On Mon, Nov 5, 2012 at 7:58 AM, Matthew Flatt <mfl...@cs.utah.edu> wrote: > At Sun, 4 Nov 2012 21:03:15 -0500, Sam Tobin-Hochstadt wrote: >> > but, anyway... I think >> > that benchmark turns out to measure mostly allocation. Racket in 32-bit >> > mode, where pair and vectors take up half as much space, runs almost >> > twice as fast as Racket in 64-bit mode. >> >> Is the Gambit allocator that much faster than ours? Or does it use >> less memory for pairs and vectors? > > I think Gambit uses less memory for pairs and vector --- one less word > per object. > > That wouldn't explain the x2.2 difference, and on further > investigation, it turns out that I was running a 32-bit Gambit build. > Enclosed is a comparison of 32-bit Gambit and 32-bit Racket. (Switching > my Chicken installation seems like too much work for a sanity check.) > You'll see that Racket still x1.41 Gambit's time for "paraffins", which > is a closer match for the difference in allocation sizes.
Thanks for the extra investigation. It's impressive to note the change in relative performance for Racket over past 3 years since you published the benchmarks on the blog -- Racket has gone from slower than Gambit on the majority of benchmarks, sometimes by a significant margin, to faster on most of them, and never more than 2x slower (except ctak). Thanks, Matthew! ____________________ Racket Users list: http://lists.racket-lang.org/users
