Re: AMD low power hacks
Just confirmed this works on the KT333 as well. Aaron On Tuesday 30 July 2002 03:00 am, Gary Jennejohn wrote: > Michael Nottebrock writes: > > The following is an OpenPGP/MIME signed message > > created by Enigmail/Mozilla, following RFC 2440 and RFC 2015 > > --enig8A086DA17DCB77CC40984CC4 > > Content-Type: text/plain; charset=us-ascii; format=flowed > > Content-Transfer-Encoding: 7bit > > > > I've been wondering lately why my AthlonTB runs at a quite high > > idle-temperature and I came across this page: > > > > http://vcool.occludo.net/VC_Theory.html > > > > Does someone feel like getting something similar into our kernel? > > If you have a VIA KT266A chipset then you can do something like this: > > # turn on HALT bit in register 0x95 of the KT266a -> CPU runs much cooler > # NOTE: the register had 0x1c when I checked it > echo Enable halt bit in KT266A > /usr/sbin/pciconf -w -b pci0:0:0 0x95 0x1e > > which I have in /etc/rc.local. My Athlon runs about 15 C cooler with > this. Bit 1 of register 0x95 controls idling of the CPU. > > Here's a step-by-step description: > > Do the following as root: > 1) pciconf -l -v > this lists all the PCI chipsets found at boot time. I see > > agp0@pci0:0:0: class=0x06 card=0x30991106 chip=0x30991106 rev=0x00 > hdr=0x00 > vendor = 'VIA Technologies Inc' > device = 'VT8366/A Apollo KT266/A,KT333 CPU to PCI Bridge' > class= bridge > subclass = HOST-PCI > > So I have a KT266(A) at pci0:0:0 > > 2) pciconf -r -b pci0:0:0 0x95 > > 0x1c > > Bit 1 isn't set > > 3) pciconf -w -b pci0:0:0 0x95 0x1e > > turns on bit 1. > > > --- > Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] > > > > To Unsubscribe: send mail to [EMAIL PROTECTED] > with "unsubscribe freebsd-current" in the body of the message To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: AMD low power hacks
Michael Nottebrock writes: > The following is an OpenPGP/MIME signed message > created by Enigmail/Mozilla, following RFC 2440 and RFC 2015 > --enig8A086DA17DCB77CC40984CC4 > Content-Type: text/plain; charset=us-ascii; format=flowed > Content-Transfer-Encoding: 7bit > > I've been wondering lately why my AthlonTB runs at a quite high > idle-temperature and I came across this page: > > http://vcool.occludo.net/VC_Theory.html > > Does someone feel like getting something similar into our kernel? > If you have a VIA KT266A chipset then you can do something like this: # turn on HALT bit in register 0x95 of the KT266a -> CPU runs much cooler # NOTE: the register had 0x1c when I checked it echo Enable halt bit in KT266A /usr/sbin/pciconf -w -b pci0:0:0 0x95 0x1e which I have in /etc/rc.local. My Athlon runs about 15 C cooler with this. Bit 1 of register 0x95 controls idling of the CPU. Here's a step-by-step description: Do the following as root: 1) pciconf -l -v this lists all the PCI chipsets found at boot time. I see agp0@pci0:0:0: class=0x06 card=0x30991106 chip=0x30991106 rev=0x00 hdr=0x00 vendor = 'VIA Technologies Inc' device = 'VT8366/A Apollo KT266/A,KT333 CPU to PCI Bridge' class= bridge subclass = HOST-PCI So I have a KT266(A) at pci0:0:0 2) pciconf -r -b pci0:0:0 0x95 0x1c Bit 1 isn't set 3) pciconf -w -b pci0:0:0 0x95 0x1e turns on bit 1. --- Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: AMD low power hacks
Michael Nottebrock wrote: > I've been wondering lately why my AthlonTB runs at a quite high > idle-temperature and I came across this page: > > http://vcool.occludo.net/VC_Theory.html > > Does someone feel like getting something similar into our kernel? Note that this can not be made to be reliable without the confidential errata for A4, A5, A6, A7, and A9. Otherwise, you can/will get spurious system hangs. I rather expect that this is the reason it's "off by default", on most systems, and why there is a "BIOS override" that disables it, on others. -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message