leowt changed the topic of #linux-rockchip to: Rockchip development discussion | http://linux-rockchip.info | http://irclog.whitequark.org/linux-rockchip
vinifr has joined #linux-rockchip
arokux2 has quit [Quit: Leaving]
hipboi has joined #linux-rockchip
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-rockchip
Galland has quit [Quit: Saliendo]
vinifr has quit [Quit: Saindo]
bethania has joined #linux-rockchip
bethania has quit [Remote host closed the connection]
Astralix1 has joined #linux-rockchip
Astralix has quit [Ping timeout: 264 seconds]
mrueg has joined #linux-rockchip
atiti has quit [Ping timeout: 264 seconds]
Theueip has joined #linux-rockchip
atiti has joined #linux-rockchip
Topgun100 has joined #linux-rockchip
rellla has joined #linux-rockchip
hipboi_ has joined #linux-rockchip
AstralixNB has joined #linux-rockchip
hipboi has quit [Ping timeout: 260 seconds]
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-rockchip
mmind00 has joined #linux-rockchip
Topgun100 has quit [Quit: Topgun100]
<hipboi_> hmm... the rk serial chips can boot from sd card
<AstralixNB> hi hipboi
<AstralixNB> 2918 and 3188 can boot from SD, yes
<AstralixNB> 3066 does not
<AstralixNB> that is, what rk told
Topgun100 has joined #linux-rockchip
vinifr has joined #linux-rockchip
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
hipboi_ has quit [Quit: Leaving]
tonikasch has joined #linux-rockchip
tonikasch has quit [Client Quit]
tonikasch has joined #linux-rockchip
vinifr has quit [Quit: Saindo]
Topgun100 has quit [Quit: Topgun100]
Topgun100 has joined #linux-rockchip
Galland has joined #linux-rockchip
<Galland> hi
<tonikasch> hi, hola
<Galland> :) ;)
<Galland> just discovered yesterday that the RK3188 does not have SATA output, and its Ethernet interface is just 10/100 (not gigabit)
<Galland> should've checked earlier the p.brief :(
<tonikasch> :s
<Galland> so in the end the only alternative interface for integrating any of those is the USB2.0
<tonikasch> ufff, 2.0 is not a good deal
<tonikasch> :(
<Galland> but one of the worse things is the lack of a temperature sensor inside the chip
<Galland> there is no way to know if CPU freq should be lowered based on that
<Galland> dangerous overclocking :S
<tonikasch> and adding a temperature sensor and connecting it to GPIO would be a hard task, I guess
<Galland> I was thinking on sticks, where there are no pins accessible
<Galland> but then... radxa board is coming (RK3188 + expansion headers)
<Galland> one of my ideas for it is to do just what you proposed tonikasch :)
<tonikasch> o
<tonikasch> oh, yes, in the radxa board GPIO pins are accesible :D
<Galland> and one LCD output, and SPI/I2C, ...
<Galland> I'm really looking forward to it
<Galland> connect all that to a quad core CPU and 2 GB RAM and... wow
<tonikasch> hehe, I was only wondering about customs taxes
<Galland> hahahaha had a recent experience with that
<Galland> almost doubled my purchasing price for some components :(
<tonikasch> yeah, a year ago I bought a chinese mobile phone and, well, I didn't have to pay as much as the double, but somehow something expensive for me
AstralixNB has quit [Ping timeout: 264 seconds]
<tonikasch> well, i go to the swimming pool, cu
Topgun100 has quit [Quit: Topgun100]
Topgun100 has joined #linux-rockchip
AstralixNB has joined #linux-rockchip
Topgun100 has quit [Quit: Topgun100]
Topgun100 has joined #linux-rockchip
Topgun100 has quit [Client Quit]
Topgun100 has joined #linux-rockchip
Topgun100 has quit [Client Quit]
Topgun100 has joined #linux-rockchip
arokux2 has joined #linux-rockchip
atiti has quit [Read error: Operation timed out]
atiti has joined #linux-rockchip
tonikasch has quit [Quit: Bye!]
AstralixNB has quit [Quit: Leaving.]
atiti has quit [Ping timeout: 246 seconds]
naobsd has joined #linux-rockchip
mmind00 has quit [Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/]
tonikasch has joined #linux-rockchip
Omegamoon has joined #linux-rockchip
<Omegamoon> Is that @Galland over here?
<tonikasch> yes
<tonikasch> but perhaps not on keyboard...
