The Great Startup Problem

Per Bothner per at
Sat Aug 23 19:25:37 UTC 2014

On 08/22/2014 01:08 PM, Charles Oliver Nutter wrote:
> What are the rest of you doing to deal with these issues?

Start-up does not appear to a problem for Kawa:

I tried running the 'meteor' benchmark program ('2098 solutions found').
which is 29352 characters (844 lines) long.  On my i7-4700MQ laptop
running a pre-compiled meteor.class the time command reports:
0.57user 0.04system 0:00.34elapsed 180%CPU (0avgtext+0avgdata 37852maxresident)k
0inputs+0outputs (0major+9390minor)pagefaults 0swaps

Running from meteor.scm source (read, parse, compile, defineClass, execute), I get:
2.68user 0.09system 0:00.97elapsed 284%CPU (0avgtext+0avgdata 71124maxresident)k
0inputs+0outputs (0major+24436minor)pagefaults 0swaps

What kind of start-up times are you talking about?  Is it a problem even for
smallish programs (like meteor), or is the problem with significantly larger programs?
	--Per Bothner
per at

More information about the mlvm-dev mailing list