<plaes>
and the design library which they provide seems to include only A10 part
<libv>
*sigh*
<plaes>
and it's for eagle :)
arokux2 has joined #linux-sunxi
<arokux2>
wens: which lubuntu was it exactly, can you post your dmesg?
<plaes>
libv: wonder whether it's possible to as Cadence guys whether Itead people are using properly licensed software..
<arokux2>
Turl: have you got your ct?
<libv>
plaes: no idea
<libv>
plaes: i am however now going to fill out a wiki page on them, and i am going to find how i can lock that page down as well
adb has joined #linux-sunxi
<arokux2>
tomboy64: have you tried lubuntu?
<tomboy64>
arokux2: tell me how to boot it from a usb-stick and i will
hglm has joined #linux-sunxi
deasy has joined #linux-sunxi
Quarx has quit []
Quarx has joined #linux-sunxi
<arokux2>
tomboy64: cannot tell you, sorry.
<hglm>
I am surprised how well/fast this A20 tablet is running after being configured properly -- the default script.bin file provided was totally screwed up and Android was unbelievably crippled (lockups, low speed, power-grabbing governor, but completely avoidable it seems).
Quarx|2 has joined #linux-sunxi
<hglm>
The memory clock was set to 432 MHz in script.bin which is totally unstable on this device! Changed it to 408 MHz and now everything is rock-solid so far...
<plaes>
hglm: please add to wiki
<hglm>
OK, I will.
Quarx has quit [Ping timeout: 272 seconds]
<hglm>
Also the ondemand governor works very well but the default threshold settings are awful -- change them and you get a very fast/responsive system that still revs down to low speed when idle (unlikely the erratic behaviour of the default fantasy governor).
<hglm>
Clock for clock single threaded performance seems to be quite a bit faster (about 80%) on the A20 than the A10 at the same clock speed (probably due to Cortex A7 vs A8 and better caches/memory interface).
vrga has joined #linux-sunxi
vrga has left #linux-sunxi [#linux-sunxi]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<ncrmnt>
Hello all, just wonering do recent A20 do ddr frequency scaling like RK3188?
<libv>
i will set this page's protection to admin only later today, after some feedback has been gathered on this page
<hglm>
ncrmnt: Don't know but in linux-sunxi tinkering with memory timings is only done in the bootloader (not kernel), although the kernel does support CPU voltage and frequency scaling.
<ncrmnt>
hglm: Hm... Good that's not become a tradition. In rockchip kernel they are doing ddr freq/voltage scaling in runtime with a super-special buggy kernel hack.
<ncrmnt>
Stock sunxi kernel from BSP (before all the fun with linux-sunxi started) now looks neat and polished to me after I've seen stuff rockchip does.
tomee^ has joined #linux-sunxi
<plaes>
libv: maybe add oshw_abuse category?
Fusing has joined #linux-sunxi
<libv>
plaes: i doubt that an assholes category will be acceptable, so "oshw_abuse" seems like a better option
vinifr has joined #linux-sunxi
ncrmnt has quit [Ping timeout: 241 seconds]
tomboy64 has quit [Ping timeout: 240 seconds]
n01 has quit [Remote host closed the connection]
jemk has joined #linux-sunxi
maz_ has quit [Ping timeout: 272 seconds]
rellla has joined #linux-sunxi
tomboy64 has joined #linux-sunxi
Fusing has quit [Ping timeout: 245 seconds]
Fusing has joined #linux-sunxi
ncrmnt has joined #linux-sunxi
hglm has quit [Quit: leaving]
<oliv3r>
atsampson: there's probably more odities :)
<oliv3r>
Turl: wens, gerwin_j: A80 will be powerVR
<wingrime>
jemk: I think, according a20 manual cedar have internal alphamixer
<wingrime>
jemk: in ISP
adb has quit [Ping timeout: 252 seconds]
hglm has joined #linux-sunxi
<jemk>
wingrime: yes, but have you found any code that uses it to find out how it works?
TomiK has joined #linux-sunxi
<hglm>
tomee: Regarding those glmark2-es2 benchmarks I think it is probably an A10 with the vsync locked at 60 Hz. I don't think the Mali in the A20 is that much faster than the one the A10 (A20 has low Mali clock speed) but I got something like 150 fps final score running glmark2-es2 on my A20 device with vsync unlocked and fb0_framebuffer_num set to 3 in script.bin.
<wingrime>
jemk: there is some parts in h264 encoder
<wingrime>
jemk: that defenetly use ISP
<wingrime>
jemk: But I not sure alpha blend
<wingrime>
jemk: I think, ISP not so difficult, and we can figure out register usage only guessing
<wingrime>
jemk: TRIG+CTRL+INBUF+OUTBUFF
<tomee^>
hglm: yeah, exactly, my score is 152.
