You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think we should do a baseline compilation for a large portion of javascript code instead of creating SSA and allocating registers.
The idea is (at least in my head) to generate CFG using ssa.js and perform some alternative register allocation algorithm on it, thus skipping the SSA transformation and Linearscan.
The register allocation could just assign rax, rbx, rcx ... to inputs and spill everything else right after the instruction. Should work just fine for the compilation speed reasons.
The text was updated successfully, but these errors were encountered:
I think we should do a baseline compilation for a large portion of javascript code instead of creating SSA and allocating registers.
The idea is (at least in my head) to generate CFG using ssa.js and perform some alternative register allocation algorithm on it, thus skipping the SSA transformation and Linearscan.
The register allocation could just assign
rax
,rbx
,rcx
... to inputs and spill everything else right after the instruction. Should work just fine for the compilation speed reasons.The text was updated successfully, but these errors were encountered: