Theodore Omtzigt

TVM is respectable, but we believe too DL specific. Full disclosure, we basically are working on a competing technology that scales to a broader range of applications. Having said that,...

@dendisuhubdy that is awesome. Do you have any results that use tiny-dnn that you can share?

@edgarriba that indeed is the $64,000 question. There is a large divide opening up between training and compiling the trained network down for efficient inference. I don't know if this...

@Dwood15 concern noted. The discussion about ENAS would be a layer on top. The issue that @edgarriba brings up is the innovations that have taken hold on the network training...

@olbender very nice! big fan of the OpenDLV lib, header-only, minimal dependency software. This is shaping up to be quite encouraging; the tinny-dnn community stepping up to solve the succession...

@edeforas it would be great if we can have tiny-dnn build reliably and without warnings on gcc, clang, and msvc. The VSCode environment is being proliferated and offers great functionality...

Isolated it to be related to the argument --wave=posit_add.ghw, that is, removing the wave generation yields a working simulator. Still working on getting a bug reproduction example for testing....

here is a reproduction: $ sh repro.sh ******************** GHDL Bug occured **************************** Please report this bug on https://github.com/tgingold/ghdl/issues GHDL release: 0.35 (tarball) [Dunoon edition] Compiled with GNAT Version: GPL 2017...

@Paebbels @diogratia thank you for your help with this issue. I installed the 0.36-dev blob that @Paebbels posted above, and added the architecture to the elaboration, but unfortunately, the setup...

@markusgft thank you for the guidance. For hw accelerated MPC co-development, where/who/what should we be connecting to?