FuriosaAI's low-level software stack

I am currently exploring the use of the RNGD. I am reaching out to inquire whether it is possible to perform low-level kernel optimization on the RNGD, similar to writing custom CUDA kernels for GPUs. Does FuriosaAI provide an API for developing custom kernels that can be executed directly on the RNGD NPU? Alternatively, are there other methods available to achieve similar custom kernel execution or low-level optimization on the RNGD?

Thanks for your interest! We’re currently developing a low-level programming interface for RNGD. The initial version will be released on third quarter of this year.

1 Like