<arete74>
09:35 < oliv3r> well we might have ahci soon
<arete74>
09:48 < hno> mnemoc, only because there was not time to produce the needed
<arete74>
amount of boards in time.
<arete74>
bad paste ;-)
<oliv3r>
arete74: very bad :p
<hno>
ganbold, more on that in 2 weeks.
popolon has joined #linux-sunxi
* ganbold
thinks hno runs factory that makes beaglebones
<hno>
I don't.
<ykchavan>
hno, for one a10 tablet, I am getting different information in script.fex and bootinfo
<ykchavan>
dram_chip_density = 2048 in bootinfo but script.fex says dram_chip_density = 1024
<hno>
ykchavan, that is normal. You should also see difference in dram_io_width.
<ykchavan>
what are the key fields that should match?
<hno>
livesuit mistedetct 8-bit dram chips as 16-bits.
<hno>
and if visual inspection of the board shows that the dram chips are 8-bit then we correct script.fex.
<hno>
density=2048, io_width=16 is equal to density=1024 io_width=8
<hno>
with same bus_width.
<rellla>
ssvb, jemk: to sum up - libcedarv/linux-armhf2 is the newest we got from aw but isn't used in any implementation. neither xbmc nor vlc. both use libcedarv/linux-armhf. xbmc speeks with it directly from source, vlc via willswangs wrapper libcedarx.
<rellla>
it would be the time, to give the new blobs a chance ;)
<mnemoc>
hno: can't cubietech get some factory time faster when doing a larger request like yours?
<mnemoc>
:(
hramrach has quit [Ping timeout: 240 seconds]
massi_ has joined #linux-sunxi
<soldoKyn>
ssvb: tried right now performance cpu governor. glmark2-es2 score running in x11 with 800x600 window rises from 113 to 131!
<ssvb>
soldoKyn: if you are not running on a battery and don't care too much about power consumption, then using ondemand governor makes no sense, it it is really bad for the desktop responsiveness
<ssvb>
soldoKyn: performance is the way to go
<rm>
won't it overheat?
<ssvb>
rm: I don't think so, power consumption also depends on the cpu load, the system does not consume much when idle
<ssvb>
rm: and if you are running something like cpuburn, the ondemand governor does not help anyway
<rm>
sure
<rm>
but in normal usage, with the right tweaks, I think it can reduce frequency without noticeably "getting in the way"
<ykchavan>
hno, which fields are important when matching script.fex of two tablets? (besides product version and machine)
hramrach has joined #linux-sunxi
<soldoKyn>
ssvb: rm: for now I'm only testing the average/peak level of performance that the board can reach. the heat for now is not so much so it's ok! I'll try that tweaks if the heat becomes too high! thank you!
<hno>
mnemoc, just almost.. was a couple weeks short.
<hno>
ykchavan, product version & machine is very often not filled in correctly so quite useless to match on. Which leaves the rest...
<mnemoc>
hno: :|
<arokux>
hno, what bones, have I missed something? I feel like ppl here are having yet another source of info..
<arokux>
:)
<ykchavan>
how about dram_para, disp_init and lcd0_para?
* hno
have all his bones intact. None missing or broken today.
<arokux>
hno, beaglebones :p
<hno>
ykchavan, all are relevant. And all can differ slightly even for the same hardware..
<ykchavan>
So, Do we need to create sys_config for each tablet?
<hno>
arokux, it's a modified beaglebone black in this project. Was hoping for a10s or a20, but was to time constrained this time.
<hno>
ykchavan, yes, each product have their own fex file. Even a change in LCD panel or touch screen requires a new fex even if everything else is the same.
<ykchavan>
In that case, my table of 'Known tablets with boardtype' is irrelevant.
<hno>
u-boot do not really need as many different configurations as there only the dram_para section matters, but easier when matching.
<tgaz>
is there any chance of getting a "usb gadget" console on a cb2 right now? if not, what's missing?
<tgaz>
(in linux-sunxi 3.4)
<hno>
tgaz, should work I think, but haven't tried. Maybe something is still missing for the OTG driver to make gadgets work, don't know. It's always been shaky on all sunxi generations..
<tgaz>
k
<tgaz>
driver or hw problem?
<tgaz>
hno: what about the same in u-boot?
<arokux>
hno, I see. but where from the others know about it? :) (I remember you were talking about this a week ago here on irc)
<tgaz>
[09:20] *hno buries himself in beaglebones for a while again, have some 5K of them that need to work in about a week..
<arokux>
tgaz, so you've actually found this message in logs?
<tgaz>
arokux: yes, from today
<tgaz>
it was a /me
<arokux>
cubieboard.org drowns in spam
<arokux>
tgaz, oh, I've missed it.
<mnemoc>
arokux: poke simon in #cubieboard, i think he still the admin
<arokux>
mnemoc, i've poked hipboi several times already
notmart has joined #linux-sunxi
<mnemoc>
arokux: tom is busy with radxa while benn is in charge of cubie
<mnemoc>
weird paradox
<oliv3r>
Bennboard
<arokux>
oliv3r, there is no problem of initramfs everything is built in. btw, this is default of multi_v7
<oliv3r>
arokux: what do you mean, full ubuntu in initramfs? :)
<arokux>
oliv3r, if you meant rootfs - it is nfs mounted.
<oliv3r>
ah ok
<oliv3r>
well i have a mainline testing initramfs
<oliv3r>
without nfs; so if i'd get a more featured initramfs, i'd probably still load it as such ;)
<oliv3r>
so i'd have a big one
<arokux>
oliv3r, buildroot busybox-based commands are very limited.
<oliv3r>
mnemoc: what's with that greg-young mail?
<oliv3r>
arokux: yeah i noticed; but for my limited tests it was ok for now
<mnemoc>
oliv3r: I can't access the ml from here....
<arokux>
oliv3r, same here, it worked was fast, but now i lack features so went for the full distro, but neither alip nor nano from linaro work.
<mnemoc>
oliv3r: what is with it?
<arokux>
mnemoc, are you blocked?
<mnemoc>
arokux: yes, restrictive office...
<oliv3r>
mnemoc: 'looking for a10 consultancy work, bla bla bla'
rellla has quit [Ping timeout: 240 seconds]
<oliv3r>
mnemoc: 'i need this done fast, who wants to work for me'
<arokux>
mnemoc, you are lucky they do not know about irc
<oliv3r>
mutt :p
<steev>
how much is he paying?