<tomee^>
hglm: I wonder what it would be if I managed to build an r3p2 kernel module...
<hglm>
tomee: Yeah, there might be some improvements in more recent Mali blobs. Theoretically it may also be possible to tweak Mali GPU speed on the A20.
peje has joined #linux-sunxi
rz2k has joined #linux-sunxi
<tomee^>
hglm: someone said the improvement would be ~2x. r3p0 only uses 1 core if I am not mistaken?
<tomee^>
hglm: but aside from the blob, I need a kernel module first ;-)
<hglm>
tomee: That could well be the case. Performance on the A20 seems to be too low compared to A10 given the two cores in the Mali400MP2.
<hglm>
I have tested some of the my own GLSL shaders/3D engine and it doesn't do badly but slows down on more complex shaders (e.g. normal mapping not implemented in the most optimal way) that work fine on PC-class hardware.
<libv>
tomee^: 1PP instead of 2
<libv>
this is different from "a" core.
<libv>
mali is a split shader design
<tomee^>
libv: didn't read the datasheet
<hglm>
Yeah so triangle rate is unchanged from MP1 to MP4 at a given clock speed, but fillrate is basically quadrupled.
<tomee^>
libv: thanks for pointing that out
shineworld has joined #linux-sunxi
<hglm>
libv: Does the shader compiler in the Mali driver optimize to the extent that regular OpenGL shader compilers do? Or does it need more tuned shaders?
<libv>
hglm: ?
<libv>
hglm: "regular openGL shaders"? means what?
<hglm>
I mean PC class like nvidia/AMD opengl drivers.
<libv>
hglm: what do you want to find out with this?
<libv>
yes and no.
peje has quit [Quit: Konversation terminated!]
<libv>
the ARM compiler is more compile time optimized than it is run-time optimized
<libv>
which is a necessary tradeoff on this hw
<libv>
there are a few corners where it could do better
<hglm>
OK
<libv>
but imho that is not important
<libv>
most of the speedups in mesa are not to be gained from the shader compiler
<tomee^>
did anyone try building that: DX910-SW-99002-r3p2-01rel2? downloadable from cubieboard.org and the Mali vendor
<libv>
i just got 50% more out of specific tests just by tracking whether index buffers were dirty, and by not recounting the indexes
<libv>
these sort of easy optimizations are still left open in mesa
<rz2k>
tomee^: r3p2 will have the same problem on cubietruck as you have on r3p0
<libv>
and i will be pushing to get at least that change upstream
<tomee^>
rz2k: the memory issue? yeah, I never counted it would not.
<tomee^>
rz2k: but frankly, a 50% performance boost in Mali is more important to me than 100% boost of available memory ;)
<rz2k>
tomee^: ok, then you need to contact ssvb, he already ran r3p2 on A20 a lot
<libv>
tomee^: it's a mostly 100% boost for normal workloads
<libv>
especially when the buffersize goes above like 800x480
<tomee^>
rz2k: yeah, I just pulled his kernel from github and applied a gmac driver to it.
<tomee^>
rz2k: I wonder if it will even boot ;)
<tomee^>
libv: that does sound tempting
<rz2k>
tomee^: also you need to use r3p2 branch of fbturbo (plus fix a little bug there - stray return when checking version)
<rz2k>
or use the official drivers
<rz2k>
for xorg part, I mean.
Quarx has joined #linux-sunxi
<libv>
ah, of course, you will not see the equivalent change in framerate if you depend on the cpu to copy the final buffer over
Quarx|2 has quit [Read error: Connection reset by peer]
<tomee^>
rz2k: can you elaborate on which bug I must fix?
<tomee^>
libv: I'm not sure what you mean. I mean I know what you mean but I am not sure what you are referring to - UMP platform vs. OS backend?
<libv>
tomee^: you will not be able to fully exploit the 2x increase in what the mali will deliver, as other parts are involved when using X11
<tomee^>
libv: ah, that, yes, of course.
<tomee^>
libv: my need for x11 is mostly because of the vdpau anyway
<hglm>
libv: Regarding Mali drivers in framebuffer mode, I do see some tearing or region copy effect (some screen areas seems to be updated earlier than other parts), it doesn't seem to be doing tear-free page-flipping). Is that correct?
AreaScout has quit [Ping timeout: 245 seconds]
JohnDoe_71Rus has quit [Ping timeout: 245 seconds]
AreaScout has joined #linux-sunxi
Fusing has quit [Ping timeout: 246 seconds]
<tomee^>
boo, /dev/g2d is missing from that kernel
ykchavan has joined #linux-sunxi
<tomee^>
but the overall score seems a bit better...
