CPU on Vostro 1500

Any solutions to the CPU whine on the vostro 1500? Searched the forum but couldn't find anything conclusive.

An odd thing about the whine is that when I scroll up and down rapidly using the touch pad the whine goes away.

Thanks

 

It is caused by the CPU entering the C4 sleep state. HP released a BIOS update that fixes it for their machines (presumably by disabling C4 sleep) but there's no fix for Dell. Rapid mouse movement works because it forces the CPU to think enough to not enter C4 sleep.

For some reason, possibly not being 100% compatible with Merom CPU's, switching to Windows XP fixed the problem for me on AC power (I don't recall the noise while running Ubuntu either but didn't run it long enough to be sure), but it still occurs on battery power. Another solution, that does work on battery power, is to run an application that uses a good amount of CPU power in the background. One such example is GPU-z 0.07 (the newer versions fixed this error); another is SkiFree 1.0 (the newer version fixed this as well). Both are free (Google them); run them in the background and you won't get any noise. The tradeoff is significantly reduced battery life.

 

Ok...quick lesson to clear CPU noise talk. (i learned all this trying to configure my linux to for Hpet - now i get 10-12 watts power consumption vs. 15-18)

C4 is an extended version of CPU C3 sleep. C4 is special because it lowers the voltage to the CPU to reduce the amount of power "leaked" by the transistors on the CPU. (note that new penryns have the new High-K insluator that should reduce leaking). This lowering happens many times a second, and for some electro-mechanical reason this high frequency translates into a high pitched noise/EMI.

Apollo 13 is right that when scrolling the CPU cannot sleep long enough to enter C4. (C1 -> C2 -> C3/C4).

Some background info: when a program is running, and requires some sort of function to be performed regularly, they set timers. They tell the OS to set a timer for them.
The OS uses the hardware to set a timer, which will wake the CPU up and run the OS kernel and decide if there are jobs to be done.
Between these timers, and if the CPU doesn't have any jobs currently running, it will sleep, and therefore, if the sleep is long enough (and the time between the wakeups long enough), C4 will be entered.

The hardware timers come in two varieties: HPET and PIT.
PIT can be programmed to interrupt at regular itervals, such as 100hz, 250hz, 500hz, and 1000hz (choices given by the linux kernel config). This means when a program asks a timer to be set, let's say @ 5 times a second, the OS will count each timer "tick" until .2 second has elapsed.

However, there is a new timer, called HPET. it is variable, so it can be set dynamically to whatever time the OS wants.
Therefore, when a program needs a timer, the OS can just set the timer @ that specific interval, and therefore, with our example, @ 5Hz.

======================
Windows XP and Linux 32-bit kernels prior to 2.6.21 do not support HPET, so their processors wake up at a regualar interval. Since this frequency is regular, we interpret the cpu noise, which is generated on a regular interval as "hiss" or something like that.

Windows Vista and Linux 32-bit kernels 2.6.21 and on support HPET (the hardware timer has been around since ICH3), so they can set the timer variably, so with these OSes, you will hear the CPU noise, because it is not regular, and so sounds like "morse code". That is why when you run a program, such as Appolo 13's suggesstions, they will keep the CPU awake and processing, and also set frequent timers to keep the CPU from sleeping.

(for linux users, 64 bit kernels 2.6.24 and on support HPET).

RMCLOCK with HLTCOMMAND - this actually runs a thread that keeps the CPU awake all the time - never lets it sleep.

HPET is advantageous because it lets the CPU sleep longer, and therefore provides extra battery life. - if your programs don't need to wake up very often, then the CPU can sleep for say 30ms at a time, instead of regular PIT interrupts. - perhaps 1 ms sleep.

 

lambchops468,

Good explanation... although it makes me wonder why a MB replacement works for some people... shouldnt all cpus with an HPET timer perform similarly and create the 'morse code' or whine when entering C4 sleep repeatedly.

This seems to be a back and forth... my old 600m centrino used to whine... while my e1505 was always whisper quiet... but the vostro is back to the whining... booted ubuntu 7.10 on the vostro and it was whine free.

 

Hey, Checked and found DELL has released an update for the Vostro 1500 BIOS. Any clues if it is required? and better still, will it nail this whine issue?

 

Ack. So that's why I discovered the whine when I used my Vostro on battery power for the first time. Completely annoying 'cause I used it in another room in the house, which was totally quiet. But that was just about the only time I really heard it loud enough to be annoying so...

 

Me too @ AC Power

@Ckhurana: I still have this problem with the latest A05 bios

 

does this happen on the 1700 aswell?