Running SpeedFan at startup on XP

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
quietdragon
Posts: 31
Joined: Thu Sep 18, 2003 9:04 pm
Location: Vancouver, BC, Canada

Running SpeedFan at startup on XP

Post by quietdragon » Wed Feb 25, 2004 5:09 pm

Does anyone know if it's possible to get SpeedFan to run immediately on startup on XP?

In its standard configuration, SpeedFan only starts when a user logs in.

This means that any fans it controls spin noisily at 100% (with an idle CPU) until someone logs in.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Wed Feb 25, 2004 5:32 pm

It's pretty simple to get any program that runs from an executable to run as a service using MS's netsvc. Rather than explaining it all here, just read and follow this guide, just replace Speedfan.exe for the F@H exe.

Overclockers.com guide to running F@H as a service

(if that doesn't make sense, or it doesn't work, let me know :lol: )

quietdragon
Posts: 31
Joined: Thu Sep 18, 2003 9:04 pm
Location: Vancouver, BC, Canada

Post by quietdragon » Thu Feb 26, 2004 12:27 am

Rusty075 wrote:It's pretty simple to get any program that runs from an executable to run as a service using MS's netsvc. Rather than explaining it all here, just read and follow this guide, just replace Speedfan.exe for the F@H exe.

Overclockers.com guide to running F@H as a service

(if that doesn't make sense, or it doesn't work, let me know :lol: )
I followed the instructions, and can start the SpeedFan service manually provided I check the box "Allow service to interact with the desktop". If I don't check this box, SpeedFan doesn't get anywhere.

However, when I reboot the computer and wait at the login window, I eventually hear a "beep" --- but don't see anything.

Logging in, I then see the dialog box:

SpeedFan: System Error: Code 5: Access Denied.

Clicking Ok, SpeedFan proceeds through to starting.

So it seems SpeedFan has started, but is waiting for me to acknowledge the error!

Any ideas?

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Thu Feb 26, 2004 1:51 am

It seems that speedfan really wants to draw things on the desktop, and since you don't have a user desktop yet it will get an error when trying to do this.
To run things as a service you must stop the program from interacting with the desktop.

quietdragon
Posts: 31
Joined: Thu Sep 18, 2003 9:04 pm
Location: Vancouver, BC, Canada

Post by quietdragon » Sat Feb 28, 2004 12:41 am

silvervarg wrote:It seems that speedfan really wants to draw things on the desktop, and since you don't have a user desktop yet it will get an error when trying to do this.
To run things as a service you must stop the program from interacting with the desktop.
I couldn't get SpeedFan to do what I wanted --- it evidently keeps wanting to scribble on the desktop even when there isn't one, and appears stops running when I log out even when it's a service.

I subsequently discovered FanSpeed at http://www.twistedfury.com/fanspeed/ which sensibly is written in two parts: a service to control the fans, and a GUI to control the service.

Luckily it works with the NF7-S which has a Winbond W83627HF controller.

Somehow the configuration got messed up, and I had to edit GUINT.cfg to read:

[RpmMap]
Fan1Rpm = "1"
Fan2Rpm = "2"
Fan3Rpm = "3"
Fan4Rpm = "4"

(During installation Fan1Rpm got set to 2, and Fan2Rpm got set to 1! This ended up confusing me and the program.)

This program isn't as mature as SpeedFan or MBM, so lacks many of their features. In particular the FanSpeed GUI doesn't divide down the fan pulse counts, resulting in reported RPMs that are way too high.

For the moment, I run the FanSpeed service to control the fans immediately the OS starts, and use the SpeedFan GUI after login to display the current speeds and temperatures in the tray.

I'm not sure if the W83627HF can or will get confused if the two programs overlap when accessing the registers.

clive
Posts: 47
Joined: Mon Jan 26, 2004 3:27 pm
Location: Sydney Australia

Post by clive » Sat Feb 28, 2004 2:50 am

I am running both Speedfan and Fanspeed in win2k with no problems

I like the GUI in Fanspeed and I find the control of the fanspeeds better than Speedfan as it allows a progressive increase/decrease rather than the more abrubt change in Speedfan

PorBleemo
Posts: 90
Joined: Fri Jan 09, 2004 11:23 am

Post by PorBleemo » Sat Feb 28, 2004 5:53 am

That FanSpeed is an interesting program. I think I'll give it a spin. But as to running SpeedFan at start up you need to paste a shortcut to the program in the Start folder. Then right click the shortcut and then click Properties. Then select the Run as Minimized. This will have it start in the taskbar.

