diff options
author | Josh Blum | 2012-09-14 09:04:26 -0700 |
---|---|---|
committer | Josh Blum | 2012-09-14 09:04:26 -0700 |
commit | 516be5f5836da2d93e4ce72945d9c0c1339a26f2 (patch) | |
tree | 4645a6a2a5d88ee99cb81372f465c136cc8aff69 | |
parent | f48974cbfc5304003ad5296d8df610579768d362 (diff) | |
download | sandhi-516be5f5836da2d93e4ce72945d9c0c1339a26f2.tar.gz sandhi-516be5f5836da2d93e4ce72945d9c0c1339a26f2.tar.bz2 sandhi-516be5f5836da2d93e4ce72945d9c0c1339a26f2.zip |
cleanup and adding to TODO.txt
-rw-r--r-- | TODO.txt | 19 |
1 files changed, 19 insertions, 0 deletions
@@ -5,22 +5,41 @@ * generate block task loops with fixed lengths ** need a lookup/jump table ** prove that compiler does unrolling + * handle forecast ** not 100% sure here + * add hooks to specify input reserve ** automatically calculate from output multiple and rel rate + * gr stream mux is on drugs + * handle calculating noutputitems using ninputs as a constraint + * allocation ** hooks for advanced allocation ** override the default allocator (output token, recommended bytes) + * python wrapper for block will come from grextras + * bring in numanuma ** thread prio ** thread affinity + * track memcpys for block usage stats + * runtime history, input/output multiple change? ** resend the hint to the allocator and possibly re-allocate + * determine when outputs should be flushed, or continual filling + * hooks for setting the "active framework" ** set this before creating blocks to give them an affinity + +* deal with settings changing at runtime +** example, block calls set_history + +* issue with marking done on multiple inputs, 1 done, 1 not +** see channel model qa as example + +* put this todo into wiki format checklist |