baokim1802/fpgapy

NumPy-like API accelerated with FPGA

FpgaPy: NumPy-like API accelerated with FPGA

FpgaPy is an implementation of NumPy-compatible multi-dimensional array on FPGA. FpgaPy consists of the core multi-dimensional array class, fpgapy.ndarray, and many functions on it. It supports a subset of numpy.ndarray interface.

To take advantage of the FPGA parallelism, FpgaPy is built on top of asyncio.

Example

import asyncio
import fpgapy as fp

event_loop = asyncio.get_event_loop()
fp.set_event_loop(event_loop)

async def main():
    a = fp.array([0, 1, 2])
    b = fp.array([1, 1, 1])
    c = a + b
    d = b + a
    print(a)            # array([0, 1, 2], dtype='int64')
    print(b)            # array([1, 1, 1], dtype='int64')
    print(c)            # <Task pending coro=<get_add() running at ...>>
    print(d)            # <Task pending coro=<get_add() running at ...>>
    c = await c
    print(c)            # array([1, 2, 3], dtype='int64')
    print(d)            # <Task finished coro=<get_add() done, defined at ...> result=array([1, 2, ...dtype='int64')>

event_loop.run_until_complete(main())