<mnemoc>
arokux: from the phone
<oliv3r>
steev: well he wants mtd drivers to be made proper i think
<oliv3r>
steev: he wants axp events to create input events
<steev>
ah
<oliv3r>
steev: he wants autologin for agetty :p
<arokux>
mnemoc, facepalm
<steev>
autologin on agetty isn't hard
<oliv3r>
yeah i faceplamed when i read that
<oliv3r>
looks like to me, someone is being lazy, hasn't done their research, and now wanna hire someone for some 'fast consultancy work'
<steev>
tell him autologin on agetty will cost $2000usd
<oliv3r>
i deleted the message allready
<steev>
too bad
<oliv3r>
if he doesn't even do some basic research, then I don't wanna spend time helping him
<steev>
i've actually gotten decent work from stuff like that
<steev>
got 7000 usd for 40 hours of work
gnufan has joined #linux-sunxi
<oliv3r>
if it was like 'looking to hire someone to intreate and fix mtd for a10' then that would be a serious request
<steev>
some people just have more money than time
<oliv3r>
oh i agree
<oliv3r>
but if you read the mail it reads a lot like 'we need this done, yesterday and we wanna be able to put it all on you'
<steev>
yeah
<oliv3r>
badly research request for support imo
<n01>
gh I'm an underpaid consultant ... 7000 usd for 40 hours of work ... sigh
<steev>
aww man, i want the cubietruck
<oliv3r>
steev: I should have mine soon!
<steev>
All this plus our application(s) must be installable from a single insertion of an SD card.
<steev>
wat
<n01>
We need all I/O systems functioning properly :))
<n01>
"are you even trying bro?" (cit.)
<steev>
ugh
<steev>
what IS the proper way to make an sdcard for the cb2?
<steev>
because what is in the wiki is wrong. dead wrong
jemk has joined #linux-sunxi
<arokux>
steev, what is wrong?
<steev>
every time i follow it... the first partition ends up completely wack
<steev>
[102821.834550] EXT4-fs error (device mmcblk0p1): htree_dirblock_to_tree:920: inode #2: block 350: comm gvfs-udisks2-vo: bad entry in directory: directory entry across range - offset=0(0), inode=3851235332, rec_len=12400, name_len=132
<steev>
maybe there's something wrong with my sfdisk
<arokux>
steev, that worked fine for me several time.
<arokux>
times*
<oliv3r>
steev: the bsp doesn't even work for me in that regard, on gentoo i _need_ partprobe
<oliv3r>
and yeah 'all i/o system functioning properly' really they just wanna hire someone to make the entire board fully workable :)
<steev>
oliv3r: well, i may be doing things a little differently here, i'm gonna do it without sfdisk here
<oliv3r>
well actually, if I can quite my daytime job and work fulltime on sunxi, sure; we can talk. I doubt they can afford that though
rellla has joined #linux-sunxi
<oliv3r>
steev: i have been doing it manually for ages now :)
<oliv3r>
rellla: have you gotten cedarX working on a20? patrick wood posted some patches
<steev>
oliv3r: exactly
<rellla>
oliv3r: i haven't tried it yet, i first want to make libvdpau and xf86 module work
<oliv3r>
rellla: fair nuff
<oliv3r>
rellla: i jus twant an XBMC sd card ;0
<rellla>
oliv3r: once successful there are some more tasks: try new cedarX blobs with xbmc, try pat's patches with a20 and vlc/xbmc/vdpau? and then look at xbmc with libvdpau
<rellla>
but smells like time-eating ;)
<oliv3r>
rellla: yeah
<oliv3r>
if everything where done with lima + cederus; you'd be done in a jiffy :)
* rellla
should wait for lima and RR
<rellla>
argh
<rellla>
*RE'ed cedar and the OS-XBMC-image oliv3r will then build :p
<oliv3r>
rellla: you are the xbmc man :p
<oliv3r>
why else would you get free dev boards! :p you are the xbmc-dev ;)
hno has left #linux-sunxi ["Lämnar"]
hno has joined #linux-sunxi
<rellla>
yeah, xbmc-dev. uahh. still not finished my "C for beginners" ;)
<steev>
good thing it's written in c++!
<rellla>
but already I know what a pointer is :)
<rellla>
oliv3r: i'll start/continue xbmc work as soon i find some free evening...
<oliv3r>
rellla: no rush :)
<oliv3r>
rellla: only the whole world waiting for you :)
<oliv3r>
and you don't need to code to get it working :)
<oliv3r>
just connect parts together; if it works, yay
* rellla
at least wants to tweak some piece of code
<jemk>
rellla: still no success with vdpau?
<rellla>
yesterday i had no time to try. maybe today after $work$
<rellla>
i'm sure, i'll succeed ;)
<oliv3r>
yeah $work; yuk
<oliv3r>
i should work on some patches in the meantime
<jemk>
rellla: :)
<oliv3r>
so much to do, so little time
<oliv3r>
oh, right, my raspbmc doesn't do h264 when streaming from hts
<oliv3r>
feck
<arokux>
oliv3r, i wonder if you have time to actually watch something :)
<oliv3r>
LOL no
<oliv3r>
but the GF does :)
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
ykchavan has quit [Read error: No route to host]
ykchavan has joined #linux-sunxi
ykchavan has quit [Quit: Leaving]
jemk has quit [Remote host closed the connection]
panda84kde has joined #linux-sunxi
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
soldoKyn has quit [Quit: Konversation terminated!]
<focus_it>
The lcd settings in the script.fex file doesn't seem to be taken
<focus_it>
Any clues what this might be down to?
<oliv3r>
very strange and unsupported LCD resolution?
<oliv3r>
the disp driver is very figity
<oliv3r>
so i wouldn't be supprised if that would be the case
arete74_ has joined #linux-sunxi
arete74_ has quit [Client Quit]
fredy has quit [Excess Flood]
<atsampson>
focus_it: the "Division by zero in kernel" bit of your log would be the place to start looking...
<atsampson>
the disp code is probably doing some maths to work out how to drive the LCD, and your parameters are breaking that
fredy has joined #linux-sunxi
<atsampson>
add some printks so you have a better idea of what it's doing in that function?
<atsampson>
hm, although I'd guess it's "lcd_fclk_frq = (info->lcd_dclk_freq * 1000*1000)/((info->lcd_vt/2) * info->lcd_ht);" in Lcd_Panel_Parameter_Check, and that'll only blow up if those params are really 0...
arete74 has quit [Quit: leaving]
<atsampson>
are you sure that your script.bin actually contains those values (bin2fex it again to see), and that the kernel's really using it?
<focus_it>
atsampson: i have those values in script.fex - but as you say best check script.bin to make sure, and also I chain from uSD to SATA when booting, and its possible, it is picking up some script.bin from uSD and if it reads it again later, its picking up a different script.bin from sata drive?
arete74 has joined #linux-sunxi
<oliv3r>
focus_it: sscript.bin is loaded into ram by u-boot
<oliv3r>
focus_it: the file should be ignored by the kernel; the kernel just checks the memory
<focus_it>
hmm... so if it loaded once, that would be it, it won't read it again from sata drive which i believe has an older script.bin
arete74 has quit [Client Quit]
arete74 has joined #linux-sunxi
<focus_it>
atsampson: i tried different values from http://linux-sunxi.org/Fex_Guide but all give same error - it does pick up the screenx_output_type setting so i know it must be reading then correct script.bin
<focus_it>
the error is like as if something did not read/initialise data structures and set lcd_x, lcd_y etc to 0 and the division by 0 starts the ball rolling with errors.
<atsampson>
focus_it: there are a couple of a20 .fex files in sunxi-boards that have 800x480 LCDs -- worth checking how they differ from yours?
<atsampson>
looks like they have lcd_vspw and lcd_hspw specified, for example
<focus_it>
ah! ok, that be good idea - i did try default settings form fex-guide, but as i say they all produce same problem.
<focus_it>
I will try suggestions later this evening.
<focus_it>
thanks
Black_Horseman has quit [Remote host closed the connection]
<atsampson>
failing that, have a poke around in the code that reads the configuration in Disp_lcdc_init
<arokux>
mnemoc, you have a bdd som what do you need more? :p
<mnemoc>
arokux: the high-res pictures are for the community
<mnemoc>
so everyone can see what's inside
<mnemoc>
and i have no use for the bdd som as I can't make carrier boards
<oliv3r>
arokux: where do we have an A20 picture
<mnemoc>
and the project on which my former employer was going to use it, died
<oliv3r>
i'll get an m5 soon, hopefully this year, because i want a nice STB
<mnemoc>
:)
<oliv3r>
rellla will have a nice xbmc image for it by that time
<oliv3r>
:)
<arokux>
oliv3r, we do not have one yet. i've asked at a lot of places - no response.
<arokux>
oliv3r, only M5 - A31 :p
<oliv3r>
ohh m5 is a31
<oliv3r>
i thought m9 is a31
<oliv3r>
yeah i want m9 pic hen :)
<mnemoc>
documentation fail
fredy has quit [Excess Flood]
<mnemoc>
rellla: 512MB :'(
<mnemoc>
err
<mnemoc>
Turl: 512MB :'(
fredy has joined #linux-sunxi
<Turl>
mnemoc: A23 + 2G would'be been nice right?
eebrah has quit [Read error: Operation timed out]
<arokux>
small home media server? :(
<arokux>
imho such small server need 2xEthernet and Wifi module in access point mode
eebrah has joined #linux-sunxi
<Turl>
that'd be a router and not a server
eebrah is now known as Guest82372
<arokux>
Turl, a router too
<Turl>
a mele under your cablemodem qualifies as server :)
<arokux>
I just like to have 1 (one) device for everything.
<ssvb>
mnemoc: if lime can beat rpi in terms of price, then 512MB is fine
<arokux>
NAS+router+WiFi network.
<arokux>
and if it indeed beats it i think our community will grow.
deasy has joined #linux-sunxi
<mnemoc>
ssvb: true
<rm>
> optional 4GB NAND Flash
<rm>
yay
<rm>
finally someone understood that NAND is useless
<rm>
btw the GK802 is classy
<rm>
2x microSD slots
<rm>
internal and external
<rm>
no NAND
<Turl>
rm: CT has an option for 2xSD too I think
<Turl>
or 1xSD and 1xtSD or some weird stuff, that's like nand exposed as an SD
<Tsvetan>
rm NAND is not useless
<Turl>
Tsvetan: it's cheap, not useless :p
<Tsvetan>
rm Android 4.2.2 runs x3 times faster from NAND vs SD-card
<Turl>
Tsvetan: what card?
<rm>
"vs a slow SD-card"
<Tsvetan>
when we have MTD we can compare Linux from NAND and SDcards ;)
<Tsvetan>
class 10 25MB card
<Turl>
Tsvetan: how did you fit android on a 25M card? :p
<Tsvetan>
25MB/s transfer :)
<Tsvetan>
NAND is x4 times faster
<Turl>
Tsvetan: maybe the nand you use is high quality
<Turl>
Tsvetan: but my china tablet nand didn't do more than a couple of MB/s
<Tsvetan>
hehehe
<Tsvetan>
with Linux we do this experiment
<Turl>
I should benchmark my olinuxino A10s nand one day
<arokux>
I have an idea in mind to take GPL'd shematics and reproduce the board using online services. I wonder what the price will be.
<Tsvetan>
install OpenCV on several Class 4 cards from different vendors it takes about 45-60 minutes
<Tsvetan>
then install OpenCV on several Class 10 cards from different manufacturers it takes about 5-10 minutes
<Tsvetan>
many file create / open / write / close operation
<arokux>
Tsvetan, is A10 actually any cheaper than A20?
<Tsvetan>
arokux a10 and a20 are same price, but Lime will be with 512MB vs 1GB on A20
<mnemoc>
arokux: the thing with lime is *support*
<Tsvetan>
and we squized the PCB size to the max
<mnemoc>
A10 is much better supported than A20
<Turl>
Tsvetan: do you aim at (sub-)rpi pricing?
<arokux>
mnemoc, ah, ok.
<Tsvetan>
Im tempted to see if A20 can work on Lime but with only 512MB I think there will be no advantage of the dual core
<Tsvetan>
Turl we try to get as close as we can
<Tsvetan>
one of the problems is that RPi and e14 screw up their distributors and give them 5-10% discount from retail price
<Tsvetan>
we cant do this
Guest82372 is now known as eebrah_
<Tsvetan>
BBB does the same
eebrah_ is now known as eebrah|FTS
<arokux>
Tsvetan, just curios. demand makes rpi so cheap or only its hardware?
<Turl>
Tsvetan: rpi hidden price was pretty big last I checked (shipping)
<Turl>
Tsvetan: also lack of case
<Tsvetan>
arokux I have no idea, ask Broadcom how much is their chip :)))
<arokux>
Tsvetan, but selling in thousands does make a product cheaper, doesn't it?