<hglm>
tomee: Some of those subtest benchmark scores are much higher than what I got with my A20. For example the jellyfish benchmark I think was 40fps and your result shows 126 fps. The actual real-world performance increase may be larger than indicated by the "glmark2 Score".
<tomee^>
hglm: would probably be even better if I managed to transplant g2d into that kernel
<hglm>
tomee: Not sure, I think the Mali integration in fbturbo uses the "disp" layer interface, not g2d (which is 2D graphics bitblt acceleration).
<tomee^>
oh
shineworld has quit [Quit: Leaving]
<tomboy64>
will i want ION memory management for building linux? or is that something android specific? (A20/mail 400)
derethor is now known as printallthething
derethor_ has joined #linux-sunxi
derethor_ has quit [Client Quit]
printallthething has quit [Quit: Leaving]
printallthething has joined #linux-sunxi
FunkyPenguin has quit [Ping timeout: 265 seconds]
FunkyPenguin has joined #linux-sunxi
<tomee^>
and the spdif is broken in the r3p2 branch, too
<tomee^>
*sighs*
<tomee^>
looks like I have to go the other way around
<rz2k>
tomee^: to use disp-accelerated hw layers with fbturbo, check what fbturbo said to you in /var/log/Xorg.0.log, I bet you have the message there about it
jemk has quit [Quit: bye]
<rz2k>
tomee^: nobody stops you from figuring out what r3p2 branch from ssvb is forked from and merge the changes for spdif from stage/sunxi-3.4 above.
<tomee^>
rz2k: [ 1053.575] (II) FBTURBO(0): Creating default Display subsection in Screen section [ 1053.597] (II) FBTURBO(0): using sunxi disp layers for X video extension [ 1053.632] (II) FBTURBO(0): enabled display controller hardware overlays for DRI2
<rz2k>
interesting, usually with standard script.bin it will complain
<tomee^>
rz2k: yeah, I know. but I can see that led interface is also missing ;)
FunkyPenguin has quit [Ping timeout: 246 seconds]
FunkyPenguin has joined #linux-sunxi
<oliv3r>
wingrime: a60 is rumoured to be mali
hglm has quit [Quit: Lost terminal]
torindel has quit [Ping timeout: 272 seconds]
torindel has joined #linux-sunxi
<buZz>
will a60 also be a bigLITTLE system?
<Wizzup>
or just a littleBIG
<buZz>
:P
<libv>
oliv3r: rumoured or hoped?
Quarx|2 has joined #linux-sunxi
Quarx has quit [Ping timeout: 246 seconds]
<arokux2>
has anybody connected ct with vga cable+vga2dvi converter?
FunkyPenguin has quit [Ping timeout: 272 seconds]
<wingrime>
arokux2: vga works but badly
<wingrime>
arokux2: wifi interfere with VGA
<arokux2>
wingrime: I see only black screen with android logo
<wingrime>
arokux2: android worked fine
<arokux2>
wingrime: did you have this case: vga cable+vga2dvi ?
<wingrime>
arokux2: no, I used only VGA
<arokux2>
ok, this can be a problem
FunkyPenguin has joined #linux-sunxi
<arokux2>
wingrime: have you flashed lubuntu to uSD?
ncrmnt has left #linux-sunxi ["Once you know what it is you want to be true, instinct is a very useful device for enabling you to know that it is"]
vinifr has quit [Ping timeout: 245 seconds]
<wingrime>
arokux2: I use own made debian
<arokux2>
wingrime: does wifi works for you in this debian?
<ZetaNeta>
and i am compiling it through adb over tcp....
<ZetaNeta>
"Great!"
<ZetaNeta>
i guess next thing il do, is get the gentoo on the tablet.
wingrime has quit [Ping timeout: 248 seconds]
<libv>
rz2k: yeah, that was truly astounding
rm has joined #linux-sunxi
<tomboy64>
ZetaNeta: you have given the -mcpu switch, meaning if xscale has one itsy bit of optimization that a10 doesn't, the binary will crash
<tomboy64>
ZetaNeta: so, unlikely.
<ZetaNeta>
tomboy64, sad....
<ZetaNeta>
but... what a armv5 cpu can what a A10 cant?
<ZetaNeta>
armv5tel
<ZetaNeta>
PXA255
<arokux2>
Turl: do you have your ct already?
<tomboy64>
ZetaNeta: hmm. think of a diesel car from 1950 and a petrol car from 2010. the 1950 car can't do anything that the 2010 can - but it will still not run on diesel.
<ZetaNeta>
which was made only satan him self knows when
<ZetaNeta>
tomboy64, Maybe you wanted to say that the new one wont run on petrol?
<ZetaNeta>
oh...