-Por

quietdragon
Posts: 31
Joined: Thu Sep 18, 2003 9:04 pm
Location: Vancouver, BC, Canada

Post by quietdragon » Sat Feb 28, 2004 6:48 pm

PorBleemo wrote:That FanSpeed is an interesting program. I think I'll give it a spin. But as to running SpeedFan at start up you need to paste a shortcut to the program in the Start folder. Then right click the shortcut and then click Properties. Then select the Run as Minimized. This will have it start in the taskbar.
-Por
Thanks for the tip. What I want is to control fan speeds from boot.

Shortcuts you put in the Start folder do not take effect until you login. For my purposes that way too late. The fans spin at 100% whilst no-one is logged in and the computer is mostly idle!

quietdragon
Posts: 31
Joined: Thu Sep 18, 2003 9:04 pm
Location: Vancouver, BC, Canada

Post by quietdragon » Sat Feb 28, 2004 11:14 pm

clive wrote:I am running both Speedfan and Fanspeed in win2k with no problems

I like the GUI in Fanspeed and I find the control of the fanspeeds better than Speedfan as it allows a progressive increase/decrease rather than the more abrubt change in Speedfan
I did some research, and there is a potential conflict when running two monitoring programs using on the WinBond chipset. I looked specifically at the WinBond W83627HF.

The problem is that this chip (and presumably others in this family) uses two IO ports to access a myriad of internal registers. You write the internal address of the register into the first IO port, and read/write the data you want in the second IO port.

Problems arise if one monitoring program gets interrupted in-between the two IO port accesses, and the other monitoring program gets to perform its access. When control returns to the first monitoring program, it will very likely read/write from the wrong address!

MBM and some other programs use a mutex as described in http://mbm.livewiredev.com/mutex.html to collaborate and not interfere with each other at this critical time.

This thread http://www.livewiredev.com/bbs/showthre ... eadid=5441 says that MBM and SpeedFan will play together.

My analysis of FanSpeed shows that it doesn't use this mutex, and so will not collaborate properly with either MBM and SpeedFan :(

clive
Posts: 47
Joined: Mon Jan 26, 2004 3:27 pm
Location: Sydney Australia

Post by clive » Sat Feb 28, 2004 11:23 pm

quietdragon wrote: I did some research, and there is a potential conflict when running two monitoring programs using on the WinBond chipset. I looked specifically at the WinBond W83627HF.
I have not had a conflict yet running Fanspeed and Speedfan. I do not have Speedfan controlling the fans
quietdragon wrote: This thread http://www.livewiredev.com/bbs/showthre ... eadid=5441 says that MBM and SpeedFan will play together.

My analysis of FanSpeed shows that it doesn't use this mutex, and so will not collaborate properly with either MBM and SpeedFan :(
This is opposite of my experience - I had problems with MBM and SpeedFan

quietdragon
Posts: 31
Joined: Thu Sep 18, 2003 9:04 pm
Location: Vancouver, BC, Canada

Post by quietdragon » Sun Feb 29, 2004 12:34 am

clive wrote:I have not had a conflict yet running Fanspeed and Speedfan. I do not have Speedfan controlling the fans
I'm running both together too, and haven't noticed anything yet. FanSpeed controls the fans, and SpeedFan monitors the readings. However, just because we haven't seen any conflict yet doesn't mean it doesn't exist!

The possibility of conflict exists (as per my description above), but the probability of conflict is low. Presumably SpeedFan and FanSpeed will access the WinBond registers at most a few times per second, and the programs are not synchronised with each other.

The window of conflict is only a few machine instructions wide.

So in their current state, if the programs run long enough, they will hit the window, and either get confused, or worse.

Post Reply