Date: Wed, 28 Apr 2004 14:18:26 -0700 From: "Kevin Oberman" <oberman@es.net> To: Nate Lawson <nate@root.org> Cc: Jeffrey Katcher <jmkatcher@yahoo.com> Subject: Re: Fan Control Success on IBM T40? (another quick Q) Message-ID: <20040428211826.A02195D07@ptavv.es.net> In-Reply-To: Your message of "Wed, 28 Apr 2004 13:15:13 PDT." <20040428131247.K88558@root.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> Date: Wed, 28 Apr 2004 13:15:13 -0700 (PDT)
> From: Nate Lawson <nate@root.org>
> Sender: owner-freebsd-acpi@freebsd.org
>
> I forgot to add -- the reason the clock rate announced at boot time is
> different is because the test for CPU TSC is done before acpi is enabled.
> At some point when the SMI is disabled and acpi enabled, the clock rate is
> switched by your BIOS to the lower rate. You can see this because the
> clock rate announced by the TSC timecounter will be ~600 mhz while the
> initial boot clock rate will list ~1600 mhz.
This all makes sense, but it conflicts a bit with my
observations. (Probably implies something bad about my powers of
observation.)
I boot and the system (T30) is running at 1.8 GHz. I throttle the CPU.
Testing clearly shows that the throttling is working. I use the test you
suggested of calculating an MD5 hash of a big string of zeros.
But, when I "count cycles" to test the CPU speed (code appended), I
still see 1.8 GHz.
Why don't I see the speed reduced when throttling? I suspect my lack of
fundamental understanding of the interactions of throttling and the ACPI
clock.
--
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman@es.net Phone: +1 510 486-8634
/*
* $Id: MHz.c,v 1.4 2001/05/21 18:58:01 davej Exp $
* This file is part of x86info.
* (C) 2001 Dave Jones.
*
* Licensed under the terms of the GNU GPL License version 2.
*
* Estimate CPU MHz routine by Andrea Arcangeli <andrea@suse.de>
* Small changes by David Sterba <sterd9am@ss1000.ms.mff.cuni.cz>
*
*/
#include <stdio.h>
#include <sys/time.h>
#include <string.h>
#include <unistd.h>
__inline__ unsigned long long int rdtsc()
{
unsigned long long int x;
__asm__ volatile (".byte 0x0f, 0x31" : "=A" (x));
return x;
}
char *estimate_MHz()
{
char *buffer;
struct timezone tz;
struct timeval tvstart, tvstop;
unsigned long long int cycles[2]; /* gotta be 64 bit */
unsigned int microseconds; /* total time taken */
memset(&tz, 0, sizeof(tz));
/* get this function in cached memory */
gettimeofday(&tvstart, &tz);
cycles[0] = rdtsc();
gettimeofday(&tvstart, &tz);
/* we don't trust that this is any specific length of time */
usleep(100);
cycles[1] = rdtsc();
gettimeofday(&tvstop, &tz);
microseconds = ((tvstop.tv_sec-tvstart.tv_sec)*1000000) +
(tvstop.tv_usec-tvstart.tv_usec);
buffer = malloc(sizeof(char)*512);
sprintf(buffer, "%lldMHz",
(cycles[1]-cycles[0])/microseconds);
return buffer;
}
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040428211826.A02195D07>