[Bug 1547528] Re: How to handle specific HW needs for dep8 tests in CI

[ChristianEhrhardt]

Sharing a few good statements from a chat with infinity with all participants on this bug
And from me thanks Adam for these insights to part of Ubuntus build/test history.

Quoting Infinity:
"We'd want scalingstack buildds and autopkgtest to match if we're changing machine type.  Having builders and testers mismatched would be super confusing.
[...]
Well, we're already failing if builders/testers are meant to actually be our base ISA
PPro for i386 that is.  For amd64, it's a bit muddier, you'd need to go find the exact specs of the first Athlon64, which is what the base ISA *should* be.
(Minus 3dNow!, since Intel never supported it).
I think that boils down to i686 + mmx + sse + sse2 + longmode, ish.

So, it comes down to sorting out what we think is reasonable.
Given we always used to build on bare metal (again, on Sandy+ machines), it's not like we've ever tried to keep the builds confined to a baseline at build time.

Anyhow, as I said, we've never stuck to this well, as we only started building on all-virt very recently.
So, I'd be fine with setting a modern base (say, SandyBridge, or just '-cpu host' even) for builders and testers, and say that packages and toolchains are responsible for not overstepping ISA bounds, as we've always done."


** Changed in: dpdk (Ubuntu)
       Status: New => Triaged

** Changed in: dpdk (Ubuntu)
       Status: Triaged => Incomplete

** Changed in: dpdk (Ubuntu)
   Importance: Undecided => Medium

-- 
You received this bug notification because you are a member of Ubuntu
Server Team, which is subscribed to dpdk in Ubuntu.
https://bugs.launchpad.net/bugs/1547528

Title:
  How to handle specific HW needs for dep8 tests in CI

To manage notifications about this bug go to:
https://bugs.launchpad.net/auto-package-testing/+bug/1547528/+subscriptions