iooi has quit [Read error: Connection reset by peer]
iooi has joined #fedora-riscv
davidlt has joined #fedora-riscv
jcajka has joined #fedora-riscv
<rwmjones_>
morning
rwmjones_ is now known as rwmjone
rwmjone is now known as rwmjones
<davidlt[m]>
morning
davidlt has quit [Ping timeout: 252 seconds]
davidlt has joined #fedora-riscv
<davidlt[m]>
rwmjones: experiment, another 10 boards are online (only during the day [fan replacements still to happen]).
<davidlt[m]>
So far peak bandwidth was ~44MB/s (<400Mbps)
<davidlt[m]>
But close to 4GB of data have been moved via a switch pretty much within minutes.
<davidlt[m]>
This is for SRPM building
<davidlt[m]>
I see my uplink port utilization spiked to 70%, that probably means a jump to 700Mbps (?)
<davidlt[m]>
It's switching to buildArch now.
<davidlt[m]>
Almost 5G of data moved
<davidlt[m]>
Uplink is trying to stabilize at <30% utilization
<davidlt[m]>
5.5G
leah2 has quit [Ping timeout: 244 seconds]
leah2 has joined #fedora-riscv
leah2 has quit [Remote host closed the connection]
leah2 has joined #fedora-riscv
leah2 has quit [Ping timeout: 264 seconds]
leah2 has joined #fedora-riscv
<davidlt[m]>
rwmjones: jive.home.annexia.org is dead
<davidlt[m]>
No pings since yesterday
<davidlt[m]>
Uh, new peak 63.2MB/s (~500Mbps)
<davidlt[m]>
10+G an hours so far transferred
<davidlt[m]>
s/hours/hour/
<davidlt[m]>
The noise is terrible, that I can confirm too. I guess, vibrations cause most of that.
<Esmil>
davidlt[m]: is that 10 unmatched boards, because that noise really would be terrible :)
<davidlt[m]>
Esmil: yes (and there will be more)
<davidlt[m]>
Fan replacements will come later
<davidlt[m]>
I am really surprised how that tiny crap could even vibrate my table
<Esmil>
ah, nice. yeah, the first thing I did when I got my unmatched was to get the chassis fan working, but unfortunately that wasn't enough to keep the cpu temp down
<davidlt[m]>
I am probably lucky. My main board can work 40-60 minutes on heavy load without a fan before thermals kill it.
<davidlt[m]>
In a closed PC case a slow fan is enough to keep it running.
<davidlt[m]>
rwmjones had troubles keeping his boards working properly during heat waves in UK this summer.
zsun has joined #fedora-riscv
<rwmjones>
davidlt[m]: looking ..
<rwmjones>
davidlt[m]: on nujive, kojid is still running but there are no jobs
<rwmjones>
I'll restart it, one sec
<rwmjones>
.. done
<davidlt[m]>
probably it's kojid crapping out :)
<davidlt[m]>
I hate that it does it, it's kinda random, and there could be multiple reasons for it
<davidlt[m]>
it has no jobs, but again not talking to Koji for the last 2 hours
<somlo>
davidlt: building the latest linux on my 8-core, 32-GB qemu VM (with `make olddefconfig` against the standard fedora kernel's config file
<somlo>
been at it for 14 hours, still cooking :)
<davidlt[m]>
somlo: nice!
<davidlt[m]>
So the board is faster ;)
<somlo>
but once it's done, I'm hoping I can just make small changes to the uart driver and rebuild, and it should take much less time :)
<davidlt[m]>
Yup
<davidlt[m]>
well it depends
<somlo>
I know, if I change debug flags like e.g PROVE_LOCKING I will have to rebuild the whole thing
<somlo>
but just small modifications to a single driver should be cheap
<somlo>
one of these days I'll have to ask for a recommendation on a good hardware board to buy, but for qemu it is (and it's (much?) faster than my fpga, even the one running at 100MHz :)
<somlo>
but it's really not about the speed for me, it's about the self-hosting :)
<davidlt[m]>
Cheap: JH7110
<davidlt[m]>
Otherwise SiFive/Intel Horse Creek
zsun has quit [Quit: Leaving.]
<davidlt[m]>
On native development you want speed :)
<somlo>
davidlt[m]: are these available for ordering anywhere?
<davidlt[m]>
No, nothing is :)
<somlo>
I'm likely to try to guilt my boss into ordering one for me, so straightforward order forms would help (as opposed to kickstarter campaigns and such like :)
<davidlt[m]>
No yet, but at soonish (?) // weeks, well more months away
<somlo>
well, there you go, it's qemu for the likes of me :)
<davidlt[m]>
RISC-V Summit is soon :)
<somlo>
I should get out more, I know :)
leah2 has quit [Ping timeout: 260 seconds]
leah2 has joined #fedora-riscv
<rwmjones>
davidlt[m]: again?!
<rwmjones>
checking ..
jcajka has quit [Quit: Leaving]
<rwmjones>
davidlt[m]: it looks completely fine
<rwmjones>
there are no jobs running but I see no errors or anything from kojid
<rwmjones>
I can restart kojid if you like
<davidlt[m]>
restart
<davidlt[m]>
is the network on? :)
<rwmjones>
I hope so, I'm accessing it over the network too :-)
<rwmjones>
shall I restart kojid and then you can try sending a job straight after?
<rwmjones>
ok I restarted kojid, can you send something?
<davidlt[m]>
it's on and talking to Koji
<rwmjones>
it's weird that it keeps losing the connection, afaik nothing is happening here
<rwmjones>
& nufive is on the same network, so you'd think if something was affecting one it would affect the other
iooi has quit [Read error: Connection reset by peer]
iooi has joined #fedora-riscv
iooi has quit [Read error: Connection reset by peer]
iooi has joined #fedora-riscv
iooi has quit [Read error: Connection reset by peer]
iooi has joined #fedora-riscv
skip77 has quit [Ping timeout: 264 seconds]
davidlt has quit [Ping timeout: 246 seconds]
<djdelorie>
davidlt[m]: if I add a second robo-builder to my unmatched, will things go badly when two big builds happen at once?