libosmocore[master]: core/conv: implement optimized Viterbi decoder

This is merely a historical archive of years 2008-2021, before the migration to mailman3.

A maintained and still updated list archive can be found at https://lists.osmocom.org/hyperkitty/list/gerrit-log@lists.osmocom.org/.

Tom Tsou gerrit-no-reply at lists.osmocom.org
Wed Mar 15 23:46:20 UTC 2017


Patch Set 5:

(2 comments)

https://gerrit.osmocom.org/#/c/1337/4/src/viterbi.c
File src/viterbi.c:

Line 539:  * the interval specified by the decoder.
> Is there a more specific function name?
The 'len' value and loop refers to moving across the horizontal length of the trellis. A more specific name that matches Viterbi algorithm nomenclature could be something like trellis_traversal() or forward_traverse().


Line 564: 	int8_t depunc[dec->len * dec->n];
> Maybe we could do it inside the init_vdec (current name: alloc_vdec)?
The API limits us to having decoder object construction and teardown on on every message decode. Sylvain and I had a discussion on this topic years ago and we agreed that reducing the overhead wasn't worth the complexity of an API change.

Where the decoder construction/destruction takes place behind the API isn't very important. It just has to happen somewhere and be transparent to the application calling into libosmocore.


-- 
To view, visit https://gerrit.osmocom.org/1337
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I74d355274b4176a7d924f91ef3c96912ce338fb2
Gerrit-PatchSet: 5
Gerrit-Project: libosmocore
Gerrit-Branch: master
Gerrit-Owner: Vadim Yanitskiy <axilirator at gmail.com>
Gerrit-Reviewer: Alexander Chemeris <Alexander.Chemeris at gmail.com>
Gerrit-Reviewer: Harald Welte <laforge at gnumonks.org>
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou <tom at tsou.cc>
Gerrit-Reviewer: Vadim Yanitskiy <axilirator at gmail.com>
Gerrit-Reviewer: tnt <tnt at 246tNt.com>
Gerrit-HasComments: Yes



More information about the gerrit-log mailing list