Having a boost_counter >1 as default?

[Idokras]

Hello everybody!

First off thanks to all the people that have made Rockbox possible - it's tremendous to be able to a) just use a file-browser for MP3s, b) not having to use iTunes and c) having so many options all around in the software!

I've been using Rockbox for my 4G U2 ipod and for my gf's 1G mini ipod for a few weeks now, but unless we change the CPU frequency to 75Mhz and up the boost_counter to a number bigger than 10 or so our ipods crash pretty quickly. I'm aware that a boost_counter of 1 and any other number should be equal as it's a true/false category, but I've come to realize that numbers bigger than 1 cause fewer crashes.

As of now, we have to adjust the boost_counter everytime we start up the ipod - this is pretty tedious. Rockbox itself only saves the 75Mhz, but not the boost_counter level. Is there any way to compile a version that has a high boost_counter on default? Or is it possible for future Rockbox versions to be able to save the boost_counter level chosen and - even though that level changes during the usage - that it starts off with that level everytime the player is being used anew?

Apart from that issue, this piece of software is awesome, and it would make it perfect for me if it could be dealt with.

Thanks for any advice,
Ced

[Llorean]

It *is* true false. Having a higher boost counter will *ONLY* matter if the boost counter is being decreased more often than it's increased.

Also, boost will not be increased by default, as that defeats the entire purpose of having a variable CPU speed. Rockbox for iPod is not released yet, and this is a feature that needs to be working before release. Removing the feature is not an option, especially since that removes any possibility of a non-developer figuring out the cause.


Could you do me a favour, set the boost to 10, use your iPod for about an hour (or longer), then stop music playback, and check what the boost level is? It should still be 10. It would be helpful to know if it's not.

[Idokras]

Thanks for the quick reply, Llorean.

I checked a few times after listening for an hour or more and then stopping playback, and the boost counter changed mostly from 10 to 9, sometimes it stayed at 10. When I checked in during playback, it was sometimes 9, sometimes 10. Maybe it went even lower or higher when I didn't check in between, I don't know.

After using the ipod for more than an hour, I shut rockbox down through holding play, and then restarted it. The boost_counter was at 1.

The same happened with the 1G mini.

It would be optimal if the software could save the boost_counter at two or three for me during a re-start. It seems as if the saved value of "1" goes to zero and then during that time it crashes somehow.

Can you give me any advice as far as convenience is concerned? Will I just have to set the boost_counter to a number higher than one everytime I start rockbox? Or can the value be saved for the next start?

Thanks,
Ced

[Llorean]

There is no purpose in saving the value. The software *should* handle the boosting automatically in the background.

Which is why I wanted you to set it at 10 while music wasn't playing (starting from 0) and see if it ever went below 10. If it's 0 when you set it to 10, then you start music, and it goes below 10, something is wrong.

So if you could reconfirm that was what was happening (as it should go 10-11 back and forth if you started from 0, not 9-10) it would be useful.

[mikeage]

There is no purpose in saving the value. The software *should* handle the boosting automatically in the background.

Which is why I wanted you to set it at 10 while music wasn't playing (starting from 0) and see if it ever went below 10. If it's 0 when you set it to 10, then you start music, and it goes below 10, something is wrong.

So if you could reconfirm that was what was happening (as it should go 10-11 back and forth if you started from 0, not 9-10) it would be useful.

I'm not sure if this is still relevant, but I've been testing this a lot lately, and my boost counter does _not_ have a net drop. In other words, if I start at 10, it alwys fluctuates between 10 and 11.

That said, my 4G crashes quite frequently (every 10-45 minutes, usually), so I'm running a build without frequency scaling.

[Llorean]

Well, it's good to know there's not a net drop.

Unfortunately, it's still unknown exactly what causes the crash in the frequency scaling code.