Rockbox

Tasklist

FS#11597 - AMSv1: improve battery life

Attached to Project: Rockbox
Opened by Rafaël Carré (funman) - Thursday, 02 September 2010, 12:51 GMT
Last edited by Rafaël Carré (funman) - Sunday, 05 September 2010, 15:34 GMT
Task Type Patches
Category Battery/Charging
Status Closed
Assigned To No-one
Operating System All players
Severity Low
Priority Normal
Reported Version Release 3.6
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Gives ~ +20% battery life

Changes cpu freq by CGU_PROC instead of using fastbus/synchronous
Also removes setting for asynchronous

Tested on clipv1/fuzev1

Problem: long home press seems to be broken on fuzev1?
This task depends upon

Closed by  Rafaël Carré (funman)
Sunday, 05 September 2010, 15:34 GMT
Reason for closing:  Accepted
Additional comments about closing:  committed v4 patch with some comments: r28000
Comment by Rafaël Carré (funman) - Thursday, 02 September 2010, 13:06 GMT
This is more correct, in synchronous mode cpu clock must be > pclk/hclk (and an integer multiple of it)
Comment by Rafaël Carré (funman) - Thursday, 02 September 2010, 14:11 GMT
My home button has a problem with SVN too so it's not related
Comment by Rafaël Carré (funman) - Friday, 03 September 2010, 06:31 GMT
Clipv1: 10h runtime
Comment by Michael Chicoine (mc2739) - Friday, 03 September 2010, 11:25 GMT
e200v2: 19:22:49 runtime
Comment by Jack Halpin (FlynDice) - Friday, 03 September 2010, 12:13 GMT
fuzev1 SVN r27965 12:23, with patch 16:35
Comment by Rafaël Carré (funman) - Friday, 03 September 2010, 15:07 GMT
I got 22hours on fuzev1 but something definitely went wrong

2 consecutive lines are:

01:08:12, 04092, 092%, 13:40, 4088, -, -, -
14:15:46, 51346, 091%, 13:36, 4075, -, -, -

It seems somehow the player was sleeping in some state all night until I looked at the battery level in the morning (I didn't press resume, and the config has nothing weird)

Comment by Rafaël Carré (funman) - Friday, 03 September 2010, 15:57 GMT
- Base PCLK off PLLA and not FCLK, so PCLK is not divided by 4 when unboosted

Not sure if the battery gains will still be there
Comment by Rafaël Carré (funman) - Friday, 03 September 2010, 18:33 GMT
Another idea to try:

- reduce FCLK to maximum (use maximum dividers and the 24MHz crystal as source) when unboosted

Also it might be interesting to try a build with PCLK set to 15.5MHz to see if it compare equally to previous results
Comment by Bertrik Sikken (bertrik) - Friday, 03 September 2010, 21:33 GMT
I got 13h30 with the v2 patch on clipv1. An older benchmark for this player gave me 10h59m.
Comment by Rafaël Carré (funman) - Saturday, 04 September 2010, 12:41 GMT
With v4 patch:
- fuzev1 15h30
- clipv1 8h50
Comment by Michael Chicoine (mc2739) - Saturday, 04 September 2010, 19:45 GMT
e200v2 with v4 patch: 16:45:56 runtime

Loading...