Bug#1070788: mesa: Package teflon - TensorFlow Lite delegate library
Hi, On Wed, 15 May 2024 11:18:48 +0100 Christopher Obbard wrote: > Hi, > > [+Timo and X team] > > On Thu, 09 May 2024 08:42:54 +0100 Christopher Obbard > wrote: > > Source: mesa > > Version: 24.1.0~rc2-1 > > Severity: wishlist > > X-Debbugs-Cc: obba...@debian.org, to...@tomeuvizoso.net > > > > Dear Maintainer, > > > > Mesa 24.1 contains a TensorFlow Lite delegate that can make use of NPUs > > to accelerate ML inference. It is implemented in the form of a external > > delegate, a shared library that the TensorFlow Lite runtime can load at > > startup[0]. > > > > We should consider packaging this for users. > > > > It should be a matter of just passing -Dteflon=true to the build config > > > > Since it builds a shared library, libteflon.so, we should package this > > as a separate library. How does libteflon0 sound? > > > > We can use a description similar to the first paragraph, which has been > > taken from [1]. > > > > I will prepare a merge request on Salsa for this over the coming weeks. > > > I've made an initial implementation for the packaging for the Teflon library > at https://salsa.debian.org/xorg-team/lib/mesa/-/merge_requests/39 I've reworked the packaging for Teflon and rebased on unstable; it's available at https://salsa.debian.org/xorg-team/lib/mesa/-/merge_requests/39 @Timo, Can you review the changes once more, please? Thanks!
Bug#1071222: mesa: FTBFS due to missing build dependency python3-pycparser
Source: mesa
Version: 24.1.0~rc3-1
Severity: minor
Tags: ftbfs
Dear Maintainer,
On my local machine, mesa FTBFS unless I install python3-pycparser:
Running command: /usr/bin/python3 -c '
try:
from packaging.version import Version
except:
from distutils.version import StrictVersion as Version
import pycparser
assert Version(pycparser.__version__) >= Version("2.20")
'
--- stdout ---
--- stderr ---
Traceback (most recent call last):
File "", line 6, in
ModuleNotFoundError: No module named 'pycparser'
../src/etnaviv/hwdb/meson.build:17:2: ERROR: Problem encountered: Python
(3.x) pycparser module >= 2.20 required to build mesa.
It seems to install the dependency and build fine on Debian buildd, so
marking the bug as minor. Since this package seems to be a direct
dependency for the etnaviv hwdb bits, I think we should add
python3-pycparser as a direct build depends.
-- System Information:
Debian Release: trixie/sid
APT prefers unstable
APT policy: (500, 'unstable'), (1, 'experimental')
Architecture: amd64 (x86_64)
Kernel: Linux 6.7.12-amd64 (SMP w/16 CPU threads; PREEMPT)
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8),
LANGUAGE=en_GB:en
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled
Bug#1070788: mesa: Package teflon - TensorFlow Lite delegate library
Hi, [+Timo and X team] On Thu, 09 May 2024 08:42:54 +0100 Christopher Obbard wrote: > Source: mesa > Version: 24.1.0~rc2-1 > Severity: wishlist > X-Debbugs-Cc: obba...@debian.org, to...@tomeuvizoso.net > > Dear Maintainer, > > Mesa 24.1 contains a TensorFlow Lite delegate that can make use of NPUs > to accelerate ML inference. It is implemented in the form of a external > delegate, a shared library that the TensorFlow Lite runtime can load at > startup[0]. > > We should consider packaging this for users. > > It should be a matter of just passing -Dteflon=true to the build config > > Since it builds a shared library, libteflon.so, we should package this > as a separate library. How does libteflon0 sound? > > We can use a description similar to the first paragraph, which has been > taken from [1]. > > I will prepare a merge request on Salsa for this over the coming weeks. I've made an initial implementation for the packaging for the Teflon library at https://salsa.debian.org/xorg-team/lib/mesa/-/merge_requests/39 Reviews welcome! Thanks, Chris
Bug#1070788: mesa: Package teflon - TensorFlow Lite delegate library
Source: mesa Version: 24.1.0~rc2-1 Severity: wishlist X-Debbugs-Cc: obba...@debian.org, to...@tomeuvizoso.net Dear Maintainer, Mesa 24.1 contains a TensorFlow Lite delegate that can make use of NPUs to accelerate ML inference. It is implemented in the form of a external delegate, a shared library that the TensorFlow Lite runtime can load at startup[0]. We should consider packaging this for users. It should be a matter of just passing -Dteflon=true to the build config Since it builds a shared library, libteflon.so, we should package this as a separate library. How does libteflon0 sound? We can use a description similar to the first paragraph, which has been taken from [1]. I will prepare a merge request on Salsa for this over the coming weeks. [0]: https://www.tensorflow.org/api_docs/python/tf/lite/experimental/load_delegate. [1]: https://docs.mesa3d.org/teflon.html Thanks! Chris -- System Information: Debian Release: trixie/sid APT prefers unstable APT policy: (500, 'unstable'), (1, 'experimental') Architecture: amd64 (x86_64) Kernel: Linux 6.7.12-amd64 (SMP w/16 CPU threads; PREEMPT) Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8), LANGUAGE=en_GB:en Shell: /bin/sh linked to /usr/bin/dash Init: systemd (via /run/systemd/system) LSM: AppArmor: enabled
Bug#1022058: spirv-headers: Please upgrade to spirv-headers >=1.3.224
Package: spirv-headers Version: 1.6.1+1.3.216.0-1 Severity: normal Dear Maintainer, mesa 22.3 in debian-experimental FTBFS with the following tests failing: Failed Tests (4): LLVM_SPIRV :: transcoding/AtomicFMaxEXT.ll LLVM_SPIRV :: transcoding/AtomicFMaxEXTForOCL.ll LLVM_SPIRV :: transcoding/AtomicFMinEXT.ll LLVM_SPIRV :: transcoding/AtomicFMinEXTForOCL.ll Turns out we need to bump spirv-headers to >= 1.3.224 and make sure llvm-spirv-translator gets built against that.
Bug#1021666: mesa: Add wayland-protocols version >= 1.24 to Mesa's Build-Depends
Source: mesa Version: 22.2.0-1 Severity: normal Dear Maintainer, With older version of wayland-protocols package available (1.20), the build starts but fails late with: Dependency wayland-protocols found: NO found 1.20 but need: '>= 1.24' Invalid version of dependency, need 'wayland-protocols' ['>= 1.24'] found '1.20'. CMake binary for 1 is cached as not found CMake binary for machine 1 not found. Giving up. Run-time dependency wayland-protocols found: NO ../meson.build:2057:2: ERROR: Invalid version of dependency, need 'wayland-protocols' ['>= 1.24'] found '1.20'. Can we add a direct version to the wayland-protocols dependency of >= 1.24 to catch this error earlier? -- System Information: Debian Release: bookworm/sid APT prefers unstable APT policy: (500, 'unstable') Architecture: amd64 (x86_64) Kernel: Linux 5.19.0-1-amd64 (SMP w/8 CPU threads; PREEMPT) Kernel taint flags: TAINT_WARN Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8), LANGUAGE=en_GB:en Shell: /bin/sh linked to /usr/bin/dash Init: systemd (via /run/systemd/system) LSM: AppArmor: enabled
