From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Piel Subject: Re: [PATCH] transition latency of speedstep-ich (third try) Date: Tue, 29 Nov 2005 00:35:07 +0100 Message-ID: <438B942B.2090100@tremplin-utc.net> References: <88056F38E9E48644A0F562A38C64FB600671185B@scsmsx403.amr.corp.intel.com> Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: <88056F38E9E48644A0F562A38C64FB600671185B@scsmsx403.amr.corp.intel.com> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: cpufreq-bounces@lists.linux.org.uk Errors-To: cpufreq-bounces+glkc-cpufreq=gmane.org+glkc-cpufreq=gmane.org@lists.linux.org.uk Content-Type: text/plain; charset="iso-8859-1"; format="flowed" To: "Pallipadi, Venkatesh" Cc: cpufreq@lists.linux.org.uk, Dominik Brodowski , davej@redhat.com 28.11.2005 23:19, Pallipadi, Venkatesh wrote/a =E9crit: >>-----Original Message----- >>From: Mattia Dongili [mailto:malattia@linux.it]=20 >> >>I just build speedstep-ich with transition_latency =3D 250 >>I'll let you know how it goes. Are you interested? Actually, the ondemand governor is quite safe on this side and multiply=20 this value by 1000, so it shouldn't bring any problem anyway :-) >>Maybe it's already been discussed, but how bad would be trying to >>measure transition latency at runtime? >> >=20 >=20 >=20 > Yes. I was thinking the same thing. As this latency may be different on P= 3-M and P4-M, may be we can do with a measurement at boot time and use it, = rather than hardcoding these values. However, I think we will have to do th= at by using gettimeofday or some such higherlevel call instead of using pmt= imer as pmtimer may not be available on all systems at all times. Otherwise= , we may end up all sort of pmtimer/pit/hpet timer code in this driver. >=20 Sounds interesting :-) Especially as speedstep-ich already does at least=20 2 frequency transitions while initialising, it should be fairly easy to=20 implement. I'm more concerned about the time measurement. What could be=20 done is to use gettimeofday() and check that the result is reasonable=20 (not too big nor too small), if it's not we just put something=20 conservative, like 500us. Eric