From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757876Ab0JSVSZ (ORCPT ); Tue, 19 Oct 2010 17:18:25 -0400 Received: from rcsinet10.oracle.com ([148.87.113.121]:52052 "EHLO rcsinet10.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754774Ab0JSVSX (ORCPT ); Tue, 19 Oct 2010 17:18:23 -0400 Date: Tue, 19 Oct 2010 14:16:27 -0700 From: Randy Dunlap To: Bruno Randolf Cc: akpm@linux-foundation.org, kevin.granade@gmail.com, blp@cs.stanford.edu, linux-kernel@vger.kernel.org Subject: Re: [PATCH v2] Add generic exponentially weighted moving average function Message-Id: <20101019141627.3bac9593.randy.dunlap@oracle.com> In-Reply-To: <20101019083635.32294.67087.stgit@localhost6.localdomain6> References: <20101019083635.32294.67087.stgit@localhost6.localdomain6> Organization: Oracle Linux Eng. X-Mailer: Sylpheed 2.7.1 (GTK+ 2.16.6; x86_64-unknown-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 19 Oct 2010 17:36:35 +0900 Bruno Randolf wrote: > This adds a generic exponentially weighted moving average function. This > implementation makes use of a structure which keeps a scaled up internal > representation to reduce rounding errors. > > The idea for this implementation comes from the rt2x00 driver (rt2x00link.c) > and I would like to use it in several places in the mac80211 and ath5k code. > > Signed-off-by: Bruno Randolf Reviewed-by: Randy Dunlap > -- > v2: Renamed 'samples' to 'weight'. Added more documentation. Use avg_val > pointer. Add a WARN_ON_ONCE for invalid values of 'weight'. Divide > and round up/down. > --- > include/linux/average.h | 37 +++++++++++++++++++++++++++++++++++++ > 1 files changed, 37 insertions(+), 0 deletions(-) > create mode 100644 include/linux/average.h > > diff --git a/include/linux/average.h b/include/linux/average.h > new file mode 100644 > index 0000000..55e4317 > --- /dev/null > +++ b/include/linux/average.h > @@ -0,0 +1,37 @@ > +#ifndef _LINUX_AVERAGE_H > +#define _LINUX_AVERAGE_H > + > +#define AVG_FACTOR 1000 > + > +struct avg_val { > + int value; > + int internal; > +}; > + > +/** > + * moving_average() - Exponentially weighted moving average (EWMA) > + * @avg: Average structure > + * @val: Current value > + * @weight: This defines how fast the influence of older values decreases. > + * Has to be higher than 1. Use the same number every time you call this > + * function for a single struct avg_val! > + * > + * This implementation make use of a struct avg_val which keeps a scaled up > + * internal representation to prevent rounding errors. Due to this, the maximum > + * range of values is MAX_INT/(AVG_FACTOR*weight). > + * > + * The current average value can be accessed by using avg_val.value. > + */ > +static inline void > +moving_average(struct avg_val *avg, const int val, const int weight) > +{ > + if (WARN_ON_ONCE(weight <= 1)) > + return; > + avg->internal = avg->internal ? > + (((avg->internal * (weight - 1)) + > + (val * AVG_FACTOR)) / weight) : > + (val * AVG_FACTOR); > + avg->value = DIV_ROUND_CLOSEST(avg->internal, AVG_FACTOR); > +} > + > +#endif /* _LINUX_AVERAGE_H */ > --- ~Randy *** Remember to use Documentation/SubmitChecklist when testing your code ***