<Necrosporus>
I mean under 100% load it reaches 85°C
<Necrosporus>
I create high load by running math-computing processes in bg
stikonas has joined #linux-rockchip
<dirkson>
stikonas: Thanks! That's actually what I'm trying to replace. Broke after 12 hours and pine won't RMA it. And that's not even the worst experience I've had with them. You've got it running an unmodified kernel? Didn't work at all for me, even with fairly recent kernels. Although now that I think about it, maybe that attempt was before I got the uart, so maybe there was something simple on
<dirkson>
serial console I missed.
<stikonas>
dirkson: yeah, I'm running completely unpatched kernel and u-boot
<stikonas>
there is a problem with some emmc cards
<stikonas>
but I'm not using emmc
<stikonas>
but uart cable is definitely needed (or at least another working board with GPIO headers :)
<dirkson>
stikonas: Hah, that'd explain it - I was trying to boot off emmc basically the whole time.
<dirkson>
A bit academic now I suppose, but good to know what I did wrong.
<stikonas>
dirkson: emmc's sold by pine need a kernel patch that disables emmc command queues
<stikonas>
or something like that...
<stikonas>
non-pine emmc actually work. I once had old emmc from hardkernel and it worked although after some point that emmc died (presumably from old age)
<dirkson>
stikonas: Now if you can just fix pine as a company, I'll be good to go!
<stikonas>
well, I guess it depends whether you have good or bad experience with them... It's not always bad
<stikonas>
I think rockpi also would be able to boot from mainline
<stikonas>
dirkson: I actually got NVMe for my rockpro64 instead of emmc because it is cheaper
<stikonas>
and NVMe is definitely faster
<dirkson>
stikonas: I was trying to make mine into a router, so I was using the pcie for a gigabit nic.
<stikonas>
dirkson: yeah, in that case you need emmc. I have some longer term plans to do that too...
<stikonas>
would also need to find some nic that works without blobs...
<dirkson>
stikonas: I did! Had it running as the router before it fell over and died. It was a glorious 20 minutes.
<stikonas>
which nic did you use?
<stikonas>
something with 4 ports?
<dirkson>
stikonas: intel I350t2 clone. Just 2 ports - Fed it directly into my network switch after that.
<stikonas>
yeah, I guess it's acceptable too
<dirkson>
In related news, I guess I no longer have a use for that nic, given the rockpro64 is the only open-ish computer I'm aware of that has a PCIE slot.
<stikonas>
well, you can try your luck again
<stikonas>
now that you have serial cable...
<stikonas>
actually, my first rockpro64 broke after a year or so and I bought another one
<dirkson>
Oh, no, the hardware is dead. 100% dead. And no RMA for me.
<stikonas>
this one seems to work fine
<dirkson>
Like no output on serial console no matter what I do dead.
<stikonas>
overheated?
<stikonas>
but strange that they didn't do RMA...
<dirkson>
Unclear. I had an active cooler on it, so I'm not sure what else I could do.
<stikonas>
it shouldn't overheat if there is active cooler
<stikonas>
(and spinning)
<stikonas>
if it's 4 GB board, they seem to be out of stock
<stikonas>
maybe that's why they didn't do RMA and tried to find some "reason"...
<dirkson>
stikonas: I waited like a month to actually set the thing up. 30 day warranty. Also they haven't actually said 'no', they just stopped responding to my tickets. And given my other issues with them, I'm fed up and intend to do two chargebacks.
<dirkson>
stikonas: I wouldn't think it would overheat with an active, but there are chips other than the rk3399 and memory on the board. I could see something else overheating. My original guess was some cold solder joint, but the issues seem to persist after it comes back to room temp.
ldevulder_ has joined #linux-rockchip
ldevulder has quit [Ping timeout: 260 seconds]
lopsided98 has quit [Quit: No Ping reply in 180 seconds.]
<robmur01>
dirkson: if you want RK3399 as a dual-NIC router, consider the NanoPi R4S that's just come out
<robmur01>
although maybe only the 1GB DDR3 variant, since DDR4 seems to be no end of bother with upstream U-Boot
<robmur01>
(by all accounts the DDR4-equipped NanoPi M4v2 is barely usable, whereas its DDR3 cousins like my NanoPC-T4 have been running rock-solid with the full mainline stack for years)
<stikonas>
but it only has sdcard?
<stikonas>
although, it's nice that it has two NICs...
<robmur01>
true, but USB3 for extra storage isn't *too* awful - main thing is it already has the PCIe NIC integrated
<robmur01>
the other end of the scale would probably be the Helios64 with 2.5GbE as a second interface plus lots and lots of storage
<robmur01>
not quite so "open" in terms of available documentation, but they are at least trying very hard to support a mainline-based kernel for it
<dirkson>
robmur01: I was looking at that one. It wasn't clear to me that it worked with a mainline kernel.
<dirkson>
'that one' being the nanopi R4S. I haven't looked into the helios yet
<robmur01>
I haven't seen a mainline DT for it yet, but that shouldn't be too much work - all the other FriendlyELEC RK3399 boards don't deviate much from the Rockchip reference design (or each other)
<dirkson>
Helios looks interesting, but it seem fairly expensive for its specs, and isn't currently available?
<robmur01>
I'd be very surprised if it isn't fundamentally just another "nanopi4"
<dirkson>
Probably part of the expense is the 2.5gbe, which I don't know that I have a usage case for in the next... decade?
<robmur01>
yeah, the Helios doesn't really fit into "hobbyist SBC" considerations, but I think it's fun that it's a thing at all :D
<dirkson>
Fair!
<dirkson>
I do like the idea of the nanopi r4s, but 1db ram seems like so little. So does 4 for that matter.
<robmur01>
other than that, there are a fair few RK3399 boards expandable via M.2 sockets, but nothing else with a full-size PCIe slot that I can think of
* tuxd3v
tuxd3v owns a Helios4..
<robmur01>
I don't believe RK3399 officially supports more than 4GB
<robmur01>
thankfully, the next-gen RK35xx lineup seems to be finally breaking that barrier...
<dirkson>
robmur01: I did work out I could grab a ROC-RK3399-PC, a mezzanine, and an m.2 ethernet card. That seems a little complicated, though.
<dirkson>
robmur01: Is that available in anything yet?
<tuxd3v>
robmur01, yeah but that rk3588 never comes out :/
<dirkson>
ah, delayed.
<tuxd3v>
dirkson, delayed by how much?
<robmur01>
apparently the mid-range 3568 is now sampling, and they're clearly going hard at it with BSP patches, so I'm hoping the first wave of Android boxes might start appearing early-ish next year
<dirkson>
tuxd3v: I saw articles referencing the RK3588 as coming out in q1 2020. And later articles saying q1 2021
<tuxd3v>
I am waiting for it for 1 year now :D
<robmur01>
most recent I've seen says Q3 2021, but it least it's getting a GPU upgrade out of it...
<dirkson>
Buuut it's on 8nm, which is what I think nvidia is on? And they can't meet their market demands, so I suspect smaller companies will have a hard time buying up fab time?
* dirkson
has been trying to buy a gpu! It has... not been going well.
* tuxd3v
tuxd3v has 3 to sell..
<dirkson>
Tuxd3v sus.
<tuxd3v>
:D
<tuxd3v>
Well, I am being testing nanopi r2s and sometimes the lan port hangs :/
<tuxd3v>
what could I be doing wrong :/
<robmur01>
running Linux - nothing good ever comes of that :P
<dirkson>
robmur01: *looks around at his household running nothing but linux* ... accurate.
t3st3r has quit [Remote host closed the connection]
* tuxd3v
agrees with dirkson :)
t3st3r has joined #linux-rockchip
<tuxd3v>
I tried the rockchip usb 3.0 driver
<tuxd3v>
for nanopi r2s lan port
<tuxd3v>
it works , but also hangs sometimes
<tuxd3v>
then I discovered Peter Geis patch in patchwork.kernel.org, for usb3.0, plus a dwc3 glue layer
<tuxd3v>
I was trying that, but it also hangs :/
<tuxd3v>
I get periods of 120-180 seconds were the lan port doesn't send or receive traffic
<tuxd3v>
this ofcourse tested with iperf3
<tuxd3v>
if the trafic is low it manage to work
<tuxd3v>
I was just conducting stress tests on lan port and discivered this..
<tuxd3v>
I don't know if openwrt images from nanppi also have this problems..
<robmur01>
TBH I'd be just as suspicious of the Realtek driver
<tuxd3v>
err, discivered -> discovered
<tuxd3v>
it could be
<tuxd3v>
the driver is the r8152
mps has left #linux-rockchip [#linux-rockchip]
mraynal has quit [Remote host closed the connection]