Binary patched kernels for Pipo models

[giouncino]

I thought a lot on the tables read by overclockomatic...

These are the stock tables for Pipo M6pro:

== Table 1 at offset 10199692 ==
MHz mV
816 1000


== Table 2 at offset 10203168 ==
MHz mV
816 1050


== Table 3 at offset 10203216 ==
MHz mV
200 1000
300 1050
400 1100
500 1150
600 1200


== Table 4 at offset 10203304 ==
MHz mV
180 1150
300 1100
400 1150
500 1200


== Table 5 at offset 10203400 ==
MHz mV
500 1100


== Table 6 at offset 10212588 ==
MHz mV
312 900
504 925
816 1000
1008 1075
1200 1200
1416 1250
1608 1300


== Table 7 at offset 10212652 ==
MHz mV
133 975
200 1025
266 1025
300 1050
400 1100
600 1250


== Table 8 at offset 10212716 ==
MHz mV
580 1300 1 DDR_FREQ_NORMAL


== End of tables ==

Table 6 is FOR SURE the CPU voltage/freq table. Same for table 8 that is the RAM table.
We supposed that Table 7 seems to be th GPU table... but if we change voltages nothing seems to be changed... also the same thing seems for the frequency...if I set 600Mhz@800mV all seems warking flawlessly, no artifact...

Now my questions are:
What are table 1-5 related to?
In particular, table 3 seems to have a wide frequency/voltage ranges...and if is the GPU table?

Table 4 with those voltage/frequencies seems in some way related to a particular strings of the Rk3188 android system file android_kernel_rockchip_rk3188 / arch / arm / mach-rk3188 / cpufreq.c:

...

static int test_tlb_gpu[TEST_GPU_NUM] = { 360, 400, 180 };
static int test_tlb_ddr[TEST_GPU_NUM] = { 401, 200, 500 };


...
Table 1 - 2 and 5 are completely obscure for me...

But just for curiosity, is there anyone who knows what these tables are related to? And what could happen if I modify the overclockomatic script to modify these parameters?

I think I could get a try...

[TerryH]

We supposed that Table 7 seems to be th GPU table... but if we change voltages nothing seems to be changed... also the same thing seems for the frequency...if I set 600Mhz@800mV all seems warking flawlessly, no artifact...

Changes to the voltage may not be altered by using Table 7 as per your testing, however changes to the Frequencies are definitely reflected when dmesg is run:
Table values:
100, 150, 200, 300, 400, 600
shows as:
99, 149, 198, 297, 384, 594.

Furthermore, I suspect that CPU voltage settings are also influenced by other things than Table 8 alone. The overclockomatic script has only one setting for CPU, generally DDR_FREQ_NORMAL. When dmesg is run, a setting for suspend is also shown which is not contained in the displayed Table 7.

[giouncino]

Changes to the voltage may not be altered by using Table 7 as per your testing, however changes to the Frequencies are definitely reflected when dmesg is run:
Table values:
100, 150, 200, 300, 400, 600
shows as:
99, 149, 198, 297, 384, 594.

Furthermore, I suspect that CPU voltage settings are also influenced by other things than Table 8 alone. The overclockomatic script has only one setting for CPU, generally DDR_FREQ_NORMAL. When dmesg is run, a setting for suspend is also shown which is not contained in the displayed Table 7.

the setting for CPU suspend as far as I know is driven by the voltage set on table 6 (1000mV) The Freq set at 1000mV (or the closest) is the one set for suspend, while the RAM settings for suspend is driven elsewhere because dmesg says also 198Mhz and 300Mhz about RAM not included in table 8.

[heisenberg]

My opinion: table 3 is the gpu.
Cause nothing changes, when we do anything to table 7. Table 3 has the nearest frequencies to what we are looking for.. .

[heisenberg]

My opinion: table 3 is the gpu.
Cause nothing changes, when we do anything to table 7. Table 3 has the nearest frequencies to what we are looking for.. .


