Repository navigation
Commit c3b1605
committed
docs(bench): the engine x thread matrix, and what each cell costs
Confirms the worker-thread hypothesis rather than leaving it as a card. 20k
calls per cell, one process, same trivial export:
JIT main thread 16547 ns
JIT worker thread 1181 ns
interp main thread 399 ns
interp worker thread 413 ns
The 14x JIT penalty is entirely the initial thread, so zwasm D-584 is dodgeable
cljw-side. The interpreter is thread-indifferent, which is what a cached stack
limit looks like. And the JIT is still 2.9x the interpreter on a worker thread,
which is D-585 becoming visible now that D-584 has stopped swamping it.
Engine and thread are independent levers; the engine one stays shape-dependent,
since the JIT wins 15.5x on compute-in-wasm.1 parent a149251 commit c3b1605
2 files changed
Lines changed: 47 additions & 1 deletion
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
34 | 34 | | |
35 | 35 | | |
36 | 36 | | |
37 | | - | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | + | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | + | |
| 51 | + | |
| 52 | + | |
| 53 | + | |
| 54 | + | |
| 55 | + | |
| 56 | + | |
| 57 | + | |
| 58 | + | |
| 59 | + | |
| 60 | + | |
| 61 | + | |
| 62 | + | |
| 63 | + | |
| 64 | + | |
| 65 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
0 commit comments