Idempotent optimisation of bytecode lookup. #104
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
By promoting the bytecode address and moving the load out of the bytecode array into an idempotent function we get a decent speedup of around 32% on bigloop. Is this safe? ltratt: > all we have to worry about is whether a function's bytecodes can > change without us telling yk. there are two ways this could happen: > > a. (most obvious) someone literally mutates the bytecode with > something like `f->code[2] = OP_...` > > b. (less obvious) if yklua changes a function's bytecode allocation. > notably, if it does `free(f->code)` then `f->code = malloc(...)` > and got back the same pointer, then this would be a mutation from our > perspective Edd investigates and comes back with: > I don't see any meaningful "mutations" of the bytecode array that would > matter to us. The array grows as the parser is building the program, it > mutates it locally to do peephole optimisations immediately after the parser > is done, then it shrinks the array in case the peephole optimiser eliminated > instructions. Then it's freed when the function proto dies. All of that to say, we believe this is a safe optimisation.
vext01
commented
Mar 6, 2025
| @@ -1169,12 +1169,23 @@ void luaV_finishOp (lua_State *L) { | |||
|
|
|||
| /* fetch an instruction and prepare its execution */ | |||
| #ifdef USE_YK | |||
| // Elide instruction lookup. | |||
| // | |||
| // FIXME: return type should be `Instruction` not `uint64_t`. We only do this | |||
There was a problem hiding this comment.
I'll fix this in a follow up PR.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Requires ykjit/yk#1642
By promoting the bytecode address and moving the load out of the bytecode array into an idempotent function we get a decent speedup of around 32% on bigloop.
Is this safe?
ltratt:
Edd investigates and comes back with:
All of that to say, we believe this is a safe optimisation.