Hi,
I am just starting to look at Chipyard, I am particularly interested in generating a 2 core system (Rocket and/or CVA6 based), with custom instructions for one of the cores and memory mapped accelerators.
My end goal would be to generate something suitable for ASIC, but initially going to FPGA as an interim step for real-world evaluation of performance of the accelerators.
I have been doing a bunch of reading, but I have a few questions still. They are pretty basic...
* What is the recommended workflow for versioning my customisation? System definition, custom instructions etc. Can I create my own repo and "link" it into my local ChipYard?
* When extending ChipYard cores with custom instructions, can I build a cycle accurate emulator QEMU/SPIKE etc?
* Can ChipYard aid in generating build tools that are aware of the custom instructions, C Intrinsics or so other way? (GCC and/or Clang)
Thanks for any help,
Cameron