I can absolutely prove that driving a car is not safe. The statistics are completely obvious. Anybody who thinks that driving a car is safe is insane. The easiest way to save huge numbers of lives every single day would be to close down the highway system. Who can disagree with me? Yet, I drive a car.

And I overclock. Why? Because the benefits outweigh the risks. Those who want to magnify the supposed damage of overclocking carefully steer away from mentioning the overwhelming positive experience on this site of people who overclock. I think it is debatable whether even the few incidents of supposed damage from overclocking are accurate — look at the messages and you will see that the people claiming the most damage are running everything but the kitchen sink and it's very difficult for me to tell that overclocking must be the smoking gun.

If anyone can give me instructions on how to damage my N900 with overclocking, I will strongly consider trying it out. But, in scientific discussions when you try an experiment the results of that experiment should change the beliefs of the people proposing the experiment – otherwise, the experiment is meaningless. Unfortunately, I think we are seeing a faith-based argument against overclocking. It is impervious to reason. Any one who cites their own experience is talked about with contempt, like experience has nothing to do with it.

I'm not an expert on the subject, but I've done some testing that could be related to CPU wear a year ago or so.

I ran nbench a multitude of times with several clock/voltage combinations. While running the tests, I monitored current_now as reported by the bq27x00_battery module. Unfortunately I've lost the spreadsheet with the exact results, but I do still remember my observations:

A higher clockspeed with substantially lower voltages required a lot more current from the battery than a lower clockspeed with substantially higher voltages.
Setting a lower or higher voltage for the same frequency didn't nearly affect the used current as much as setting a lower or higher frequency.
This is with SmartReflex turned off in software.

These results seems to support joerg_rw's post that the SoC does have micro regulators that affect the current. This in turn seems to suggests that the user set voltage isn't nearly as much in play as in traditional desktop/laptop overclocking (and thus overclocking the N900 can't be compared to that).

I would be grateful if someone* could reproduce (or disprove) the results of my test. Nbench for the N900 can be found here. Personally, I think the higher current_now is enough 'practical' prove that OC'ing does wear out the CPU much faster than running the cpu at stock speeds, pretty much regardless of whether you're undervolting or not.

* do we have a volunteer?
--

By the way, I don't know how accurate current_now from bq27x00_battery is, but the fact that it went up significantly with higher frequencies at least means something I guess. Testing was done with SSH over USB (offline mode, display off).

Exactly. I am quite confident that no matter how much we overclock the N900 within the boundaries of what the Power Kernel allows, the CPU will outlast pretty much anyone's life time usage of the device. Yes, you can try to be scientific and calculative about the whole theory behind overclocking and how it can damage or shorten its lifetime. But my point about overclocking being safe is far more correct "till-now" than the ones who are over-cautious and saying overclocking is dangerous. If suddenly people's N900 start burning/dying tomorrow, I'll stand corrected. But I doubt it will be the CPU that burns out first.

N900 has been out for about 23 months now and overclocking started around 18 months ago I think. So we only have 18 months of data to go by. I am fair confident in saying we so far had less than 0.1% (if that) of those who overclocked that had their N900 physically damaged by it. I'm throwing imaginary numbers out of my head so please correct me if you have more accurate numbers.

I've been overclocking CPUs since the 486 days. (My motherboad didn't have any option to overclock my 286 and 386...) I've yet to have a cpu die on me and I gone through over 20 CPUs myself and many many more for all my family, relatives and friends that I've built the PC for. (I overclocked every single one of them before letting them use it. lol)

Mobile phone overclocking is fairly new concept and we don't have the flexibility to better "cool" it or run it at different FSB etc. I've only overclocked a handful of Android phones (including the SGS2 where I got mine to 1.5GHz rock stable) and the N900 so far. So not a whole heap of personal experience yet but I might still be up there and be above-average.

I understand some people going totally against it. But I don't understand when you just go against it without any real strong evidence or statistics to back it up. I also don't have a proper statistic for you guys but the history so far is definitely in the favour of the overclockers I think..... We will know more in a few more years.... as that's the argument... I guess only time will tell...

(This is a debate where no one can win. It's like religion it seems. lol)

@jakiman

They go against overclocking and saying how their lifetime is going down the sewer, but wasn't the SGS II overclocked like half a month prior to launch? There was no time to implement a new SoC in half a month. Going by what joerg_rw says, we should sue Samsung for lowering our CPU's lifetime. Did anyone complain? No, the phone got praised for being the fastest in the mobile industry.

Based on this, the question rises, how much are this phones actually downclocked in the cheapest SoC variant? Desire Z can go from 800MHz to 1.8GHz even 2GHz, a crazy jump. Apple, iirc, officially stated that 3GS A8 was downclocked to 600MHz from the actual 833MHz (3440 also 800MHz). How many Moto Droids have actually died at 1.3GHz. People have been benchmarking the said frequency for hours and playing emulators (same with the N900)?

BTW, weren't you the one that ran min max 1GHz on your N900 for a while?