<Tsvetan>
but I guess at $35 they had no any profit when wanted to make 1000 boards
<Tsvetan>
but with 1 000 000 I guess their BOM is around $10
<atsampson>
Turl: shipping's free from CPC
<atsampson>
(to the UK, anyway, which is the target market)
<Tsvetan>
and now they make good profit but keep screwing their disti with the educational non for profit shit :)
<Tsvetan>
I saw recently video on Nicolas blog for A13 tablet which sells for $30
<arokux>
Turl, there is free shipping on amazon.de, you can get rpi for 34 EUR/40 EUR
<Tsvetan>
with 512MB RAM
<Tsvetan>
and 7 inch LCD
<Turl>
atsampson: when I was considering buying one, adding case + shipping made it 100% more expensive
<Tsvetan>
so I guess when you have volumes any pricing is possible :)))
<Tsvetan>
if you look for 7 inch LCD with touch screen you cant buy 1 pce for $30
<Tsvetan>
but you can buy A13 tablet for this price :)
<Turl>
china prices :p
<arokux>
amazing
<arokux>
Tsvetan, but you cannot buy 1 tablet :(
<atsampson>
Turl: at present, RPi model B is £28.07 +VAT, RPi model B with case is £32.80 +VAT, free delivery
<Tsvetan>
so with 1 000 000 pcs volume (if this is true)
<Tsvetan>
I guess RPi cost to manufacture less $10
<arokux>
they are making good money
<arokux>
and all the oss projects are working for them for 0.0$
<arokux>
what I miss with rpi is absence of the vision for the next rpi. guys should invest the money and manufacture a really open board.
<arokux>
rpi looks like one time project
<Turl>
they already upgraded it once
<Turl>
doubled ram I think?
<arokux>
at the very beginning
<ssvb>
arokux: you never know, and they surely are not going to announce it until you can already buy this next rpi in the shops
<arokux>
ssvb, well you knew about rpi well in advance
<Turl>
arokux: and that turned out to be chaos :p
<arokux>
ssvb, with all the fuss about community they could actually let community design the next gen rpi
<arokux>
Turl, imho lots of advertisement for them :)
<ssvb>
arokux: well, I knew about the 512mb rpi upgrade in advance (because I was in contact with the rpi people at that time), but it was a total surprise for the general public :)
<ssvb>
arokux: there was a big fuss, with people trying to cancel their pending orders to make sure that they get an upgraded board
_enrico_ has joined #linux-sunxi
arete74 has quit [Quit: leaving]
<ssvb>
arokux: I would be very surprised if their announce of the alleged next rpi board would be anything other than "you can go to these web shops and already buy it"
arete74 has joined #linux-sunxi
Cubear has joined #linux-sunxi
<rm>
the most horrible thing about R Pi
<rm>
is that it's ARMv6
<rm>
and plain Debian armhf doesn't work there
<atsampson>
I'm not sure that's the *most* horrible thing, but it's pretty horrible
<arokux>
ssvb, then I'd be surprised if all oss projects will rush to support it :(
<rm>
other than this, they can just keep making it for tens of years to come
<atsampson>
the utterly crappy USB, NAND and SD controllers are more annoying for me
<arokux>
ssvb, imho, ppls should go for olimex.
<rm>
the performance will be enough for many tasks
<arokux>
ppl*
<arokux>
rm, why keep it for many years if for the same money/power consumption you can have more perf?
<rm>
yeah, the price should come down
<rm>
e.g. for $15 shipped
docice has quit [Quit: docice]
<arokux>
yep
<rm>
but I suppose they don't plan to
<Tsvetan>
I just add picture of the box
<rm>
at some point Broadcom will make them a new SoC
<rm>
ARMv7 and maybe dualcore
<rm>
perfect for them would be if they manage to do it pin-compatible
<oliv3r>
nand is limited to 5-10 MB/s from our preliminary mtd tests
<arokux>
Tsvetan, box is cute!
<Tsvetan>
this is standard box we use for the prototypes
<oliv3r>
but i only ever tested cb1
<arokux>
rm, they won't unless a competition will force them too. they are cheapest now with whole lot of support from oss
<Tsvetan>
then when we know the openings for the connectors will make mold for plastic injection
<arokux>
so Tsvetan we believe in LIME!
Liberato1 has quit [Remote host closed the connection]
<arokux>
Tsvetan, are you actually getting something back for you GPL'd PCB schematics. I mean do ppl suggest some bug fixes or improvements?
sioxs has quit [Quit: Page closed]
<Tsvetan>
arokux actually no, I do not remember anyone to have done pull request for OLinuXino :)
<Tsvetan>
with file modification etc
<arokux>
sad.
<Tsvetan>
open hardware is not like open software
<arokux>
so what is your motivation to make PCB design GPL'd?
<Tsvetan>
giving back, teaching what you know
<Tsvetan>
lot of projects were done around our hardware designs
<Tsvetan>
what we learned doing open hardware for several years is
<Tsvetan>
that even if you open your designs if your prices are fair there is good amount of people who would not make their own board based on your files
<Tsvetan>
but they will develop more comfortable when they have access to everything in the design
<Tsvetan>
sure there will be people who will copy you
<Tsvetan>
but they will do it even if your design is closed
<arokux>
didn't know it is easily possible.
<Tsvetan>
Tom Cubie hide his designs but this doesnt stop these new Chinese companies to run same Indiegogo campain like him, so what he achieved when didnt open his CAD sources? makes developers life harder trying to trace what and where is on his PCB from his schematic :))
<arokux>
Tsvetan, you mean marsboard?
sioxs has joined #linux-sunxi
<Tsvetan>
marsboard and one new board which claim to be arduino compatible I forgot the name
<Tsvetan>
pc duino
<Tsvetan>
bla bla
<mnemoc>
Tsvetan: how hard is for an experienced EE to clone a .... cubieboard just my looking at it?
<tgaz>
and if you're using it in production: the uncertainty about whether New Hardware Company Inc will be around a year from now.
<Tsvetan>
mnemoc Allwinner have reference designs made in Allegro
<Tsvetan>
all clones including cubie are made by them
<mnemoc>
Tsvetan: so there isn't much to "protect" anyway
rz2k has joined #linux-sunxi
<arokux>
Allegro is software to design PCBs?
<Tsvetan>
yes Allegro is what left from Pads/Orcad
<Tsvetan>
cost EUR 12000 per seat or something like this
<arokux>
o_O
<Tsvetan>
probably 10 rmb in china
<Tsvetan>
:-)
<arokux>
Tsvetan, why its so expensive, does it make everything for you?
<zumbi>
I used to know how to drive OrCAD pretty well
<Tsvetan>
aroux I never used it so I cant comment :)))
<zumbi>
then Altium was gaining quite a good traction
<arokux>
Tsvetan, I bet you dream about it! :)
<Tsvetan>
but if people pay for it probably it worth
<Tsvetan>
arokux no
<zumbi>
arokux: I think the seat is valid for ever.. Altium is more appealing as seats are really cheap compare to others
<Tsvetan>
what is the reason to make open source hardware with tool which few could afford?
<zumbi>
Tsvetan: right.. open source hardware would be nice on gEDA
<Tsvetan>
Eagle non for profit licensee is EUR 100
<arokux>
Tsvetan, I'd expect every tool to export in common formats, like stl for 3D models.
<zumbi>
Eagle is no open source, iirc
<Tsvetan>
this means you get full CAD which cost EUR 1200 just for EUR 100 if you do open source hardware
<zumbi>
you can design on Altium/OrCAD and export design to gEDA :P
<zumbi>
netlists are all txt afterall
FDCX_ has quit [Remote host closed the connection]
<Tsvetan>
and was wondering what AltF2 does in windows :)
<Tsvetan>
whatever this is like which programming language is best, the experienced programmer will code in any language if necessary, every one have his preferences, but ppl are not same
<Tsvetan>
and this is good
<buZz>
no its wrong
<buZz>
everybody should like emacs
<buZz>
and nothing else!
<buZz>
;)
soldoKyn has quit [Quit: Konversation terminated!]
<buZz>
arokux: alt-f2 in winderp is mod4-r , right?
<arokux>
buZz, what is winderp?
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<arokux>
Linux needs 7-10% on desktop - it is my dream. then, I hope all driver worries will evaporate.
<buZz>
arokux: its that operating system they make in redmond
<Tsvetan>
redmond redhat whatever
<Tsvetan>
:P
sioxs has left #linux-sunxi [#linux-sunxi]
<arokux>
buZz, ok, not Win-R won't run anything... only what's in system32 if i'm not mistaken
<buZz>
arokux: yeah you need to type the full path
<buZz>
;)
atiti has quit [Read error: Operation timed out]
diego71 has quit [Ping timeout: 245 seconds]
diego71 has joined #linux-sunxi
<brain__>
questions, what are your guys who build android using for system specs? would like to know compile time also please.
<brain__>
takes me 1 hour 15 minutes on a AMD 64 bit dual core using ccache.
ZetaNeta has quit [Ping timeout: 240 seconds]
<atsampson>
my build machine's 8-core (2x E5520) -- I've not timed how long it takes from scratch
<atsampson>
takes a couple of minutes to rebuild the sun4i_crane_defconfig kernel, if that's any help?
<arokux>
atsampson, you have 2 CPU sockets?
wingrime has joined #linux-sunxi
atiti has joined #linux-sunxi
<brain__>
without ccache it takes me 4 hours to build cm9....kernel takes only about 15 minutes without ccache, have no timed it using ccache yet but once I start doing inline kernel building i'll see if its any faster. i need a better machine.
<atsampson>
arokux: yes, it's a Tyan S7002 (which at the time was about the only board I could find to buy)
wingrime_ has joined #linux-sunxi
<atsampson>
Tsvetan: that LIME design looks really cute :)
<atsampson>
can you do an A20 version with the same board?
<rm>
five drivers and/or branches and/or forks of those
<rm>
for the same device
<rm>
eebrah|FTS, works fine, I wouldn't say it's slower than a regular 'fbdev' device on the A10
<rm>
but I did not try to play any video or games
<eebrah|FTS>
rm cool
<eebrah|FTS>
I can live with that .... for now
n01 has quit [Remote host closed the connection]
<rm>
I think what I built was only for Xorg
<rm>
and it uses the kernel part that's in mainline
<rm>
<eebrah|FTS> rm: also, on a sunxi device? <- yes, first with MK802 now with the Cubieboard
<eebrah|FTS>
cool rm I have mk802 ii
<eebrah|FTS>
wonder if it would work ( well ) on an A13 device
ykchavan has quit [Read error: No route to host]
<ssvb>
rm: I find it extremely hard to believe that you can get good performance with this USB display
<ssvb>
rm: are you sure that the desktop is not laggy when using it?
<rm>
when I move windows I do not see them redraw from top to bottom
<rm>
that's good enough performance for me :D
<buZz>
i didnt see this on fbdev either (while running fluxbox and moving terminals)
<rm>
ssvb, maybe I could run gtkperf for you, or something else (packaged in Debian if possible)?
<eebrah|FTS>
I think I have drank the displaylink coolaid
<rm>
eebrah|FTS, you don't have that monitor already?
<rm>
if are you buying it, why?
<rm>
if you are buying it, why? *
<eebrah|FTS>
rm HDMI is harder to come by ... relatively
<rm>
overediting ._.
<eebrah|FTS>
and not portable
<eebrah|FTS>
I have a HDMI to VGA converter but again, not portable
<ssvb>
rm: for me it is fast enough when I can run a text editor without feeling that it is slow (basically exactly the same experience as on x86 desktop)
<ssvb>
rm: with such requirements, one can easily see that the ondemand cpu governor ruins performance
<popolon>
this driver is included in 3.4.43-sun7i+
<popolon>
at least one found on cubieboard forum
_enrico_ has quit [Quit: Bye]
<buZz>
i had a very hard time searching for sis usb devices
<ssvb>
rm: gtkperf is not very good, because it depends on the used gtk theme, but I'm getting something like "Total time: 13.55" for 50 rounds (twice better than your results) with 1280x720 resolution on cubieboard2
<rm>
it's comparable
<rm>
I tested on cb1, and the res is higher
<rm>
come to think of it, gtkperf is a really crappy benchmark
<ssvb>
yep, no disagreement about this
<rm>
could've at least had command-line switches to open in a e.g. 800x600 window
<brain__>
anyone ever see this error when booting android builds?
<brain__>
init: Unable to open persistent property directory /data/property errno: 2
ZetaNeta has joined #linux-sunxi
wingrime has joined #linux-sunxi
<wingrime>
ssvb: witch pll currently feed HDMI clock?
<montjoie[home]>
And when I use the arch bin that work, the kernel dont load...
<rm>
what's your end goal?
<wingrime>
montjoie[home]: you need uart for find whats wrong
geecko has joined #linux-sunxi
<montjoie[home]>
ok thanks I read the wiki
n01 has joined #linux-sunxi
eebrah|FTS is now known as eebrah
<wingrime>
jemk: are you there?
<jemk>
wingrime: yes
<wingrime>
jemk: I saw last patch
<wingrime>
jemk: nice
<wingrime>
jemk: but also, we suppot hi-quality resize, as you can notice
<wingrime>
jemk: kernel have FIR tables for antialiance
granite_crusher has joined #linux-sunxi
<wingrime>
jemk: so, scaler do nice work
<wingrime>
jemk: also, can you mute messages that slow down video. if you try rewind any video
<wingrime>
mnemoc: HDMI core have own version register
<wingrime>
mnemoc: 00010003 on a20
<buZz>
can cedar also mix videos?
<buZz>
like , decode 2 videos , and blend the output?
<jemk>
wingrime: are you sure these messages are reason for slowdown?
<wingrime>
jemk: on a20 yes
<wingrime>
jemk: a13 have some bug with timers, I need use profiler
<wingrime>
err
<wingrime>
on a10 yes
<wingrime>
two debug message for each frame
<jemk>
buZz: no, cedar only decodes. blending could be done by disp or g2d
<buZz>
jemk: but is it possible to decode _two_ videos at the same time?
<wingrime>
jemk: you can notice if you try rewind any video
<wingrime>
jemk: endless flood to console with mplayer
<jemk>
wingrime: i don't really believe that printing some lines is that expensive, and i don't see any performance problems here, but i'll make them once next time
rellla has joined #linux-sunxi
<wingrime>
jemk: _printig_ realy expensive if you have bash history
<wingrime>
err
<wingrime>
jemk: I don't know why, but it cause for mer
<wingrime>
*me
<wingrime>
jemk: noticeable slowdown
<jemk>
buZz: should be possible, but current kernel driver can't handle simultaneous access so userspace has to handle this
<montjoie[home]>
in ext2load mmc 0 0x43000000 boot/script.bin what does mean 0x4300000 ?
<buZz>
hmmhmm
<buZz>
guess i should play some with cedarx code
<jemk>
buZz: libvdpau should be able to do what you want once it is finished
<wingrime>
jemk: when I coded MSP430 MCU last year, I have __BIG__ slowdown when I output with sprintf
<buZz>
hmm :) that would make it more portable
<buZz>
i would _love_ to use a sunxi platform for VJ-ing
<wingrime>
buZz: ?
<buZz>
?
<wingrime>
buZz: love?
<jemk>
wingrime: msp430 is much slower than a10
<buZz>
yeah its an emotion
<buZz>
you know it?
<wingrime>
buZz: sunxi need more love
<wingrime>
buZz: for make it work
<wingrime>
jemk: a10 much slower than any Intel CPU
n01 has quit [Ping timeout: 245 seconds]
<buZz>
wingrime: i can love it longtime
<buZz>
5 dollah
<rellla>
jemk, wingrime: bunny plays fine here with vdpau, and sintel trailer without artifacts, but slowed down. so /me is ready for testing ;)
<wingrime>
rellla: try rewind video
<jemk>
rellla: uh, why slowed down. for me it played nice in full hd
<jemk>
rellla: mplayer2 fork... that isn't implemented
<rellla>
which mplayer should i use?
<rellla>
wingrime: how do i rewind via ssh-console?
<jemk>
rellla: i used the original, mplayer2 should work too, but it might stutter because of this presentation time stuff
<oliv3r>
wingrime: i found those somewhere I think (or maybe not) i don't know :)
<wingrime>
oliv3r: bad
<wingrime>
oliv3r: HDMI hardware have nice build-in colorbars
<jemk>
rellla: the whole output things are only implemented as far as it was necessary to see some decoded pictures
<rellla>
jemk: the decoded pictures are fine here! very fine.
<jemk>
rellla: this needs to be rewritten and expanded, but i still don't have an idea how to do it correctly
<rellla>
jemk: but imo it's a very good start!
<rellla>
do i have to build mplayer myself, as the wheezy one seems not to work?
<jemk>
rellla: at least for bunny and sintel, interlaced for example crashes horribly
<rellla>
jemk: yes 1080i crashed. i noticed that :p
deasy has joined #linux-sunxi
<akaizen>
Testing a new A20 HDMI stick board... fullsize usb port doesnt seem to work and OTG port powers a mouse but doesnt work
<akaizen>
Any ideas on how to fix this?
<granite_crusher>
Hello, I have a stupid question (I searched in wikis and so on...); I have A13-olinuxino; which have just 512MB of ram, as I understand memory is used by allwiner drivers for mali and this cedar thing as also g2d...
montjoie[home] has quit [Quit: Quitte]
<granite_crusher>
Do open source drivers alocates and locks memory in the similar way?
<granite_crusher>
I mean this cedarx
<oliv3r>
wingrime: but what about gmac driver for mainline? :)
<wingrime>
oliv3r: I trying get something on hdmi whith uboot
<jemk>
rellla: could you paste me a log of some seconds mplayer2 with VDPAU_TRACE=1
fredy has quit [Excess Flood]
<rellla>
jemk: one moment
<granite_crusher>
do these RE cedarx driver needs part of ram dedicated and locked from rest of system?
<oliv3r>
wingrime: well disp has colobars too? the ones you wanted to remove?
<oliv3r>
wingrime: or are these colorbars purely for hdmi
<oliv3r>
wingrime: but colorbars, closer to the output, are very important for testing :)
fredy has joined #linux-sunxi
<wingrime>
oliv3r:as I see, hdmi IP have own colorbars
<jemk>
granite_crusher: the memory is reserved by kernel, so there is no difference yet. we need this memory too.
<jemk>
rellla: thx
<rellla>
jemk: your last commit not applied yet
<jemk>
rellla: shouldn't change much, was more of cosmetic nature for vdpauinfo
<jlj_>
Has the musb usb driver been made the default for the sunxi kernels?
<jemk>
rellla: haven't seen a player actually using this information
Super-noob has joined #linux-sunxi
<jlj_>
Or is the Allwinner usb driver still the default
guilherme_ has quit [Ping timeout: 240 seconds]
tinti has quit [Read error: Connection reset by peer]
<rellla>
jemk: i think i have to build a newer mplayer, as the debian wheezy .deb seems to miss vdpau support?!
tinti has joined #linux-sunxi
<rellla>
and most likely the log-spamming causes the video to stutter or slow down...
<jemk>
rellla: possible, but mplayer2 should be fine too, it's even better to be fair, i only was to lazy to implement things it needs
<rellla>
;)
<jemk>
rellla: then try latest commit, it stops spamming
<wingrime>
Turl: ping
<jemk>
looks like i have to build a new mplayer2 too. the one from arch repo sets presentation time 429s in future?!
<ssvb>
granite_crusher: mali and g2d no not really need memory reservation, only the reservation for cedar is justified
<wingrime>
ssvb: may libvdpau use ump device for alloc?
<wingrime>
ssvb: directly
<arokux2>
Turl, are you in mood for testing? :)
<wingrime>
Turl: so, I must warn you about HDMI PLLs
<ssvb>
wingrime: we need allocation of physically contiguous memory, which means cma
<granite_crusher>
jemk, it can be freed by kernel argument, my question was: do RE cedarx needs the reserved/locked/unusabble for other aplications part of RAM as propietary needs, or does it use memory from whole - so it could be freed with kernel boot argument?
<ssvb>
wingrime: now that the dust has settled after a20 support integration, we can give it a try for real
<ssvb>
wingrime: ump by itself is just a wrapper for memory buffers to use them by mali
<jemk>
granite_crusher: we need it, but see what ssvb said
<wingrime>
Turl: accoring hsang note in hdmi_core.c , PPL settings in ccmu not worsk
<wingrime>
Turl: HDMI module have own regs for this
<granite_crusher>
thanks :)
<rellla>
jemk: sintel trailer runs out of sync if i can read the values right and mplayer remembers me, that my system is too slow to play this video. no problems with hopping bunnies
guizamboni has joined #linux-sunxi
<jemk>
rellla: yes, thats because vdpau always reports current time when mplayer asks for the time of last frame display, so it thinks it's slow
<jemk>
rellla: would try to fix that, but my mplayer2 does strange things^
ZetaNeta has quit [Quit: ZetaNeta]
<arokux2>
wingrime, you've promised to help with testing. :p what board do you have?
pacopad has quit [Quit: pacopad]
<wingrime>
arokux2: what you want to test?
<arokux2>
wingrime, usb host ehci
<wingrime>
for mainline
<wingrime>
?
<arokux2>
wingrime, yes.
<wingrime>
arokux2: I have one a13 tablet
<wingrime>
one a10 tablet
<wingrime>
cb2
<wingrime>
and olimex-micro a20
<arokux2>
wingrime, ok, cb2 and olimex-micro will do.
<arokux2>
wingrime, tablets do not have host, afaik
<wingrime>
arokux2: my a10 have
<techn_>
arokux2: a13 has also
<techn_>
one host
<techn_>
one otg
<wingrime>
arokux2: oh yes
<arokux2>
wingrime, ah, ok. but those are mini usb? do you have cables?
<wingrime>
arokux2: wifi
<wingrime>
arokux2: I have otg cables
<wingrime>
arokux2: but wifi connected internaly
<arokux2>
wingrime, so are you up to test?
<techn_>
wifi is connected to internal host
<arokux2>
techn_, true.
<wingrime>
arokux2: not currently, I have 3 mmc cards
<wingrime>
arokux2: I maybe need buy more one for mainline
<arokux2>
wingrime, set up nfs, mmc cards are slow to stick out / burn / stick in
<wingrime>
arokux2: also, I need solder uart to internals
<arokux2>
:$
<wingrime>
arokux2: we have no disp
<wingrime>
arokux2: so, I curentrly can test cb2 and olimex withour solder
<wingrime>
arokux2: a13 possible, but with pain
<techn_>
wingrime: uart on mmc slot works fine
<arokux2>
wingrime, cb2 and olimex could be fine for start. do you still need a spare mmc for those too?
<techn_>
wingrime: then nand rootfs and fel-boot :)
<wingrime>
techn_: nand boot ugly
<techn_>
wingrime: how's that?
<wingrime>
techn_: I need kill android stock
geecko has quit [Ping timeout: 260 seconds]
<techn_>
sure you can upload whole rootfs when you fel-boot.. but that's slower
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
<wingrime>
techn_: last time I tryed fel boot it not worked
<techn_>
or.. use that android rootfs
<wingrime>
arokux2: mainline have little problems for me
<arokux2>
wingrime, ok...
<wingrime>
arokux2: But cb2 and olimex micro I can test for you
<wingrime>
arokux2: but around tomorrow, I wan't not kill working systems, so I need more one mmc
<arokux2>
wingrime, cool. have you tested mainline already?
<arokux2>
wingrime, ok.
<wingrime>
arokux2: yes, I have pending IR driver for mainline
<wingrime>
arokux2: but I also can't send it until clocks not finished
<arokux2>
wingrime, perfect, so you know how to pass a dtb etc.?
<wingrime>
arokux2: yes, but I have do it again
<arokux2>
wingrime, no problem I will write very good instructions.
<wingrime>
arokux2: wiki wiki
<arokux2>
wingrime, it's there, if you know how to use them..
<arokux2>
wingrime, anyway. find a spare mmc ;)
<wingrime>
arokux2: last time it need some more love thats in manual
<wingrime>
*neded
<wingrime>
*needed
<mnemoc>
wingrime: can we use the hdmi register to distinguish a13 from a10s?
<mnemoc>
wingrime: relying in the SID sucks
<wingrime>
mnemoc: after HDMI module clocked, you can read that reg
<wingrime>
mnemoc: in case a10s without hdmi thats should read as 0x0
<wingrime>
mnemoc: *err, inverse
<wingrime>
mnemoc: a10s - reads, a13 not reads
<arokux2>
oliv3r, will you plz test usb ehci with mainline?
<oliv3r>
i will, next week :)
<oliv3r>
i really can't this week
<arokux2>
oliv3r, ok
<wingrime>
arokux2: I think we need someone with marsboar
<oliv3r>
too much to do @ $work
<wingrime>
*marsboard
<granite_crusher>
I could test it tomorow on A13-olimex
<oliv3r>
wingrime: talk more about the HDMI pll's
<arokux2>
granite_crusher, have you tested mainline already?
<granite_crusher>
still not
<granite_crusher>
but I have empty SD
<wingrime>
oliv3r: a20 man have some regs that was undocumented early
<arokux2>
granite_crusher, ok. but do yourself a favor and set up nfs. recommended.
<granite_crusher>
and earlier or later I wanted to try mainline
<wingrime>
arokux2: I thinking begin playing with disp, for basic support
<arokux2>
wingrime, mainline?
<wingrime>
arokux2: yes
<arokux2>
wingrime, what is disp?
