Found out the hard way that NativeBOINC won't run on 5.0 and up. The manager loads, but the core client wont... fails to connect. Guess the developer would have to recompile for PIE compliance. Summa Summarum... back to BfA. Definitely OK with 5.02, left alone 95% or better valids on OET1. Sometimes they bomb within the first minutes, middle of the night at that... this period where a result is uploaded and a new one is started. Something as on Linux not handled well.

No luck here. I tried and same as you had the client fail to connect.

OET through stock Boinc on my original Nexus 7 runs well. I've not kept stats as you, but I can't remember when I last received an error. This is the same device which I quit running FAAH tasks out of frustration. Nearly all FAAH tasks errored. I haven't tried FAAH post 5.0, but I'm happy with my valid result % on OET, so not pressing my luck.

Hi guys.
Here's my feedback regarding Lollipop and boinc/wcg.
I am using 10 smartphones of the brand MLAIS, model M52 that I got a week ago. The phones are running on Lollipop.
Cheap ARM mediatek octacore-phones that run at 1.7Ghz each! (just google for prices, ok, I am not here to advertise. Use youtube to check reviews). So 10 phones made me add 80 cores... 80 computing days PER 24 HOURS.
My feedback is that I have running FAAH(Vina) at an average of +-23.000 credits for every 3 day of crunching per device (so a total of 10 phones at 23k credits each every 3 days.) That is what WCG-device statistics is telling me.
All devices run latest boinc 7.4.41 soft.

Now for the price I have paid for those smartphones, that is absolutely WHOPPING!!!!!! Initially, I wanted to go with raspberries but they couldn't keep up with the benches (with focus at the badges!) :D :D :D

Now, FAAH Phase1 is ending, the devices have crunched some OET-units. This works perfectly fine as well, BUT the credits seem wrong. The devices are only requestion about 2-3 credits for one unit.

Example:
OET1_0000886_xEBGP-FW_rig_18595 1.29 / 1.30 (CPU times) 2.2 / 3.4 (claim/credit)

on the other hand, I have:
OET1_ 0000864_ xZAGP-F_ rig_ 64957_ 0-- 12.17 / 12.44 (CPU times) 101.3 / 0.0 (claim/credit)
NB: No credit yet, since pending validation.

So you can see a huge difference between the first unit that the latest boinc agent only requested about 2 credits for over one hour of work, and the second one where the agent requested about 100 credits for 12 hours of work.

I think, it would be better to have this a littble bit more steamlined if that is possible. I also regret that WCG lacks GPU-capabale projects, but that would be a different topic. ;)

Enjoy guys, crunch on.
Chris!

Hi Sek! As for me, I thought of the GPU in general, not specially on the smartphone. Dev team would do well to further port projects to GPUs (no matter if AMD/NV/Intel, discrete or APU) so I do not expect them to bring anything like that over to Android anytime soon! I also think, cycles are shorter there, making it harder to have sth stable for each chip. I could be wrong though...

Anyway I would wonder if that would require any forced cooling, because running 8 cores 24/7 is already getting devices into some temps. Adding the temp of a non-stop running MALI-chip maybe not the very best idea, though I did not look up the max temp spec the ICs I own (760). ARM-website didn't mention temp-spec and I didn't want to make a deep search. About your charger, if USB-out is rated 2.1amps, that's as high as you can get... right now.

I am looking forward the next Boinc-agent for Android to see if it will bring any improvement on the "credit request"-side.

Chris!