Pretty much. In Docker it’s a much smaller difference than it is on Mac using the homebrew version. Looks like this PR will fix the issues though.
I ran the test from to_i64 takes very long to release-compile · Issue #9654 · crystal-lang/crystal · GitHub with the newly released 0.36.1 and it seems to deff be faster in the codegen bc+obj phase; which took 8.5s versus 39.5s on 0.36.0. Also used just under 2x less memory. I’d be curious to see what affect LLVM11 would have over 10.
Just an update here that 0.36.1 did indeed resolve the issues! 
Great work team! 