Currently, extraction uses a lot of memory. Function inlining makes extraction slower. Running cargo run -r tests/passing/small/fib_recursive.bril --run-mode compile-brilift --optimize-egglog true --optimize-brilift true --interp true on its own, for example, takes 7 seconds with 3 iterations of inlining. When run with all other tests (default 8 threads), too much memory is used and this test takes over a minute.
Function inlining is at 2 iterations currently so tests pass in a reasonable amount of time. Once extraction is less memory-intensive, function inlining should be bumped to 3 iterations (in config.rs).
Currently, extraction uses a lot of memory. Function inlining makes extraction slower. Running
cargo run -r tests/passing/small/fib_recursive.bril --run-mode compile-brilift --optimize-egglog true --optimize-brilift true --interp true
on its own, for example, takes 7 seconds with 3 iterations of inlining. When run with all other tests (default 8 threads), too much memory is used and this test takes over a minute.Function inlining is at 2 iterations currently so tests pass in a reasonable amount of time. Once extraction is less memory-intensive, function inlining should be bumped to 3 iterations (in
config.rs
).