[LLVMdev] Speeding up llc

Bill Wendling wendling at apple.com
Fri Dec 4 17:41:26 PST 2009

On Nov 19, 2009, at 11:22 AM, Jon Harrop wrote:

> Are there any command-line options I can set to get llc to run faster?
'-O0' should help. Using the local register allocator "regalloc=local"  
will go faster than linearscan, but produce worse code. And other  
similar flags. :-)


More information about the llvm-dev mailing list