Date: Thu, 24 Jan 2013 12:30:36 -0800 From: Adrian Chadd <adrian@freebsd.org> To: Jason Evans <jasone@freebsd.org> Cc: freebsd-arch@freebsd.org Subject: Re: [rfc] enabling MALLOC_PRODUCTION on -HEAD for now, until jemalloc has been taught to have some run time selectable debug options Message-ID: <CAJ-Vmok_6bSVg6%2B1VFypQQ1KZrDPkDJiP3p%2BespJbMXeiESLgQ@mail.gmail.com> In-Reply-To: <A803FFD3-E195-43C9-8B5D-24664707584C@freebsd.org> References: <CAJ-VmomY_jy5s_pgjpjDXZpN54HpKykD-5tWjU6TG6Z7eR=eOQ@mail.gmail.com> <A803FFD3-E195-43C9-8B5D-24664707584C@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 24 January 2013 10:40, Jason Evans <jasone@freebsd.org> wrote: > I imported jemalloc 3.3.0 last night, which I think will restore non-MALL= OC_PRODUCTION performance to roughly what it has been for years. The key c= hange was a patch by Ian Lepore: > > http://www.canonware.com/cgi-bin/gitweb.cgi?p=3Djemalloc.git;a=3D= commitdiff;h=3D14a2c6a698a207ac3f3825443cf3441c8842e990 > > In practice, the zeroed memory validation should no longer execute at all= on FreeBSD due to madvise(...MADV_FREE) not guaranteeing zeroed memory. > > I don't know how significant the impact will be on constrained/older syst= ems, but for the many buildworld repetitions I did during testing it improv= ed throughput by >1.5X. With this change in place, I'm not keen on enablin= g MALLOC_PRODUCTION by default, because the assertions are useful both for = detecting bugs in jemalloc (heaven forbid), and for indicating that crashes= within jemalloc are application errors. The only other significant source= of overhead I'm aware of is junk filling, which we've had enabled by defau= lt on -current for ages, probably since the introduction of phkmalloc. If = you don't like it: > > ln -s 'junk:false' /etc/malloc.conf > > If a substantial performance difference remains for you (MALLOC_PRODUCTIO= N vs non-MALLOC_PRODUCTION with junk:false set), please let me know. Sweet, thanks. I'll do up a new -HEAD MIPS, ARM and PPC test builds tonight and let you know how it all goes. I'd love to keep MALLOC_PRODUCTION off on -HEAD by default. :-) Thank you to both you and Ian for spearheading this change to your malloc library. :) Adrian
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-Vmok_6bSVg6%2B1VFypQQ1KZrDPkDJiP3p%2BespJbMXeiESLgQ>