« Cadence spots the elephant! | Main | DAC:2010 - Vowels, Gnats and Hubs. »

05/25/2010

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

David Murray

Hi Steve,
Thanks for the feedback! TLM synthesis is definitely an avenue to enable the virtual platform. The ESL aspects can sometimes be quite focused on the 'Transaction' and how accurate it is in relation to 'time'. Another axis that I think is very important is how 'functionally' accurate the transaction target is. If from the start, we can keeping all models 'register' accurate, without too much overhead, this would allow much better HW/SW integration.

Effectively you are creating a common based-reference model that both the Virtual model and RTL models can relate to, a kind of intermediate ESL model, sychronized to both.

regards
- Dave

David Murray

Hi Michal, I agree. Alignment between the various models would help streamline HW/SW converence. It is important to ensure at least the foundation of the HW/SW interface is coherent before you build virtual prototypes or FPGA prototyess on it
regards
- Dave

Steve Brown

I enjoyed reading your white paper and seeing how relevant the EDA360 industry vision is to your offering. It's a very impressive solution! We should discuss TLM synthesizability to connect the virtual platform world to the implementation flow and shorten schedules further.

Msiwinski

Dave - fully agree with your assessment on elements to fulfill the EDA360 vision for the industry. Rapid assembly of SoCs and systems, with means of effective trade-offs and optimizations, is necessary to enable the growing set of Integrators to succeed ...and drive towards higher profitability. The industry solutions will also need to comprehend various levels of abstractions to provide an open and connected path to rapid convergence between software and hardware. - Michal

The comments to this entry are closed.