<KotCzarny>
but if you expect it to idle most of the time it wouldnt be bad to switch to ondemand/interactive governor
<[TheBug]>
Also you do have at least a heatsink right?
<sunshavi>
no heatsink from always
<KotCzarny>
governor "performance"
<[TheBug]>
ohh sorry missed you answering that before
<[TheBug]>
but yeah at least a small heatsink would be probably a help
<sunshavi>
TheBug: no problem
<KotCzarny>
its good if the box is meant to be always in work
<KotCzarny>
but if it's going to idle, better to switch to ondemand (even if you lose a bit of performance)
<sunshavi>
well. the box has worked really fine it is almost always on
<mru>
an h3 sitting on my desk, naked board w/o heatsink, constantly at 1GHz is idling at about 45°C
<KotCzarny>
it's not about 'always on', it's about 'always busy' or 'sometimes busy'
<KotCzarny>
mru: check with 1.2
<mru>
the kernel doesn't offer that speed
<sunshavi>
KotCzarny: them. I am going to play with it. and changing the governor manually according to what I am doing
<KotCzarny>
mru: hack it
<mru>
why would I do that?
<KotCzarny>
mru: for science?
<[TheBug]>
sunshavi: just gives you an idea, I have 10+ H3s and every single one has 30mmx30mm heatsink and a 30mmx30mm 5v fan on top. stays 35-48c max in general on all boards under full load.
<mru>
this is a commercial product
<KotCzarny>
mru: is it used commercially without any enclosure?
<mru>
of course not
<sunshavi>
take in account that opipc without case goes 1.3 without any issue
<KotCzarny>
so its not a com. product, but a dev one
<mru>
same software
<mru>
that's the point
<sunshavi>
also without heatsink
<sunshavi>
TheBug: nice arrangement
<[TheBug]>
I have improved a lot of them since that pic actually but gives you an idea of how I had heatsink and fan at least
<[TheBug]>
sunshavi: easiest application is using thermal glue to adhere the heatsink so you don't have to use some type of downward force on it and then you can attach fan right on top with m3 nylon screws
<[TheBug]>
you can get thermal glue for cheap like ~$1 from china, usually in white tube
<[TheBug]>
And when our power combine, we become Captain SafeZone
<KotCzarny>
cpt. friendzone
<[TheBug]>
LOL
<[TheBug]>
Thats a good one, actually made me LOL :D
<[TheBug]>
KotCzarny: would suck if you were a teenager and your girlfriend was a big fan of Captain Snowflake -- Damn that Captain Friendzone, I will never love my virginity!
<[TheBug]>
s/love/lose*
<[TheBug]>
but I digress.
\\Mr_C\\ has quit [Quit: (Read error: Connection reset by beer)]
<sunshavi>
Trailblazer seems and AMD processor
Nemo_bis has quit [Remote host closed the connection]
jstein has joined #linux-sunxi
RichardG867_ has joined #linux-sunxi
RichardG867_ has quit [Changing host]
RichardG867_ has joined #linux-sunxi
RichardG867 has quit [Ping timeout: 272 seconds]
tuxd3v has joined #linux-sunxi
netlynx has quit [Ping timeout: 256 seconds]
netlynx has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
macc24 has joined #linux-sunxi
<willmore>
Oh, you'll love it.
<sunshavi>
:)
<sunshavi>
60 degress with ondemand
macc24 has quit [Ping timeout: 265 seconds]
Nemo_bis has joined #linux-sunxi
macc24 has joined #linux-sunxi
tuxd3v has quit [Quit: Leaving]
xes has joined #linux-sunxi
<mru>
sunshavi: btw, last night you mentioned a weird compiler error
<sunshavi>
but on my defense i need to say compiling same source is quickier on patassd
<sunshavi>
source code
<sunshavi>
how that is possible?
<mru>
writes might be faster on the ssd
<mru>
or it can handle more transactions per second
<sunshavi>
alos opipc on hdd boot faster than uSD
<mru>
matters for small reads/writes
<sunshavi>
boot on hdd is faster on pata ssd also
<mru>
sd cards are slow
<sunshavi>
half the time
<sunshavi>
@least
<mru>
rest of the time they're dead
<sunshavi>
then on production. pata-ssd is faster than emmc. On test things seems to be different
<mru>
sequential read speed is only one of several parameters
<sunshavi>
that's it mru numbers are not matching reality
<mru>
those read speeds are probably accurate
<mru>
but as I said, it's likely different for writes and small, random accesses
<sunshavi>
i got this one for my thinkpad x30
<mru>
ssd specs usually quote 4 numbers: sequential read/write speed and number of read/write operations per seconds
<sunshavi>
so. I just wanted to test it on opipc(sunxi-fel). and I got very surprised with the performance i got from it
<sunshavi>
sata is not plug and play. I have a sata disk also. But it need to be plug and play
<mru>
huh?
<sunshavi>
i use same disk on opipc (as main hdd). and on opipc as chroot
<mru>
with a usb adapter, it all looks the same
<sunshavi>
for sata i need to plug to cables
<sunshavi>
s/to/two/
<sunshavi>
I would be nice testing pata-ssd vs sata
<sunshavi>
s/sata/sata-sdd/
<willmore>
mru, I used some PATA SSDs back in the days of thin clients. I took the boxes out of service a while back when their power supplies started going flakey. So, maybe they were from the early 2Ks or late 1990's.
<mru>
willmore: SSDs were certainly not a consumer thing back then
<sunshavi>
mru: Yes. that one is nice.
<sunshavi>
s/nice/fine/
gaston1980 has quit [Ping timeout: 260 seconds]
<mru>
there are many similar products
<mru>
I have a nameless one I got from amazon
<sunshavi>
BTW. i first tried a normal hdd. But I think it needed more power for working
gaston1980 has joined #linux-sunxi
<mru>
yeah, even a 2.5" hdd needs more power than a usb port can (or should) supply
<mru>
at least while it spins up
<sunshavi>
Yes. It should be handy having one of them
<mru>
can't say I use it much though
<sunshavi>
actually with those speeds I am getting. I would need to book some time for trying BSD
<sunshavi>
But. yesterday's compiler error. Should be cos of long cable. When the device is used with uSD there are not this problems. Before having the hard disk. I used to do on memory compilations on /tmp on this device
<mru>
I doubt the cable is to blame
<sunshavi>
mmmm?
<mru>
usb error rate is pretty low
<mru>
and there are checksums to catch any that do occur
<mru>
ram bitflips would go undetected
<sunshavi>
them. ur theory is memory issue?
<sunshavi>
s/them/then/
<mru>
something corrupted your data, that's obvious
warpme_ has joined #linux-sunxi
<sunshavi>
But file on hdd was fine
<mru>
yes, so it was mangled somewhere between the drive and the cpu
<sunshavi>
right
<mru>
and ram is the only place that doesn't have checksums
<sunshavi>
then it could be
<sunshavi>
but also when uncompressing kernel sometimes I am getting ^@^A on source code files
<sunshavi>
compilation fails and I fix source code manually
<mru>
could be ram acting up there too
matthias_bgg has quit [Quit: Leaving]
Putti has joined #linux-sunxi
<sunshavi>
i am going to try to compile mesa on chroot on opi+2e and see how it goes. Last time I did on opipc it failed with the abeve mentioned characters
<sunshavi>
perhaps too few people is trying opipc with hdd attached. And nobody has noticed it
<mru>
or yours is faulty
<mru>
bad ram happens
<sunshavi>
You are right
<mru>
heat makes it worse
<sunshavi>
I bought. this second h3 device for unloading jobs from opipc
<sunshavi>
but opipc goes 1.3 without hiccups (no case)
<sunshavi>
and opi+2e (with case). 1.3 hangs the machine
<sunshavi>
i am going to check temperature on opipc
<sunshavi>
on opipc is 42
lurchi__ is now known as lurchi_
<sunshavi>
but I would need to check in the middle of kernel compilation
<mru>
I was going to say that
<mru>
it'll be much higher
<sunshavi>
yes. Sure it is going to be higher
<sunshavi>
I have just enabled sun8i_thermal on my kernel. That is just a sensor. Or it helps on throtling the cores
<sunshavi>
?
<mru>
it will if you have those options enabled and the necessary dt bits
<mru>
you should see some files with trip_point this and that
qschulz has joined #linux-sunxi
<sunshavi>
mmm. i am going to check it again then
<sunshavi>
on sun8i-h3.dtsi there is just one include and not two like in the patch
Putti has joined #linux-sunxi
Putti has quit [Changing host]
<sunshavi>
that commit has been applied?
<mru>
it's in 5.7, not 5.6
<mru>
you should apply the patch to your kernel
<willmore>
mru, indeed. The ones I had were 4 to 16 MB. Enough for a kernel and maybe a small initrd. Everything else came over the network. Well, some of them had other storage--USB particularly. USB 1.2, so not exactly blazing speeds.
<mru>
they had some sunray thin clients at my university around that time
<mru>
I guess they would have had some simple storage like that