New Feature: Segmented Memory

[colinoflynn]

ChipWhisperer is adding segmented memory - see https://github.com/newaetech/chipwhisperer-jupyter/blob/master/demos/Using%20Segmented%20Memory%20for%20Hardware%20AES%20(STM32F4).ipynb for the best example of it. You need to use the debug FPGA bitstream in that example.

See https://chipwhisperer.readthedocs.io/en/latest/api.html for details in:

• fifo_fill_mode talks about enabling this mode.
• capture_segmented and get_last_trace_segmented are needed instead of capture and get_last_trace

This is only useful for short captures (think - hardware AES). But you can bump up your capture rates in the range of 1000-4000 traces/second in my testing. More may even be possible, there is still a lot of overhead right now that isn't optimized away. But at some point I'm not sure how much faster becomes useful, since you can capture 1M traces in ~15 mins, 1B traces in ~a week. You'll need a lot of GPUs to process those traces.

Keeping this issue tracker open to get issues/feedback on it until it goes into the main FPGA build.