IRC channel logs

<damo22>there is no reason large language models need to be so big, i dont know why people are boiling the oceans just to train the biggest model

<homo>to make free software look good, as it is not so resource hungry

<damo22>Pellescours: ping, can we figure out the missing symbol for rumpusbdisk? then we could ask y oupi to merge the new rump

<gfleury>damo22: hi i have tested smp with flavio's script. 1) i build gnumach with --enable-ncpus=2 --disable-linux-groups 2) run qemu with -smp 2. For the first run it stuck at irq 10. Then i change to qemu -smp 3 it reboot constaly with is espected. I kill it and rerun with -smp 2 and this this it boot and i can access to

<gfleury>Internet i do ping 1.1.1.1 it works

<gfleury>nproc gives 2

<damo22>i usually compile with --enable-cpus=8 so it can be run with any -smp up to 8

<damo22>there is still a race condition at boot i think with smp 2

<damo22>we need to fix pci, acpi, rumpdisk starting up

<damo22>i can take a look at this now

<gfleury>Will check with -smp 3 since my Host has 4 core

<damo22>you can use -smp 4 just dont stress it

<gfleury>Okay

<damo22>but you will need to recompile gnumach with more cpus

<gfleury>Yes i know

<Pellescours>damo22: pong, yes i need to reckmheck but it's a dma... function

<paulz>Hello at All

<paulz>I hope you had good holidays during this "Noël"

<Pellescours>damo22: maybe this commit can be usefull https://github.com/NetBSD/src/commit/7584142d813548eb76f151a26c5adfbf66b659f4

<Pellescours>do we use wg drivers for network ? if yes I can re-update the rumpkernel. I no, we can keep the current update

<homo>wg is a wireguard vpn, it is not tied to any hardware

<gfleury>damo22: smp 2 work well when using qemu serial and console=com0

<gfleury>sneek later tell damo22: that smp 1 and 2 on console=com0. perhaps the race could be in VGA driver in device dir

<sneek>Got it.

IRC channel logs

2024-12-30.log