<Galland> hi Omegamoon
<Galland> :D
<Omegamoon> That's weird, thought he was tired of rockchip ;-)
<tonikasch> haha
<Galland> nah, it was more that I'm overblown with work on the dayjob
<Omegamoon> Work... Forget about that, which is a good sign
<Galland> so can't dedicate as much time as before to RK, at the moment
<Galland> :)
<Galland> I'll throw a question out in the air:
<Galland> at times I get the screen blank for a half-second while on Linux on the RK31 (also happened in RK30)
Omegamoon has quit [Read error: Connection reset by peer]
<Galland> it happens much less when I clocked the DDR at 500 MHz instead of 400
Omegamoon has joined #linux-rockchip
<Galland> and seems to happen more when I do something memory intensive
<Galland> so I was thinking the blanking was due to the fb not getting enough "memory-time" to get the pixels out to the screen
<Galland> in normal apps that only means they are slower, but the screen is a realtime thing, so it would be a reasonable outcome for it to go blank
<Omegamoon> Sounds to me more of a sync problem
<Galland> and was wondering if somebody knows if/where does the kernel give more "memory-access-priority" to one task over another
<Galland> hmmm sync
<Omegamoon> Vertical sync
<Galland> could be, actually the problem is a half-second black screen every 1 hour or so
<Galland> with lower memory frequency it happens much more often
<Galland> and once I got it to be repeatable if I opened a certain zip file
<Omegamoon> Sync problems could lead to this behavior in the end
<Galland> it would become blank just when file-roller was opening
<Galland> that was what made me think of the memory
<Omegamoon> Havent seen it... Yet
<Omegamoon> Increasing the freq is just a workaround I guess
<Galland> but talking about sync, the communication with the external hdmi tx chip (ITE...) is done, afaik, by the lcdc IP core inside the rk31
<Galland> once it is configured with a certain screen modeline
<Galland> why would it fail a certain vsyncs?
<Galland> :S
<Galland> it's possible, but then the file-roller thing...
<Galland> I'll keep trying to get a repeatable setup
<Galland> I'll see if I can find a memory intensive app to discard one point or another
<Omegamoon> You know what the hdmi driver looks like do you?
<Omegamoon> It's crap!
<Galland> :D
<Galland> actually phjanderson's fix for the video smoothness problems in android
<Omegamoon> @tonikasch, have you experienced this problems?
<Galland> was related to a vsync interrupt
<tonikasch> Omegamoon, mmmm, I have Linux3188 kernel compiled for a mk808 but I don't use the video in it :(
<tonikasch> only remote access
<Galland> I do code on my rk3188 stick :) gotta love that complete silence at night
<Galland> I'm writing from it right now
<tonikasch> :D
<Galland> plus I recently measured power consumption
<tonikasch> I'll move the rk3066 to my tv and let the newcoming rk3188 do the server stuff
<tonikasch> how much was it, Galland?
<Omegamoon> But this "fix" had side effect didn't it?
<Galland> thinking it would be higher than on the rk30 (see http://hwswbits.blogspot.com/2013/03/rk3066-measy-u2c-stick-power-consumption.html )
<Galland> but it turned out to be lower
<Galland> rk3066 is implemented on 40nm sillicon
<Galland> while rk3188 targets 28nm
<Galland> that must be it, even with 4 cores
Omegamoon has quit [Read error: Connection reset by peer]
<Galland> gotta blog about it some of these days
<Galland> I'll wait for a MK802IV that is coming my way to make the measurements using the internal wifi
<tonikasch> :) I'm waiting for a minix x7
<Galland> pretty box that is :)
Omegamoon has joined #linux-rockchip
Omegamoon has quit [Read error: Connection reset by peer]
Omegamoon has joined #linux-rockchip
<tonikasch> :)
<Omegamoon> Are there any efforts in here to get Mali working in custom kernels?
<Galland> over Linux?
<Omegamoon> Since it sucks to keep on using the android modules
<Omegamoon> Over linux... And android also
<Omegamoon> Fully build custom kernel from source I mean
<Omegamoon> Not needing any rockchip modules any longer
<Galland> understood, no libMali.ko
<Galland> :)
<Galland> other than the cusstom built one
<Omegamoon> How would we be able to upgrade to lets say 3.1+ kernels
<Galland> Mali did work from source on Linux for RK3066 devices
<Omegamoon> I'll trough in a 3.0.72+ kernel when I'm back home...
<Omegamoon> And then the problems begin already
<Galland> but porting the same changes to the RK3188 kernel codebase did not work
<Omegamoon> Changes should be minimal
<Omegamoon> Rockchip uses the libmali for rk30 as well as rk31
<Omegamoon> The same one I mean
<Omegamoon> So there's no new magic in there
<Galland> hmmmmmm
<Galland> using https://github.com/ssvb/tinymembench to stress test memory access
<Galland> the app seems to be cpu limited (25% usage, that's one full core)
<Galland> but no black screen so fatr
<Omegamoon> Wasn't the trick on rk30 just to add some hooks here and there?
<Omegamoon> For mali?
