RockBoy Speed

[Danycacks]

Is there a way to speed up the games on the iPod. Because there slow.

[zajacattack]

To my knowledge, there is not; however, I might be mistaken.

[safetydan]

If there was it'd already be done. There's room for optimisation, but no one has stepped up to do it.

[grossaffe]

i'm sure if someone went to the length of overclocking their player it could be done... but that's a complete waste.

[Sanek]

i'm sure if someone went to the length of overclocking their player it could be done... but that's a complete waste.

I'm actually curious - how feasible would overclocking of a 5.5g iPod by say 5-10 MHz be? I know its an embedded device and normally should not be overclocked because of possible overheating, the battery life would go down, etc.

In theory, how much can an iPod be overclocked without risking overheating?

The reasoning for this would be to get just a tiny bit more power out of the CPU when needed (only when needed mind you, to save some power), so things like RockBoy would run better.

[Llorean]

Why not optimize Rockboy. I really don't like the idea of "Abuse the hardware, rather than doing the actual work needed to get things running well."

Those of you who want faster Rockboy are more than welcome to contribute patches optimizing things, but it's unlikely Rockbox will ever intentionally abuse hardware just so people don't have to work on things.

[zajacattack]

Well, why not overclock? Wouldn't that give an overall performance boost?

[safetydan]

The specifications that we have for these chips say that their maximum clock is 80 MHz. Do you really want to go outside that? Apart from heat and battery life issues there's also the danger of breaking the CPU in some way.

[zajacattack]

Oh, so the risks end up outweighing the benefits?

[Llorean]

In case the key word "abuse" didn't hand it to you on a silver platter, overclocking is a good way to cause permanent damage to the hardware.

Overclocking is not a solution, it's a way to avoid solving the problem properly. Any performance gains would be offset by an increased chance of permanent hardware failure.

Why not optimize code? Wouldn't that give performance boosts?

[grossaffe]

i was simply stating it could be done, but definitely should NOT be done.  first off, i don't know of any way to do it through software (don't know if the rockbox developers could do it), which means it would have to be done with an oscillator crystal soldered to the main board, most likely not fitting in the designed case.  and on top of all of that, like previously stated, it would be a terrible strain on the chip as well of the lack of adequate cooling which would no doubt cause irreparable harm do the device.  but if you REALLY wanted that little performance boost...

[Llorean]

It can be done in software fairly easily actually.

[Sanek]

It can be done in software fairly easily actually.

Just a little follow-up question:
On an iPod, when CPU frequency increases, does the voltage change with it (proportionally) or is the voltage always constant?

My guess would be that it changes all the time depending on the CPU utilization as to save some energy, but is that really the case?

RockBoy was just an example since thats what the thread was about. The reasoning I had behind the overclocking question is that it would allow running more demanding applications on the same targets, but I guess it could possibly be too dangerous and ultimately not worth the risk to go beyond the specifications of the CPU - which brings me to my next question...

The specifications that we have for these chips say that their maximum clock is 80 MHz. Do you really want to go outside that? Apart from heat and battery life issues there's also the danger of breaking the CPU in some way.

According to Wikipedia, 5th Generation iPods use PortalPlayer 5022 CPU
http://en.wikipedia.org/wiki/PortalPlayer

And according to the PortalPlayer 5022 CPU Specifications in the Rockbox wiki:

Up to 100 MHz processor operation per core with independent clock-skipping feature on COP

http://www.rockbox.org/twiki/bin/viewfile/Main/PortalPlayer?rev=1;filename=5022_Brief_Mar05.pdf

So maybe there really is room for overclocking there?

[safetydan]

That's one particular model. Other's in that series support lower clock rates.

As the the clock rate is software controllable, feel free to experiment. But it's something that you're doing at your own risk. A much better solution would be to optimise rockboy.

[Llorean]

It's quite likely that, for most things Rockbox does (with the notable exception of Monkey's Audio music, which isn't even fast on the Gigabeat) we should be able to run them successfully at 80mhz with enough optimization.

Since we're going to *have* to do it anyway for most of the iPods, we should focus any effort on solving the problems.

As well, that wikipedia article is slightly wrong. The iPod Video uses the PP5021, which reports itself in software as a PP5022C, and we theorize that those are PP5022 chips that don't run stably at full speed.