Patch Set 3: Hello Alexander,
I have some questions about this. You suggested to add some fake-target for avx. Maybe I get the things wrong, but I think the things will be not just some changes in the build system. Currently the targets are compiled with -msse4.1 and -mavx. The targets are both different and result into two different .a files. The function prototypes have also different names to avoid clashing. Then the runtime CPU detection detects which CPU is installed and picks the matching implementation by populating some generically named function pointers. Thats how we switch between the different implementations dynamically. You said the avx is similar to (when I remeber correctly) sse 4.1. So technically the that code should compile with -mavx as well. But I still need the sse 4.1 support. An avx compiled sse code would not run on a pure sse machine, right? I would than have to mangle the function prototypes to avoid clashing between the sse 4.1 and the avx version. And of course we need an additional check to populate the function pointers. Maybe I am thinking this overcomplicated, please let me know if I am wrong or right. regards, Philipp -- To view, visit https://gerrit.osmocom.org/2100 To unsubscribe, visit https://gerrit.osmocom.org/settings Gerrit-MessageType: comment Gerrit-Change-Id: Iba74f8a6e4e921ff31e4bd9f0c7c881fe547423a Gerrit-PatchSet: 3 Gerrit-Project: osmo-trx Gerrit-Branch: master Gerrit-Owner: dexter <[email protected]> Gerrit-Reviewer: Alexander Chemeris <[email protected]> Gerrit-Reviewer: Harald Welte <[email protected]> Gerrit-Reviewer: Jenkins Builder Gerrit-Reviewer: Max <[email protected]> Gerrit-Reviewer: Vadim Yanitskiy <[email protected]> Gerrit-Reviewer: dexter <[email protected]> Gerrit-HasComments: No
