Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 03 Feb 2019 20:36:24 +0300
From:      Greg V <greg@unrelenting.technology>
To:        John F Carr <jfc@mit.edu>
Cc:        freebsd-arm@freebsd.org
Subject:   Re: Scheduling on heterogeneous ARM SoC
Message-ID:  <1549215384.4076.0@smtp.migadu.com>
In-Reply-To: <34290296-577D-4A05-9FC8-1142A47909F5@exchange.mit.edu>
References:  <34290296-577D-4A05-9FC8-1142A47909F5@exchange.mit.edu>

next in thread | previous in thread | raw e-mail | index | archive | help


On Sun, Feb 3, 2019 at 5:29 PM, John F Carr <jfc@mit.edu> wrote:
> Some ARM-based single board computers combine two types of Cortex=20
> processors, for example 4 A73 and 4 A 53 or 2 A72 and 4 A53.  How=20
> does FreeBSD schedule processes on these chips?  Does it use all=20
> cores at random, only the "big" or "little" set, or something more=20
> complicated?

Pretty much "at random" i.e. doesn't take the differences between cores=20
into account.

I'd really like to have a "fill up fastest cores first" mode, because=20
for building software on my RK3399 board I've had to either not do=20
anything and sometimes end up with big cores idling when little cores=20
are compiling, or to cpuset to big cores which would end up not=20
utilizing the little cores in highly parallel build phases=85

=




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1549215384.4076.0>