<davidlt[m]>
Conan Kudo: I am slowly providing access to the boards I run, and I would prefer some form of integration with FAS at some point (that would be easiest for the users). I don't run boards 24/7, which is annoying and don't have right now free resources during these mass buildings.
<davidlt[m]>
Oleg Girko: I doubt it will be affordable like RPi. It will be using Intel 4 (7nm) process from IFS. That doesn't sound cheap.
<davidlt[m]>
Not to mention there is DDR5 and PCIe Gen 5 on it :)
<davidlt[m]>
The sad part is P550, which is old by now, but the 1st OoO core.
<davidlt[m]>
Not compliant to RVA22 from not yet ratified spec.
<davidlt[m]>
In general only two companies so far announced RVA22 profile compatible core IP. That's SiFive with P670 (minor update from P650) and Alibaba with C908 (C906 is used in Allwinner D1).
<davidlt[m]>
So that probably means that 1, 2 or even 3 years away from hardware compliant with a profile.
<davidlt[m]>
Note, that RVA23 is the next major one, which will be a radical shift. I consider that ARMv8 -> ARMv9 situation.
<davidlt[m]>
We might need to discuss what to do here. That is should we redefine riscv64 == RVA22, or RVA23.
<davidlt[m]>
While technically profiles should happen every ~two years, RVA23 should happen in early 2023. RVA22 is kinda a stop gap solution, and that's why it's not considered a "major" profile.
<davidlt[m]>
RVA23 requires vector, vector crypto, support for larger pages than 4K, and something else probably too.
<davidlt[m]>
RVA23 definition technically doesn't exist yet. Only in the slides.
guerby has quit [*.net *.split]
Ferix has quit [*.net *.split]
rwmjones|hols has quit [*.net *.split]
Ferix has joined #fedora-riscv
rwmjones has joined #fedora-riscv
guerby has joined #fedora-riscv
xen0n has quit [*.net *.split]
Esmil[m] has quit [*.net *.split]
nirik[m] has quit [*.net *.split]
defolos has quit [*.net *.split]
xen0n has joined #fedora-riscv
Esmil[m] has joined #fedora-riscv
defolos has joined #fedora-riscv
nirik[m] has joined #fedora-riscv
<davidlt[m]>
Last 225 packages from the the 1st mass F37 sync left to process.
<davidlt[m]>
So two weeks for this mass run.
<davidlt[m]>
The 2nd mass sync should be way smaller, as I don't think any new packages were allowed to tagged before F37 is released now.
davidlt has quit [Ping timeout: 240 seconds]
Kevinsadminaccou has quit [Quit: You have been kicked for being idle]
ahs3[m] has quit [Read error: Software caused connection abort]
ahs3[m] has joined #fedora-riscv
davidlt has joined #fedora-riscv
<davidlt[m]>
9 packages just left! This basically means the 1st wave is done.
<davidlt[m]>
Will sync F37 today again to see what delta we have between upstream and downstream Koji instances.
<ol>
BTW, why https://riscv.rocks/ has no proper certificate? Could you use something like LetsEncrypt please?
zsun has joined #fedora-riscv
<davidlt[m]>
Yeah. IIRC this is under self-signed certificate, which used used to authenticate users (all is based on certificates).
<ol>
So, why not just use LetsEncrypt's free certificate?
<davidlt[m]>
Koji CA certificate was created a long time ago (based on Koj instructions). Replacing that means we would need to re-generated everything and replace certificates on all machines, and users.
<davidlt[m]>
If out auth layer would be based on Kerberos using LetsEncrypt TLS certificate would be easier.
<davidlt[m]>
Also I am not sure if LetsEncrypt certificate could replace Koji CA certificate. I recall that we had to do a longish config to generate it, not sure what LetsEncrypt allows to configure.
<davidlt[m]>
We are also not signing any RPMs too right now.
<ol>
davidlt[m]: NOOOOO!!!!!111 How can I install any packages if I can't ensure any integrity on neither transport nor data level?
<davidlt[m]>
Well, we cannot afford to run full Fedora infra yet. We are running bare minimum. Things slowly will improve in the future, but step by step.
<davidlt[m]>
With nirik and neil (i.e. more human hours, more eyes, more fingers to use the keyboard and more brain power) there might be improvements.
zsun has quit [Quit: Leaving.]
davidlt has quit [Ping timeout: 268 seconds]
<neil>
myself and Sokel were chatting about that a few weeks back. I'm home from my honeymoon now and though I'm traveling a bit, my work schedule is mostly freed up to get back to poking on Riscv :)
Kevinsadminaccou has joined #fedora-riscv
<neil>
sigul would probably be a pain, and it all comes down to what you're using this for, ol... afaik, it's not as if there are datacenters full of RiscV that I'm worried about transport and other security for. This is all very very early days and should be treated as such. It's not even a secondary, let alone Primary Fedora arch yet