 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Yes, that's right. THE FAN IS CONTROLLED IN SOFTWARE! How retarded is
> that? Surely this is a serious fire hazard...
No, because the machine will switch off (which is in hardware) long before
the chance of any fire breaking out.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
scott wrote:
>> Yes, that's right. THE FAN IS CONTROLLED IN SOFTWARE! How retarded is
>> that? Surely this is a serious fire hazard...
>
> No, because the machine will switch off (which is in hardware) long
> before the chance of any fire breaking out.
As I say, the laptop let itself get hot enough to randomly malfunction
without turning off. What makes you think it would turn off at some even
higher temperature?
(A more likely scenario is that whatever program it's running would
crash, causing the CPU to halt and therefore cool down again.)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> No, because the machine will switch off (which is in hardware) long
>> before the chance of any fire breaking out.
>
> As I say, the laptop let itself get hot enough to randomly malfunction
> without turning off. What makes you think it would turn off at some even
> higher temperature?
Because that's what CPUs are designed to do. It was probably the RAM that
was overheating giving corrupt data to the CPU.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 7/14/2010 3:08 AM, Invisible wrote:
> Yes, that's right. THE FAN IS CONTROLLED IN SOFTWARE! How retarded is
> that? Surely this is a serious fire hazard...
Most can be controlled by software. Usually the BIOS is in charge, but
programs can also alter the fan speed. This is useful for more advanced
monitoring programs like SpeedFan.
My system at home has 4 fan ports on the motherboard (3 case, one CPU,
the CPU and 2 of the case fans can be controlled by PWM)
--
~Mike
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 7/14/2010 4:36 AM, scott wrote:
> Because that's what CPUs are designed to do. It was probably the RAM
> that was overheating giving corrupt data to the CPU.
Yep... There's usually a thermal sensor either on the die or on the
motherboard.
Some older AMD's I believe rely on the mobo's thermal shutdown. an AMD
chip can be damaged by heat before the mobo registers a problem. On the
Intel side, the chip will throttle down to a snails pace to try to
control the heat, if that doesn't work, it'll tell the mobo to shut off.
I'm not sure about the newer AMDs.
Another interesting thing that's fun to watch: On the later Intel Core2
chips, they have an adaptive multiplier. When the CPU is idle, it will
reduce its clock rate to save on energy and keep cooler when idle.
--
~Mike
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Because that's what CPUs are designed to do. It was probably the RAM
>> that was overheating giving corrupt data to the CPU.
>
> Yep... There's usually a thermal sensor either on the die or on the
> motherboard.
>
> Some older AMD's I believe rely on the mobo's thermal shutdown. an AMD
> chip can be damaged by heat before the mobo registers a problem.
http://www.tomshardware.com/reviews/hot-spot,365-5.html
demonstrates how older AMDs had *no* thermal protection at all. They
even have a video of one of the CPUs actually smoking and glowing. Can
you spell "fire hazard"?
> I'm not sure about the newer AMDs.
They added thermal protection a long, long time ago.
> Another interesting thing that's fun to watch: On the later Intel Core2
> chips, they have an adaptive multiplier. When the CPU is idle, it will
> reduce its clock rate to save on energy and keep cooler when idle.
Yeah, most processors do that now (especially "mobile" ones). Initially
they just made it so that a certain temperature, the system powers off.
But later they had adjustable fan speeds and adjustable clock speeds,
and switching off cores and so on.
I'm still puzzled about this laptop though. Considering that fan control
is both trivially simple and utterly safety critical, why is it
controlled by the OS? Why isn't it hard-wired?
Post a reply to this message
|
|
| |
| |
|
|
From: Mike Raiford
Subject: Re: My computer doesn't like heat, what to do?
Date: 14 Jul 2010 10:46:09
Message: <4c3dcdb1@news.povray.org>
|
|
|
| |
| |
|
|
On 7/14/2010 7:34 AM, Invisible wrote:
> I'm still puzzled about this laptop though. Considering that fan control
> is both trivially simple and utterly safety critical, why is it
> controlled by the OS? Why isn't it hard-wired?
Generally the OS isn't in charge of the fan. Perhaps the linux distro
you're using detects the chipset and grabs control of the fan?
--
~Mike
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> I'm still puzzled about this laptop though. Considering that fan control
>> is both trivially simple and utterly safety critical, why is it
>> controlled by the OS? Why isn't it hard-wired?
>
> Generally the OS isn't in charge of the fan. Perhaps the linux distro
> you're using detects the chipset and grabs control of the fan?
Why should it even be possible to control it in software? The correct
algorithm is simple: set the fan to the lowest setting that gives
acceptable cooling. Why would you want to be able to control it in
software at all?
(As it happens, once Linux was installed, the fan worked normally. Just
not while the installer was running...)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Why should it even be possible to control it in software?
Because there's usually an IC on the motherboard that reads the temperatures
and controls the fan PWM. This is mapped to memory so that the BIOS can
control it. Obviously this means any software running (an OS, a special
program, etc) can also take control.
> The correct algorithm is simple: set the fan to the lowest setting that
> gives acceptable cooling.
That's what it does! But you have to take into account that there are many
different fan and CPU combinations that will be used on a motherboard, not
giving the option to change the settings would be a bit limiting. (Plus it
would make it hard for the BIOS to setup the IC, currently it just writes
the data to a certain memory address).
> Why would you want to be able to control it in software at all?
How else do you propose doing it, that isn't 10x more expensive than the
current method?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
scott <sco### [at] scott com> wrote:
> The fan/BIOS is probably designed to run "silently" at CPU-idle and only
> speed up once it gets loaded. They are probably doing this by triggering at
> speed trigger temperatures?
Actually someone else (IRL) suggested checking the BIOS settings for the
CPU fan, so I did so.
There's some kind of setting which makes the CPU fan target for a certain
maximum CPU temperature. This setting was disabled, so I don't know what the
CPU fan was doing. I tried enabling this (and the default target maximum
temperature seems to be 66 degrees, so I left it like that), and now the
PC is incredibly silent, yet the CPU keeps quite cool when idle (2500 RPM,
47 degrees celsius) even though the room temperature is still unbearably
high (30 degrees celsius at the moment).
I'm monitoring closely the CPU temperature and fan speed, but it seems to
be quite stable. And incredibly silent.
--
- Warp
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