<wingrime>
arokux2: video support
<arokux2>
wingrime, do it. i think we talk too much instead of coding :p
<oliv3r>
wingrime: i can send you 1 or 2 512 mb MMC cards :)
<wingrime>
oliv3r: no, too long
<granite_crusher>
well I dont see how NFS would work for me, because I have olimex without nand and wifi, so I guess booting from network throught usb dongle is impossible...
<arokux2>
oliv3r, lol, you'll pay more for shipping
<arokux2>
granite_crusher, no ethernet?
<granite_crusher>
on olimex A13?
<granite_crusher>
nop
<granite_crusher>
*nope
tinti has quit [Quit: Leaving]
<granite_crusher>
just throught USB dongle
<granite_crusher>
but that needs linux boot on
<granite_crusher>
or maybe I don't know something...
<arokux2>
granite_crusher, well, first you need usb support in mainline :D
<arokux2>
granite_crusher, i'm laughing, because my patches add it.
<granite_crusher>
cool. so this patch is in mainline already? or should I donwload it somethere elese and apply it?
<granite_crusher>
*download
jemk has quit [Remote host closed the connection]
atiti has joined #linux-sunxi
<wingrime>
granite_crusher: you can do uart based boot
<arokux2>
granite_crusher, but bindings are not yet added for A13, will look into it know
<wingrime>
granite_crusher: uart support X-Modem protocol for file sending
<granite_crusher>
do you mean it is possible to boot A13 throught uart? with no SD?
<granite_crusher>
wingrime: I thought files have to be loaded to RAM from SD... or from NSF or whatever, but U-boot and SPL have to be executed in SD or (nand.... which I dont have) or FEL USB?
<Turl>
howdy
<arokux2>
granite_crusher, A13 needs more work than I thought, I'll look into it later, once I know everything works.
<arokux2>
Turl, hi :)
<arokux2>
mripard, should I sent the patches to all the ppl returned by get_maintainer.pl?
atiti has quit [Ping timeout: 248 seconds]
<mripard>
yes
<oliv3r>
arokux2: nfs is not always an option :)
<arokux2>
mripard, who should be the primary recipient?
<oliv3r>
arokux2: and SD cards will probably last you very very long :) but yeah they are slower
<arokux2>
oliv3r, it is slower to burn on it. also I've noticed problems with my sd card slot
<arokux2>
mripard, you also CC ppl from allwinner, should I?
<arokux2>
mripard, shuge@allwinnertech.com and sunny@allwinnertech.com
<granite_crusher>
and SD cards sometimes gets corrupted with no (known) reason
<arokux2>
granite_crusher, this one I had too
rz2k has quit [Read error: Connection reset by peer]
<arokux2>
mripard, and who is kevin.z.m.zh@gmail.com,
<Turl>
arokux2: you needed testing I think?
<arokux2>
Turl, yes, please!
<arokux2>
Turl, but help me to sort out mainline mailing first..
<arokux2>
Turl, "ARM: sun4i: dt: Add bindings for USB host clocks"
<arokux2>
Turl, "ARM: sunxi: clk: Add support for USB Host clocks"
<arokux2>
these should be in a separate group
<arokux2>
Turl, but get_maintainer.pl returns different ppl for them. so what should I do? merge the list of ppl and and send these two patches to all of them?
<mripard>
arokux2: do whatever you want for this
<Turl>
mripard: for those two you should cc me, mripard, linux-arm and mturquette
<arokux2>
mripard, i'm just not sure what to do. do not want ppl be angry at me from the very beginning
<Turl>
arokux2: don't worry, nobody gets angry for being unnecesarily cc'ed on a patch
<arokux2>
Turl, good.
<arokux2>
Turl, can you test now?
<mripard>
arokux2: why should they be angry?
<arokux2>
mripard, no idea. they are busy.
<granite_crusher>
arokux2: I hope I can be useful tomorrow with testing A13, now my GF sad I _HAVE_ to go to sleep...
<oliv3r>
arokux2: wingrime: na shipping will be cheap; I will just tape the SD card to a post card and send that
<mripard>
arokux2: ah. for CC'ing them.
<arokux2>
granite_crusher, ok :) you GF is right. good night
granite_crusher has left #linux-sunxi [#linux-sunxi]
<mripard>
I'm not sure I would actually
<mripard>
you don't actually need to cc them, so you can leave them out of the loop if you don't want to send them your patches
<Turl>
regarding the actual patches, I'm not sure I like those _reset clocks; I'll have a look in a bit and give it a try
<arokux2>
Turl, thanks. please ping me, i'll give you some instructions
<mripard>
arokux2: well, it's precisely because it's not in mainline that you send your patches, isn't it?
<mripard>
if you mean that you depend on other patches
<mripard>
well
<arokux2>
mripard, yep, but I rely on things that are not yet in mainline, like clk
<mripard>
just say in your cover letter that you depend on those patches
<arokux2>
mripard, yes, this is what i mean.
<mripard>
and we will just merge your driver after those patches have been merged
<deasy>
dd is not ok for testing
<mripard>
basically, in your cover letter, say something like "this patch serie depends on the patches <Subject of the cover letter> from <author of the patches".
<arokux2>
mripard, the problem is those patches are not yet submitted.
<Turl>
arokux2: you're doing the test backwards
<Turl>
that has many levels of caching
<Turl>
arokux2: drop caches and do a read test
<deasy>
you can do read test with hdparm
<deasy>
-t or -T
<Turl>
mripard: the main problem I think is that they're not posted yet :) but it should be enough to mention this is the case
<mripard>
arokux2: well, then, either put the patches as part of your serie, or slap Turl
<mripard>
you're not really comparing the same thing, not even the same controller
<arokux2>
mripard, you mean on HDD side?
<Turl>
mripard: we're using dd, that should give you an idea of the seriousness of these comparisons :)
<arokux2>
Turl, i thing there is no major issue, if any with speed. so it is fine.
<arokux2>
Turl, I fail to figure out how to test keyboard or mouse, though. oliv3r said it won't work but it mouse seems to do smth. and i do not have usb keyboard only ps/2.
Tartarus__ has quit [Quit: Ex-Chat]
wingrime has quit [Ping timeout: 248 seconds]
<Turl>
arokux2: same, no usb kbrd here
<Turl>
you can probably make an app that reads input events and prints them to test mice
<arokux2>
Turl, i'm too lazy. mouse is useless for headless anyway. as for keyboard - i do not have any. so somebody will need to help with testing this one.