Overclock the N900? - Page 94 - maemo.org

Rushmore | # 931 | 2010-04-05, 13:30 | Report

Flandry, you can not argue against real science and you type the truth. a key point:

"So while there is a linear scaling for temperature within a given range of frequencies, there's an exponential decline in lifetime."

That is my point from another perspective of it (in regards to a result)

The point for which the temps spike significantly (five to ten degrees), is the operational ceiling and also the point for exponential / accelerated lifespan decline.

We are speaking the same thing, I am just trying to "hillbilly" it down a notch

I am a hillbilly BTW

Added:

I do not think the N900 will have a profound lifespan issue at 800mhz as a max clock- based on Droid testing and me not noticing a temp increase or additional battery drain (compared to 600mhz). My guess is the 125mhz min is offsetting the battery, since my device spends more time idle than active.

Last edited by Rushmore; 2010-04-05 at 13:35. Reason: Added

LondonBenji | # 932 | 2010-04-05, 13:31 | Report

So far the only convincing argument I have seen for NOT o/c the phone is Igor's statement because if you aren't prepared to believe a statement from a developer IN Nokia, then what single piece of written text on a screen would an end user who most likely hasn't got any direct contact with designing devices that make use of the processor believe?

However, one other thing still bugs me which has only really briefly been mentioned (but only in passing), if we can significantly reduce the lifetime of the processor by even simply running it at 550-600MHz all the time, then why when you connect the charger does it stop the scaling and keep the proc at the maximum 600MHz?

I don't know about anyone else but I charge my phone EVERY night and in normal usage realistically, twice a day..... that means my proc spends most of it's time at 600MHz (now 900MHz).....

This is the only thing that makes me feel it isn't so much of an issue otherwise when charging I would have thought perhaps they'd lock it to a lower frequency?

The Following User Says Thank You to LondonBenji For This Useful Post:
Faz

6sicSIX | # 933 | 2010-04-05, 13:33 | Report

Does anyone know if it's possible to check temperature of the n900 from some internal sensor?
I had mine oc'd to 900 and left it booting microxp in bochs, it was very hot when I returned so i've clocked it back down to 800 for now..

xgrind | # 934 | 2010-04-05, 13:34 | Report

While I appreciate those who tried to highlight the dangers of overclocking, my opinion is most of those who had plunged in and OCed their N900, they have already understood the potential risk and consequences from doing so and should not blame anyone if s**t happens.

I understand the risk and willing to trade a shorter phone lifespan for a better user experience with the phone

I just hope this thread will continue to flourish with new ways to improve the performance instead of ending up a flame thread.

Just my 2 cents

The Following 3 Users Say Thank You to xgrind For This Useful Post:
kmare, rpgAmazon, UNderworld

jaeezzy | # 935 | 2010-04-05, 13:35 | Report

BTW, do we have to manually set scaling_max_freq to 900000 if we are using 900MHz kernel in /etc/pmconfig? thanks.

_privateMember | # 936 | 2010-04-05, 13:43 | Report

Originally Posted by javispedro View Post

125Mhz comes from https://bugs.maemo.org/show_bug.cgi?id=7116 (as you can see, testing firmwares had scaling_min_freq set to 125Mhz)..

Just tested that on my N900 with the 125-900 oc kernel.
Got the result:
Nokia-N900-51-1:/# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies
900000 550000 500000 250000 125000

According to #7116 the 125000 shouldn't be listed. I can't reproduce the error. Or am I missing something?

Serge | # 937 | 2010-04-05, 13:46 | Report

Originally Posted by Rushmore View Post

Flandry, you can not argue against real science and you type the truth. a key point:

"So while there is a linear scaling for temperature within a given range of frequencies, there's an exponential decline in lifetime."

That is my point

The point for which the temps spike significantly (five to ten degrees), is the operational ceiling and also the point for exponental / accelerated lifespan decline.

We are speaking the same thing, I am just trying to "hillbilly" it down a notch

I am a hillbilly BTW

There is an official datasheet from TI (for omap3530 which is very similar to omap3430) with the information about the expected lifespan. Running the chip all the time at 600MHz vs. 500MHz already decreases lifespan twice, hence 600MHz is considered overdrive:

Code:

To avoid significant device degradation for commercial temperature OMAP3530/OMAP3525 devices (0°C
≤ Tj ≤ 90°C), the device power-on hours (POH) must be limited to one of the following:
• 100K total POH when operating across all OPPs and keeping the time spent at OPP5-OPP6 to less
than 23K POH.
• 50K total POH when operating at OPP5 - OPP6.
• 44K total POH with no restrictions to the proportion of these POH at operating points OPP1 - OPP6.

If this effect is exponential, then pushing the chip above 600MHz may reduce lifespan quite dramatically. I think that TI would sell chips labeled at 1GHz if they were sure that this clock frequency is safe, it is in their best interests after all, considering the tough competition.

Those who buy new smartphones each half a year may not care much, but I feel sorry for those who would buy a second hand N900, heavily worn out by some overclocker.

The Following 9 Users Say Thank You to Serge For This Useful Post:
allnameswereout, apollovy, Bundyo, buxz777, Chrome, cibyr, joerg_rw, sjgadsby, Venemo

sadfist | # 938 | 2010-04-05, 13:52 | Report

44,000 power on hours is 5 years. I'm OK with that.

The Following User Says Thank You to sadfist For This Useful Post:
cibyr

Rushmore | # 939 | 2010-04-05, 13:55 | Report

Originally Posted by Serge View Post

There is an official datasheet from TI (for omap3530 which is very similar to omap3430) with the information about the expected lifespan. Running the chip all the time at 600MHz vs. 500MHz already decreases lifespan twice, hence 600MHz is considered overdrive:

Code:

To avoid significant device degradation for commercial temperature OMAP3530/OMAP3525 devices (0°C
≤ Tj ≤ 90°C), the device power-on hours (POH) must be limited to one of the following:
• 100K total POH when operating across all OPPs and keeping the time spent at OPP5-OPP6 to less
than 23K POH.
• 50K total POH when operating at OPP5 - OPP6.
• 44K total POH with no restrictions to the proportion of these POH at operating points OPP1 - OPP6.

If this effect is exponential, then pushing the chip above 600MHz may reduce lifespan quite dramatically. I think that TI would sell chips labeled at 1GHz if they were sure that this clock frequency is safe, it is in their best interests after all, considering the tough competition.

Those who buy new smartphones each half a year may not care much, but I feel sorry for those who would buy a second hand N900, heavily worn out by some overclocker.

Good perspective. It would be cool if there was an actual, "functioning hours" to define the life. Point is the actual life of the device depends on actual usage. I use the N900 along with my phone, so the N900 is mainly a netbook and media player, though MAME and PSX are creeping in now....

If a person uses the N900 100% as their main device and use the "heck out of it", OC could be an issue. Then again, how much is the 125mhz difference in min clock helping to offset the increase? My feelings are that an 800mhz clocked 3430 with "normal" use for me will last about as long as the 600mhz clocked will (since my device spends nmore time idle than active).

Edit:

There IS an hours of use defined with the tech spec (duh)

Last edited by Rushmore; 2010-04-05 at 13:59.

The Following User Says Thank You to Rushmore For This Useful Post:
Faz

Serge | # 940 | 2010-04-05, 13:56 | Report

Originally Posted by sadfist View Post

44,000 power on hours is 5 years. I'm OK with that.

It's 5 years only if you never clock it higher that 600MHz (or 720MHz for the latest omap3530 revisions). If each extra 100MHz decrease lifespan twice, you can do the math about how long it will last running at clock speeds up to 900-1000MHz.