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
