Re: Performance assistance / advice
"Marco Antoniotti (as marco dot antoniotti at unimib dot it)" <[email protected]>
| Newsgroups | gmane.lisp.lispworks.general |
|---|---|
| Message-ID | <CAG0Nw2=dLjju4sPfCDJ8YeN38NY6Ggk8eWyyhpgJuaY0ECGxcw@mail.gmail.com> |
... which boil down to: there are two godzillion people working on C compilers (*) and two on Common Lisp compilers. Cheers MA (*) Ok: half a godzillion: the rest is now working on Rust. On Tue, Apr 15, 2025 at 3:24 PM Tim Bradshaw (as tfb at tfeb dot org) < [email protected]> wrote: > On 15 Apr 2025, at 13:49, Yuri Davidovsky (as work at disclosure dot ie) < > [email protected]> wrote: > > > > Not bad. I assume the best lisp performance version would be sbcl? How > does the C code fare without optimisations enabled for the corrected code? > > Yes, SBCL: LW is OK but somewhat slower (which is fine: LW has other > advantages!) > > With no optimisation flags C is doing about 2.2GFLOPS. There's not much > difference between -O and -O3. > > My uninformed suspicion is that C is beating SBCL because the C compiler > is emitting fmadd instructions, and there's an awful lot of x -> x + a * b > in the code of course. SBCL doesn't seem to emit those. But I'm not sure. > > --tim > > _______________________________________________ > Lisp Hug - the mailing list for LispWorks users > [email protected] > http://www.lispworks.com/support/lisp-hug.html > -- Marco Antoniotti, Professor, Director tel. +39 - 02 64 48 79 01 DISCo, University of Milan-Bicocca U14 2043 http://dcb.disco.unimib.it Viale Sarca 336 I-20126 Milan (MI) ITALY CSCE 2025 - csce.lakecomoschool.org