Active Topics

 


Reply
Thread Tools
Banned | Posts: 3,412 | Thanked: 1,043 times | Joined on Feb 2010
#11
Originally Posted by MohammadAG View Post
Based on what I know.
The chip in the N900 should run @ 550, with 600 being a default "overclock".
Running 550 gives the device a lifetime of 10 years.
600 5 years, that's half the lifetime for a 50MHz increase (rough estimate).
afaik these values are from TI (texas instruments) themselves.
My rough estimes:
650MHz, 2.5 years, 700MHz 1.25... you can guess the rest I suppose.
Also, according to joerg_rw, an engineer (not at Nokia mind you), damage to the CPU occurs due to http://en.wikipedia.org/wiki/Electromigration, not thermal factors.

What annoys me though, is that apps like CPUFreqUI to report "raw" speeds, using a sensor that according to SpeedEvil, TI themselves suggest not using, and it's not on the CPU, it's next to the battery.

Anyways, this is the info i've collected so far, what I said above is not from my personal findings (except my rough estimates).
Oh dear where was your brain when you typed this lol. HEAT is the deciding factor here and the figures your quoting are simply not true in any way or form.

I suggest you read that link properly and you might understand a bit more... good luck.
 
MohammadAG's Avatar
Posts: 2,473 | Thanked: 12,265 times | Joined on Oct 2009 @ Jerusalem, PS/IL
#12
Originally Posted by giannoug View Post
Find me a device that works at 600Mhz everyday I use my N900 one or two hours per day. That means the clock gets higher than 250Mhz ONLY those hours. It runs mostly at 250Mhz with bursts (no more than 10-20mins) at 1150Mhz. I suppose my grandchildren will love it when I give it to them as a gift

Just for the record, I have my N900 overlocked to 1.1Ghz since Lehto's kernels... Now I'm using Titan's. I've never experienced any problems, except some crashes which might not be related to OCing...
Well, yeah, the values are based on the device running that frequency 24/7 (kind of like LCD display lifetime).

About the second part, that doesn't mean it's safe now does it

Edit:
Originally Posted by abill_uk View Post
Oh dear where was your brain when you typed this lol. HEAT is the deciding factor here and the figures your quoting are simply not true in any way or form.

I suggest you read that link properly and you might understand a bit more... good luck.
Heat is a factor, but it's not the only factor as some apps suggest (see CPUFreqUI with its broken "raw temperature" sensor)

Last edited by MohammadAG; 2010-07-04 at 11:00.
 
Posts: 208 | Thanked: 91 times | Joined on Jun 2010
#13
For anyone now confused about the discussion.

You will destroy your n900s from over-clocking long after it has become obsolete and replaced.
 
Posts: 1,224 | Thanked: 1,763 times | Joined on Jul 2007
#14
Originally Posted by MohammadAG View Post
@ Matan, so values from TI are meaningless?
You do not report TI's numbers, you invent your own numbers, as you openly admit.

TI numbers are:

100K hours when operating across all OPPs and no more of 23K of those hours are at OPP5-6.

50K hours when using OPP5-6 only.

44K with absolutely no restriction on OPPs.

Where do you get those 10 years and 550MHz and 5 years at 600MHz, I can't imagine.

It is important to recall that an OPP is a predefined frequency+voltage level, so when we do not use TI's OPP (such as when we use 600MHz at a lower voltage level, or when we use higher frequency and lower voltage compared to OPP6) the effect on lifetime is unknown.

Since practically any N900 runs at more than 77% idle, we are within the parameters of the 100K hours lifetime, and the fact that a few percent of this time is spent at 900 MHz instead of 600MHz, most probably does not have a great effect on lifetime.
__________________
My repository

"N900 community support for the MeeGo-Harmattan" Is the new "Mer is Fremantle for N810".

No more Nokia devices for me.
 

The Following 2 Users Say Thank You to Matan For This Useful Post:
Posts: 842 | Thanked: 1,197 times | Joined on May 2010
#15
That 44k figure you mention(i.e. 5 years), does that mean running at 600mhz/stock voltage 100% of the time?

Also, has anyone tried to run numbers on OCing/undervolting based on -these- figures? Or, better yet, call up someone at TI and ask -them- about some of this stuff? "I'm looking for technical information on the OMAP3430 and its theoretical lifetime..."
 
Posts: 1,729 | Thanked: 388 times | Joined on Jan 2010 @ Canada
#16
so in what i've read from the first page, i am having the conclusion that the estimated time of hardware to wear is still in theory, in overclocking.

but for sure, will still reduce device's life?
__________________
if i did help you, just click "Thanks" on the lower right of my post. thanks!