This is the answer to my question about deep sleep from pipo:

And our technician was told to make a non-stop-3G-network frimware, that is why it can't go to deep sleep.You can turn off the 3G modern if you don't use it.


Best regards,


Anna
PiPo Technology Co,.Ltd(

[herrjemineh]

in the pipo m7 2.2 thread is a suggestion:

Originally Posted by val_andrio

Don't get me wrong and all this performance boost is really good to have. But how about the battery life?
I noticed (since my M7Pro) doesn't deep sleep, I loose 1% almost every hour without doing anything.

How about after this overclock/performance boosts? My old Galaxy S+ with cyanogen mod drops 2%-3% max after a whole night (8 hours) with 1/5 battery capacity comparing with my pipo.

How does cyanogen mode achieve this? Is it a matter of better kernel?

Answer:


ril-rk29-dataonly is coming back....(only for 3G version)
Move ril-rk29-dataonly from / system / lib / to your SD card. Reboot. You don't have 3G service but you win the "Deep sleep".
If you need 3G, move ril-rk29-dataonly from your SD card to / system / lib. Reboot and you will have 3G service.

Last edited by manao; 12-04-2013 at 10:09 PM.

[fazerg]

Replace original ril-rk29-dataonly with one from attached archive and you'll have no problem with deep sleep without reboot or moving files.
I've not tried it yet but all who did confirmed working, just takes a bit longer to start 3G modem.

libril-rk29-dataonly.zip

[herrjemineh]

I'll give it a try. Tell you about the result tomorrow

Am Ende ist alles gut, und wenn nicht alles gut ist, dann ist es auch nicht das Ende...

[giouncino]

FANTASTIC!!!!!

Replace original ril-rk29-dataonly with one from attached archive and you'll have no problem with deep sleep without reboot or moving files.
I've not tried it yet but all who did confirmed working, just takes a bit longer to start 3G modem.

[ATTACH]4309[/ATTACH]

IT WORKS FLAWLESSLY!!!!!!!

Thank you so much!!!!!!

The only very very little problem is that as soon as the screen goes off the 3G automatically is swithched off. But it is so much important to have deep sleep working that this is absolutely not important!!!!

[fazerg]

as soon as the screen goes off the 3G automatically is swithched off.

This is the goal, isn't it? Not exactly of course but close. Just need to increase timeout for screen perhaps.
All thanks not to me. This version of library was found in M9 thread on russian 4PDA forum.

[giouncino]

This is the goal, isn't it? Not exactly of course but close. Just need to increase timeout for screen perhaps.
All thanks not to me. This version of library was found in M9 thread on russian 4PDA forum.

DEFINITELY YES, it very close to the final goal!
I suspected from the beginning that the only responsible was that stupid library file, I looked for that file almost everywhere on internet and tried many modification on that file without success (copying the same file from other tablet models), so FOR ME thank you to you that give me finally the correct file!

[herrjemineh]

Not Working for me unfortunately. Still 1% per hour drain, battery Status still Shows that cell Standby takes 70% of the whole battery consumption.
Andy suggestions?

Gesendet von meinem M7pro mit Tapatalk

[herrjemineh]

Obviously i made a weird mistake somewhere. After redoing, e.g moving the existing lib to sd card and replacing with the one attached seems to work now. I Know it in a few hours

Am Ende ist alles gut, und wenn nicht alles gut ist, dann ist es auch nicht das Ende...

[giouncino]

Obviously i made a weird mistake somewhere. After redoing, e.g moving the existing lib to sd card and replacing with the one attached seems to work now. I Know it in a few hours

Am Ende ist alles gut, und wenn nicht alles gut ist, dann ist es auch nicht das Ende...

It is not necessary, if you install CPU spy, as soon as the screen goes off, the tablet goes in deep sleep and next time you switch the screen on, on CPU spy you will see the bar relevant to deep sleep (if the tablet never goes in deep sleep that bar does not appear).

[krucifix85]

Would love a 1-2 minute time out or something. That would be so good! Nice find btw!!