@iDont - it is obvious that runnning CPU constantly on higher frequency will increase average current (which is what bq27x00_battery measures). But average current is not the same as peak current, which is the one responsible for EM. And running syntetical benchmarks has nothing to do with real life usage patterns.

I really hope you can find your tables (or someone to do measurements again) to see how voltage affects current, because quantities as "lower" and "higher" may work in fuzzy logic, but won't work here.

Re Smartreflex - I am in process of educating myself about that technology and kernel support for it. So far it seems that it CAN be turned off, just not sure if setting sr_vdd1 and sr_vdd2 to zero is enough.

That's right. Even the iPhone4 was downclocked to 800Mhz from the 1GHz A4 which iPad2 uses. It's a matter of striking good balance between heat, battery life and performance. These mobile phone SoC's are made with much higher yield than what they actually are made to run at to ensure they are stable even while they are under a blazing summer heat with no air circulation. So it's even better than the desktop PC CPU's IMO.

Yes, I was the one running my N900 at 1.1GHz for both min/max. Heck, even Lehto (the first one ever to overclock the N900 at TMO afaik) told me he does this. But this isn't something I would recommend for everyone unless you know the risks. (possibility of instability, hand burning, exploding...just kidding...hehe..) But yeah, it should only be done if you are experimental and fearless.

BTW, I got my N900 replaced this week as my USB socket ripped out. (my kid stomped on the cable/n900 while it was still plugged in) This new one I have is even more stable than my previous one as my last one was quite unstable at 1.15GHz using ideal voltage. But this new one seems to be rock stable. My last one never skipped a beat at 1.1Ghz using xlv voltage. So this one should last me even longer maybe as long as my kid doesn't kick my N900.

I'm actually eagerly waiting for someone to overclock the N9.
I reckon 1.2GHz is just guaranteed and 1.4GHz is highly possible.
So yeah, bring it on. I'll risk my N9 for the community.

(AFAIK, I was the first person Lehto gave his custom kernel to to overclock the N900 other than his own)

@jakiman - just have in mind that 3630 SoC uses 45nm, not 65nm as 3430, i.e. it is more fragile and susceptible to EM

hehe. Yes. But it also means it's is more efficient. Smaller fabrication so far has pretty much always shown that it allows faster frequency for the same amount of power. (or even less power) If the 3630 SoC is anything like Samsung's SoC used in SGS1 (similar Cortex A8), it should be able to handle at worst 1.2Ghz. But I'm an optimistic so I want my N9 running at at least 1.3GHz. Should make rendering web pages, loading apps, multi-tasking a bit smoother and faster.

Heat and power consumption are stopping them from upping the frequency at a larger fabrication process (remember, it's a phone). Anyhow, you'll surely get 1.3GHz out of the 3630 since the Droid 2/Milestone 2 are capable. Also, 3630 can come out of the factory as 1.2GHz; clock range is 600-1200MHz.

http://en.wikipedia.org/wiki/Texas_I...ts_OMAP#OMAP_3

I sure as hell wanna know what the actual clock of the A9 is inside the Vita.

Very interesting thread!

To understand why anecdotal evidence is really useless here, one has to remember that both the quality of a single chip (expected time to failure) and the time to failure itself are random. This is why TI uses margins in the clock speed, and this is why some chips will stand for a very long time OCed, while other will fail quickly. (randomness of quality is obvious when you look at maximum stable clock: we have a lot of feedback on them, and they vary from 700 to 1200 MHz for different devices)

exemple 1 (N900 related, invented numbers):
let's say that the average chip reaches an expected time to fail of 4 years at 800MHz, but the variance is such that 1% of the devices achieve these 4 years at only 600MHz, while another 1% will do it at 1000MHz.
Most of the feedback on OC at 800MHz would be largely positive (especially after 1 or 2 years), and yet such OCing would be unthinkable for the manufacturer, and the lifespan of a significant number of devices would be reduced.
I have fabricated the numbers, but you can't discard this possibility with today's knowledge.

example 2 (not N900 related, real numbers):
To show that quality of processors in a manufacturing process vary VASTLY between individual chips, within the same batch, factory, etc.. Let's take a look at the recent apparition in the market of "unlockable cores": 2, 3 and 4 core chips are now identical, produced by the same chain, and all ship with 4 physical core. BUT, there are so many cores that fail the post production quality tests that you they can base almost all of their 2 and 3 core production on the failures/remains of the 4 cores chips.
Failling the quality test does not mean that the chip won't work. It doesn't even mean that it won't work as long as one that has passed. It means that its failure probability is too high (probably like 1 or 2% for the warranty period)
Now, think of this in the context of "same chip, different clock speed, different price".

well, my point is, whatever people told you or experience:

• OC is safe for some devices
• OC is not safe for some devices
• You can't tell in which category yours is.

PS: lights bulbs mostly wear when switching them on, because of the mechanical constraint due to the violent, uneven temperature changes. The wear by evaporation of the heated metal is much less significant. So if you never switch a light bulb on and off, it is much more likely to last 100 years