Re: Running up that (HIP) hill
On 2/16/22 16:43, M. Zhou wrote:
Hi,
On Wed, 2022-02-16 at 16:30 +0100, Maxime Chambonnet wrote:
- imported in a MUT (vendoring other github repos) way
gbp import-orig --component=clr --component=opencl \
--component=hip --pristine-tar --uscan
TBH I very much dislike squashing multiple sources into a single one.
They aren't tightly coupled modules, right?
What's the motivation for multi source?
I was on your side two months ago, and then I gave up. [1]
I made basically no progress in understanding HIP past two months.
One would need quite a "bare-metal" understanding of the program
to un-bundle the 4 components. Gentoo devs are still getting mad about
it on rocclr 4.5.2. I do not mean that it is impossible!
There is at least a circular cmake dependency between opencl [2] and
rocclr [3], both of which I started packaging (MUT too...) aside.
[1] https://lists.debian.org/debian-ai/2021/12/msg00026.html
[2] https://salsa.debian.org/rocm-team/rocm-opencl-runtime
[3] https://salsa.debian.org/rocm-team/rocclr
So the package currently installs in GNUInstallDirs -
/usr/lib/x86_64-linux-gnu, /usr/include/hip. There are two main
shared
objects: libamdhip64-5 and libhiprtc-builtins-5.
lgtm
- device-side (GPU kernels), allegedly rocmgdb ioctls are not in
the amdgpu.ko of the debian kernel yet.
The first upload of this package will enter experimental. Have you
checked the kernel (5.17) in debian/experimental?
not yet
rocmgdb advertises "seamless debug sessions between CPU and GPU",
so is fixing dwz even a requirement?
- I am not sure if the shared objects will work for any other GPU
than gfx906 currently...
Basically, there is work, at least for me, to undestand what hip is
about still: I don't understand clearly what is in the elf files and
I would need to dive in elf spec more for dwz, for the lintian
warning
symbols-file-contains-current-version-with-debian-revision, and
finally in order to look for the presence of GPU-code in there.
debian/*.symbols file should not contain lines like ".* 5.0.0-1",
where the "-1" is the debian revision. Just removing that should
eliminate that warning.
I understood this with previous packages, but here I took the time
to manually update the symbols from 4.5.2 to 5.0.0, without
-1~exp1 and still have the warning. FWIW, David Bremner on
OFTC#debian-mentors was intrigued. Maybe won't you have
this warning...
There are also at least two packages in the hip department that would
be nice packaging IMO: the hip-examples, and the hip-testsuite repos.
Maybe also hipcc and hipcpu?
the testsuite sounds great as a high-level sanity checker for the
whole stack of packages in debian.
Agreed
This was me throwing all I have about this package, not knowing very
well what to do next / knowing that $DAYJOB will be taxing here next
weeks.
Just take your time.
Reply to: