![]() |
2010-04-19
, 16:01
|
Posts: 1,729 |
Thanked: 388 times |
Joined on Jan 2010
@ Canada
|
#3132
|
This is wrong and misleading information!
The min. freq. is the one used for low load, like updating widgets, decoding MP3s etc.
In his case the permanently high load at 1GHz probably killed the battery
but the brain dead 850 minimum is a sign of irresponsibility and how little
he understands this device.
If we had access to the core temperature it would be one factor that determines
the damage of the SOC.
Several people here mistakenly believe that this sensor value would indicate
the CPU temperature. I does NOT!
If your device gets really hot, than it's a sufficient, but not a necessary sign that
you're doing something wrong.
So don't just rely on the temperature.
If you believe temperature doesn't matter, please perform an experiment
and bake your N900 in an oven @200C. It may void your warranty, however...
![]() |
2010-04-19
, 16:17
|
|
Posts: 658 |
Thanked: 117 times |
Joined on Jan 2010
|
#3133
|
The Following User Says Thank You to NokiaRocks For This Useful Post: | ||
![]() |
2010-04-19
, 16:25
|
Posts: 310 |
Thanked: 383 times |
Joined on Jan 2010
|
#3134
|
The N900 doesn't idle at 850 or 500 or 250 or anything else, when idle, the cpu clock simply doesn't run. And I don't think the cpu is even powered.
![]() |
2010-04-19
, 16:29
|
Posts: 310 |
Thanked: 383 times |
Joined on Jan 2010
|
#3135
|
so you are saying that when the cpu temp goes 41 C (like i saw on youtube), its "safe?" coz if that is the case, i will continue or elevate from 850mhz to 1000mhz! my cpu temp goes 38C @ 1Ghz, and 26C @800-900mhz (Lehto's)
im totally having fun with Lehto's kernels. hope his kernels are PR1.2 (or other updates) compatible like Titan's
and the CPU's power supply, is that the battery? (noobe me)
The Following User Says Thank You to nightfire For This Useful Post: | ||
![]() |
2010-04-19
, 16:38
|
Posts: 1,729 |
Thanked: 388 times |
Joined on Jan 2010
@ Canada
|
#3136
|
I believe it's safe for the CPU. It'll shorten its lifespan, but we're talking for 12 years to 10 years (assuming typical <1% high clock rate usage over 24 hours).
The CPU is fed by a switch-mode power supply that converts the main rail's voltage (probably battery voltage) to whatever the VDD1 is configured for. So that could be from 3.4V -> 1.5V. This I would be concerned about frying, not the CPU.
![]() |
2010-04-19
, 16:40
|
Posts: 310 |
Thanked: 383 times |
Joined on Jan 2010
|
#3137
|
This is wrong and misleading information!
The min. freq. is the one used for low load, like updating widgets, decoding MP3s etc.
In his case the permanently high load at 1GHz probably killed the battery
but the brain dead 850 minimum is a sign of irresponsibility and how little
he understands this device.
If we had access to the core temperature it would be one factor that determines
the damage of the SOC.
Several people here mistakenly believe that this sensor value would indicate
the CPU temperature. It does NOT!
If your device gets really hot, than it's a sufficient, but not a necessary sign that
you're doing something wrong.
So don't just rely on the temperature.
If you believe temperature doesn't matter, please perform an experiment
and bake your N900 in an oven @200C. It may void your warranty, however...
![]() |
2010-04-19
, 16:50
|
Posts: 1,729 |
Thanked: 388 times |
Joined on Jan 2010
@ Canada
|
#3138
|
But, rendering said widgets happens 2 or 4 times as fast, limiting the amount of time spent off idle. Sure, it uses more power and thus creates more heat. But nowhere near the amount that running anything significant at even 600mhz does.
That is, playing an mp3 at 850mhz means 5% on CPU, 95% idle. Heat is 5% of maximum dissipation @ 850mhz, so say 3W * .05 = 0.15W.
At 250mhz, it means ~15%, plus the occasional bout up to 850mhz anyway. Say the processor draws 0.5W at full load @ 250mhz. 15% of 0.5W = 0.075W, for a difference of 0.075W thermal dissipation. Hardly anything to worry about.
It may reduce battery life slightly, but I bet 0.075W pales in comparison to the rest of the power draw (radios, DSP, amplifier, etc).
Compare that with running an intensive process at 600mhz. Assume the processor draws 1.5W at this speed. 1.5W @ 100% for 10 minutes represents 0.25Wh of heat dissipation. That's the same as a 5% duty cycle at 850mhz for almost 2 hours. 10 minutes at 600mhz is ~10 times worse.
Of course there is the argument that even though the heat dissipation is low, the temperature changes are more rapid and may lead to premature thermal failure anyway..
![]() |
2010-04-19
, 16:51
|
Posts: 1,255 |
Thanked: 393 times |
Joined on Oct 2009
@ US
|
#3139
|
The Following User Says Thank You to Rushmore For This Useful Post: | ||
![]() |
2010-04-19
, 16:54
|
Posts: 1,729 |
Thanked: 388 times |
Joined on Jan 2010
@ Canada
|
#3140
|
Is there a LV kernel that can be installed as simply as the original OC kernels? I have been using the 125-800 kernel since the first few days the OC kernels were released. It has been super stable, but would like to try a lower volt version.
The kernel I have is nice, since it installed with the N900's built in flash app and no terminal parameters needed (dynamic OC from 125-800).
![]() |
Thread Tools | |
|
The min. freq. is the one used for low load, like updating widgets, decoding MP3s etc.
In his case the permanently high load at 1GHz probably killed the battery
but the brain dead 850 minimum is a sign of irresponsibility and how little
he understands this device.
If we had access to the core temperature it would be one factor that determines
the damage of the SOC.
Several people here mistakenly believe that this sensor value would indicate
the CPU temperature. It does NOT!
If your device gets really hot, than it's a sufficient, but not a necessary sign that
you're doing something wrong.
So don't just rely on the temperature.
If you believe temperature doesn't matter, please perform an experiment
and bake your N900 in an oven @200C. It may void your warranty, however...
Last edited by titan; 2010-04-19 at 16:03.