 | 281 | 200 |
| Related entities 2 tasks | 154 | · | 214 | 214 | 184 | 145 | 286 | 273 | 273 | 273 | 223 | 210 | 220 | 🏆 137 | 226 | 140 | 140 | 267 | 209 | 191 | 253 | 165 | 302 | 213 | 176 | 169 |
| One value 2 tasks | 43 | · | 92 | 92 | 66 | 🏆 59 | 102 | 144 | 144 | 144 | 86 | 72 | 88 | 70 | 68 | 60 | 60 | 85 | 70 | 76 | 66 | 63 | 84 | 99 | 72 | 61 |
Where the rest is
| page | what it answers |
|---|---|
| Scenario tables | every scenario's full table: time, ratio, drift, render counts, flags, who declined and why |
| Participants | each library's version, what it stores, and the audit of how idiomatically its adapter uses it |
| Scenarios | what each script models, what decides it, and the audit of how representative it is |
| Methodology | verification, the anchor, the controls, the ranking rule, the load gate, staleness |
| Running the suite | measuring in small pieces, a budget at a time, without giving the machine a day |