I am looking for help in building up model for ReRAM (Resistive RAM) for some application in Spiking Neural Network. Which tool should I prefer for this?
This will depend on the implementation of the spiking neural network itself (SNN). How we at Southampton like to deal with such tasks is to model the SNN in Cadence and then add into it our Verilog-A memristive model (see "A Data-Driven Verilog-A ReRAM Model" published in IEEE TCAD). This is an approach whereby we require our CMOS part of the system to be something that we can realistically manufacture and we want to see parasitics etc.
Alternatively, you might want to look for a SPICE model e.g. for use with LTSPICE or similar tools. These are good for getting a decent degree of hardware realism, especially if you want to mix in some standard discrete components or want to keep your simulation relatively ideal and without too much attention to the particular technology being used to fabricate the design.
Finally you could even go to a python implementation using a python-based model. This would be best if you restrict yourself to very coarse-grain modelling, e.g. if you model transistors as switches, or just want to use generic neuron models that somehow (and you don't care how) generate appropriate spike waveforms.
Note: Overall, there are a number of models out there, of course e.g. VTEAM so choose wisely depending on what precisely it is you want to model.
Thank you Serb for your helpful and valuable information.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence