Ever since the release of nueOverClockTest, I’ve got several inquiries on “why can’t there be a 600MHz setting? It would probably be much more stable”. Well, the short answer is – I haven’t found how to get to that setting yet. Since Qualcomm would rather keep information about it’s coveted MSM line a secret, I basically have to guess how to get things to work.
I’ve figured out my problem with changing the “L-Value” of BackPll0/1 – I need to change it when it isn’t selected as the ARM11 SubSystem’s core clock. However, it’s still pretty unstable, and a value of 0x5 makes the device slow as heck. What’s even more perplexing is the default value, which seems to be 0x7. If that is multiplied against the 19.2 MHz TCXO clock, we get something like 134 MHz. Perhaps I’m looking at the wrong place, or I got the meaning of the registers wrong.
I’ve found the registers that control the voltage exactly, not just levels 0-7, but writing to them seems to have no effect. It seems that I don’t understand the PMIC well enough yet to get complete control over it.
I’m still evaluating our options to figure out what’s best, but its starting to lean more and more towards figuring out how to get the ARM9 to execute code for us without hacking the radio rom.
Anyways, this raises a question that’s been on the back of my head for awhile now – why the heck does Qualcomm think it’s necessary to keep all information on the MSM chipsets confidential? I doubt its because they’re afraid the competition is going to steal all their secrets, because I doubt the documentation HTC engineers use could help a competitor of Qualcomm in any real way… Making the documentation public would probably get more developers on the device, increasing their sales volume… Alas, I fail to understand Qualcomm… Perhaps I should send them an e-mail asking for information.
Regarding 0.4, there’s nothing yet to show in a version 0.4, so relax, calm down, when I have something to show you, it’ll come out in version 0.4.