> The bug doesn't really explain why it's fine to remove all the listed
packages on ppc64el and what's the plan to bring those back if needed

The ppc64el ROCm packages that I've requested for removal have never
been shown to work on any hardware. I am the developer that fixed most
of the compilation errors that were blocking the ROCm stack from
building on ppc64el. I did that because Timothy Pearson of Raptor
Computing had enabled the ROCm driver support on ppc64el. The removal of
x86-isms was sufficient to get much of ROCm working on arm64, but
Timothy and I never found a ppc64el system that worked with ROCm.

There's no plan to bring these packages back on ppc64el. They have zero
users, zero upstream support, and zero passing tests. If the compiler
bugs causing FTBFS are fixed this week, then I might reconsider my plan
to drop the ppc64el architecture from these packages... but I'm not
expecting that to be the case.

> Also a removal request usually comes with a reverse-depends analysis
of the packages to ensure they don't have other things in the archive
needed them

What should the analysis look like? The original bug description
requests the removal of all reverse dependencies of rocblas and rocfft.
I believe I included all reverse-depends as packages to be removed in
the request.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2134241

Title:
   Please RM ppc64el binaries

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/ggml/+bug/2134241/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to