"The best way to break a Spell is to prevent it from being cast in the first place"

N900: 1000/1150mhz; sampling_rate 15; up_threshold 150000;
 
Posts: 1,427 | Thanked: 2,077 times | Joined on Aug 2009 @ Sydney
#17
Originally Posted by MohammadAG View Post
Based on what I know.
The chip in the N900 should run @ 550, with 600 being a default "overclock".
Running 550 gives the device a lifetime of 10 years.
600 5 years, that's half the lifetime for a 50MHz increase (rough estimate).
afaik these values are from TI (texas instruments) themselves.
My rough estimes:
650MHz, 2.5 years, 700MHz 1.25... you can guess the rest I suppose.
Also, according to joerg_rw, an engineer (not at Nokia mind you), damage to the CPU occurs due to http://en.wikipedia.org/wiki/Electromigration, not thermal factors.
Okay. Going by your estimates.

750 = 6months
800 = 3 months
850 = 1.5months
900 = 3 weeks
1000 = 5 days
1100 = 1.25 day

I've been running mine at 1Ghz+ for 3 months now.
So I guess mine should have died 3 months ago.

It might be true if you run it at that frequency at a constant 100% load. But that's still theory.
Also depends heavily on the voltage and the temperature. Not just frequency.
If you had it like that in blazing hot summer, maybe it'll die way quicker if ran at 100% permanently.

However in real life, it may not even affect the N900's lifetime whatsoever even when overclocked to 1GHz.
In most uses, CPU will be running at 100% for less than couple of minute bursts at a time. (more like seconds)
Maybe in bad cases, up to 15minute at 100% if you run some cpu intensive app/game constantly.
But still, it's not enough to really affect the CPU's lifetime in a noticebaly negative way.

ie. CPU lifetime isn't reduced every time the CPU reaches 600Mhz+. So if the CPU runs at 1Ghz for 1second, it might mean zero negative surplus impact to the cpu lifetime. There is always a threshold it needs to go over before it starts deteriorating exponentially. Also, those TI specs are based on he initial design. Over time, they will get better yields from manufacturing and may actually perform way better than what they initially specced out.

Last edited by jakiman; 2010-07-04 at 12:10.
 

The Following 5 Users Say Thank You to jakiman For This Useful Post:
Posts: 1,224 | Thanked: 1,763 times | Joined on Jul 2007
#18
Originally Posted by RobbieThe1st View Post
That 44k figure you mention(i.e. 5 years), does that mean running at 600mhz/stock voltage 100% of the time?
If you always run at 600MHz stock voltage (OPP5), then you are within the parameters of 50K hours lifetime.
__________________
My repository

"N900 community support for the MeeGo-Harmattan" Is the new "Mer is Fremantle for N810".

No more Nokia devices for me.
 

The Following User Says Thank You to Matan For This Useful Post:
Posts: 1,427 | Thanked: 2,077 times | Joined on Aug 2009 @ Sydney
#19
Originally Posted by MohammadAG View Post
Heat is a factor, but it's not the only factor as some apps suggest (see CPUFreqUI with its broken "raw temperature" sensor)
I'm pretty sure that the only temperature sensor we can tap into is the battery temp sensor. (this was from Igor - Nokia N900 HW engineer) Battery temp obviously goes up when more current is pulled off it. But we can never know what the CPU temp is still.
 
Banned | Posts: 3,412 | Thanked: 1,043 times | Joined on Feb 2010
#20
Originally Posted by MohammadAG View Post
Well, yeah, the values are based on the device running that frequency 24/7 (kind of like LCD display lifetime).

About the second part, that doesn't mean it's safe now does it

Edit:


Heat is a factor, but it's not the only factor as some apps suggest (see CPUFreqUI with its broken "raw temperature" sensor)
You might be suprised to know that each and every cpu will have a different lifespan due to the actual build quality, as for the heat sensor i did look on the schematics and the data sheet for the arm cpu used in the N900 and came to the conclusion it is just too much effort to install an effective temp sensor so i went for the next best thing and a heat sink will keep down the temprature but to be effective there must be better ventilation on the N900 for heat dissapation to work effectively, so a lot of headaches lol.

I would not take any notice of apps made for the N900 as the main goal is to keep heat to a minimum but if you incorporate a fan then obviously a heat sensing circuit would be beneficial but not imperative.
Only problem with a fan is the battery life would drastically decrease unless on an external power supply.

After quite some exploration on different N900's my estimation would be something like 18 month at full overclock for the pcb to get brittle enough to start giving problems with no heat sink, having said that with a heat sink applied would make that a lot longer lifespan on trouble free operation.

No way will the arm cpu in the N900 break down under normal conditions but i cannot say the same for the pcb it is soldered too.

All in all i will give the device a max of probably 5 year or more under normal speed before pcb deteriation starts to be a problem.
 
Reply


 
Forum Jump


All times are GMT. The time now is 04:52.