Re: Elisp native-comp vs. SBCL for inclist-type-hints benchmark (was: [PATCH] Re: Bignum performance)

[Emanuel Berg <incal@dataswamp.org>]

Ihor Radchenko wrote:

> I did, and the results are very different from yours

It is probably because of the Emacs batch and/or the SBCL
non-SLIME style of execution, using commands similar to
yours [last] I get the following results.

Elisp vs SBCL

inclist:            (faster 1.04 1.59) ; CL 35% slower
inclist-type-hints: (faster 1.05 0.69) ; CL 52% faster

Elisp optimization: (faster 1.04 1.05) ; Elisp 1% slower from optimization
CL optimization:    (faster 1.59 0.69) ; CL 130% faster from optimization

We see that, surprisingly, CL is slower for plain inclist.

With type hints tho, CL benefits hugely to beat the Elisp
non-optimized record.

While Elisp doesn't seem to benefit from the optimization
at all.

#! /bin/zsh
#
# this file:
#   https://dataswamp.org/~incal/cl/bench/inc2-cl

sbcl --noinform --load inclist.cl --load inclist-type-hints.cl --quit

#! /bin/zsh
#
# this file:
#   https://dataswamp.org/~incal/cl/bench/inc2-el

emacs                                                            \
    -batch                                                       \
    -l ~/.emacs.d/elpa/elisp-benchmarks-1.14/elisp-benchmarks.el \
    --eval '(setq elb-speed 2)'                                  \
    --eval '(elisp-benchmarks-run "inclist")'

-- 
underground experts united
https://dataswamp.org/~incal