<ZetaNeta>
otherway round
<ZetaNeta>
but...
<ZetaNeta>
nah, confused
<ZetaNeta>
well, lets see if the binaries will run
adb has joined #linux-sunxi
<ZetaNeta>
the unoptimized binaries for armv5tel works...
<arokux2>
rz2k: it is supposed to have everything configured, or no?
<tomee^>
arokux2: and one more thing, I haven't investigated the bluetooth part, but it seems to be connected to an UART. whether it is presented as /dev/ttyX or /dev/ttySX, no idea, but I disabled getty on /dev/tty1 to be safe.
<tomee^>
...all I wanted was a cheap board that would do 720p decoding, have SATA and toslink
<rz2k>
please, try standard solutions, like, many routres will send you to home if you have date of your system being 1970
<tomee^>
and I will end up with a god damn oscilloscope hooked to the pins of the ICs
<tomee^>
it's always like this ;)
<arokux2>
rz2k: 1970?
<tomee^>
arokux2: bad RTC/time
<tomee^>
arokux2: wifi PSK is time-based
<arokux2>
tomee^: where is it?
Gerwin_J has quit [Quit: Gerwin_J]
<tomee^>
arokux2: where is what?
<rz2k>
arokux2: if you have bad time in the system, DHCP/WPA auth will go nuts
<arokux2>
rz2k: where do you see I have bad time?
<buZz>
all allwinners have bad time :P
<rz2k>
I dont see that, that is just a standard problem
<rz2k>
that I encounter lot of times
<buZz>
i think there is just one? board with batterybacked rtc
<rz2k>
because many boards have no RTC
<rz2k>
cubietruck has battery for RTC by the way
<arokux2>
rz2k: i'm on cubietruck
<tomee^>
buZz: noo, the ct1 has a real RTC. at least in the docs ;)
<rz2k>
but it didnt have proper time from the start
<buZz>
yeah, thats the only one, isnt it?
<buZz>
tomee^: yes, but no batterybackup
<rz2k>
arokux2: attach ethernet and run # ntpdate ntp.corbina.net
<rz2k>
then try wifi again
<arokux2>
ok..
<tomee^>
... having unplugged the ehternet first
<tomee^>
or you will also end up in trouble
<tomee^>
if the dhcp server is the same
<tomee^>
buZz: none at all? no capacitor? well, nicely done.
<buZz>
yes its p00
<buZz>
p00p
<buZz>
:)
<rz2k>
tomee^: networkmanager will route things right
<tomee^>
buZz: a ds1307 breakout board with a cr232 battery holder cost me like $1 last time I bought it in retail
<rz2k>
(epoc moment: first thing in the history said about networkmanager not blaming it)
<rz2k>
s/c/ch/
<mripard>
torbenh3: wens: who worked on gmac in u-boot ? is there any public branch for it ?
<tomee^>
rz2k: networkmanager is evil.
<buZz>
tomee^: dude, 1 dollar is a lot
<rz2k>
everything is evil, lets go live in the woods.
<buZz>
if you want to make 100000 boards
<rz2k>
I did an attachment board to my WandboardQuad with DS1338 for the same purpose
<tomee^>
buZz: yes, yes, I know. "a capacitor and a pull-up? can't we just go with a resistor?"
<tomee^>
buZz: "hmm, well, the signal slope is almost within bounds. maybe it will work. we can try"
<tomee^>
ok
<tomee^>
I'm off to watch some stuff on my still impaired media center
<tomee^>
albeit showing promise, the r3p2 drivers made a difference today...
Quarx has quit []
<tomee^>
now, just implement vdpau with layers
<tomee^>
and in about a month I will have a fully working setup
<tomee^>
I will lose my daytime job in the meantime
<tomee^>
but who cares?
<rz2k>
I got my current job doing the support around linux-sunxi ML
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
<rz2k>
dude from local embedded company emailed me with demand for embedded linux developers and that my experience will be enough, turned out it was really enough.
<rz2k>
also I got lucky and this dude lives really near me and we are same age.
eebrah|mobile has quit [Read error: Connection reset by peer]
<tomee^>
rz2k: given I managed to fit a fully-buffered 16x2 menu and 2 PWM drivers into 2048kB of Atmega SRAM I think I would be able to deal with embedded linux ;) but nonetheless, they pay me for taking care of bigger toys. nevermind. o/
<tomee^>
[d]
<tomee^>
errr, s/2048kB/2048B/
<rz2k>
nice
tzafrir has joined #linux-sunxi
<buZz>
right, i figured 2MB would be a bit much :)
<arokux2>
tomee^: have you used wpa_supplicant that was in the image?
<rz2k>
arokux2: setting date didnt change anything?