<Galland> yep, on lcdc and fb parts mostly
<Galland> I tried to make separate commits for it
<Galland> let me look for them and paste them here for reference
<Galland> (fortunately I am ample with commit messages)
<Omegamoon> fb parts have changed considerably in rk31...
<Galland> hey, remember what I said about using sunxi's fb driver?
<Galland> for rk it at least takes advantage of the NEON ISA
<Galland> well, it looks like memory reads using NEON are more than twice faster than without it
<Omegamoon> Hmm, could work
<Galland> still running the tests
<Omegamoon> Too bad there seems to be no one around here interested in this
<Galland> two people better than none :)
<Galland> Rome wasn't built on a day
<Omegamoon> Is this a "lurkers only" channel? ;-)
<Galland> lol
<tonikasch> :Pç
<Omegamoon> Hey, there are others!
<Omegamoon> No, don't leave now :-)
<tonikasch> sorry, telf
<Galland> C copy prefetched (32 bytes step) : 502.3 MB/s (2.0%)
<Galland> NEON copy prefetched (32 bytes step) : 886.0 MB/s (8.2%)
<Omegamoon> Nice
<Galland> telf = phone
<Galland> ;)
<Galland> this tool is done by one of the top contributors to linux-sunxi's fb driver
<Galland> top notch developer
<Omegamoon> Ah, it's Spanish? ;-)
<Galland> telefono :D
<Omegamoon> It sound like a nice acceleration
<Omegamoon> Haven't looked into the sunxi fb code
<Galland> I'm currently using the sunxi fb driver on the rk31 (NEON)
<Galland> and no problem at all
<Omegamoon> Would it need extra adjustments to make it work on the rk31?
<Galland> can't "feel" much of a difference from the default one, but the NEON thing lured me in
<Omegamoon> Ah, that's a no then
<Galland> and it's being actively developed
<Galland> if we can some day snatch the "RK3188 RGA.pdf" or "RK30xx RGA.pdf" (2D accelerator IP core)
<Omegamoon> But it's no hatdware accelleration...
<Galland> we could add a module for it, they've coded it very nicely to do that
<Galland> RGA is actually used for accelerating 2D operations like scaling
<Omegamoon> Hatdware... Hardware I mean
<Galland> RGA is an IP core within the RK3188/RK3066, it's hardware
<Galland> just read it yesterday on the TRM
<Galland> and p.brief
<Galland> the sunxi fb driver is very nicely coded
<Galland> for example, the 2D accelerator in Allwinner chips is accessible as a module (sunxi_x_g2d.c)
<Galland> same could be done for RK's RGA
<Galland> if only we had the docs...
<tonikasch> hey, sorry, I was on the phone
<Galland> np ;)
<tonikasch> docs, docs, docs... or that git source code :P
<Galland> you mean the bugged source code? :D
<tonikasch> yes... but perhaps the full source code
<Omegamoon> The full source code...
<Omegamoon> Life would be so much easier... If only we had the full source code of it
<tonikasch> yes...
<Galland> actually something may be inferred from the code: ./drivers/video/rockchip/rga/
<Galland> that's better than nothing
<Galland> I particularly like rga_reg_info.c
<Galland> exposes clean and intersting functions
Omegamoon has quit [Quit: Bye]
<tonikasch> oh, well, I have to leave now, some friend have come
<tonikasch> see you, Galland
tonikasch has quit [Quit: Bye!]
JochenKauz has joined #linux-rockchip
atiti has joined #linux-rockchip
Topgun100 has quit [Quit: Topgun100]
fuzufaza has joined #linux-rockchip
fuzufaza has left #linux-rockchip [#linux-rockchip]
<eebrah> !quote nj3ma
tinti has quit [Read error: Connection reset by peer]
Theueip has quit [Quit: Leaving...]
<Galland> RK30XX TRM p.736: "Chapter 23 - RGA: RGA is a separate 2D raster graphic acceleration unit. It accelerates 2D graphics operations, such as point/line drawing, image scaling, rotation, BitBLT, alpha blending and image blur/sharpness."
<Galland> and p. 695: "Chapter 21 - IPP (Image Post Processor): Image Post Processing (IPP) is used doing image scaler, deinterlace and rotation. Each processing can be done independently or combined with others."
<Galland> both can scale and rotate hmmmm
<Galland> "For detailed information about RGA,please refer to RK30xx RGA.pdf."
<Galland> "For detailed information about IPP,please refer to RK30xx IPP.pdf."
<Galland> Gotta love that
arokux2 has quit [Remote host closed the connection]
Topgun100 has joined #linux-rockchip
JochenKauz has quit [Quit: Leaving.]
tonikasch has joined #linux-rockchip
vinifr has joined #linux-rockchip
Topgun100 has quit [Quit: Topgun100]