The bubble of cryptocurrencies helped to contribute to a new wave of interest in hardware-level reprogrammable computing, specifically in field-programmable gate arrays. Hardware reprogrammability offers massive speed advantages over the regular software-level reprogrammability (as most coders are used to), but today it’s limited to very limited and specific problem sets, such as stream processing/encoding, embedded devices, and cryptocurrency “mining”.
The news of Intel acquiring Altera could potentially be a massive step forward for this technology. With Intel’s advanced chip design and engineering resources brought to bear on FPGA technology, there is the possibility to highly-reprogrammable logic chips. Today the barriers to more general-purpose use of FPGA’s are the arcane steps required to reprogram them and the difficulty of actually writing code to properly target & utilize them.
However, easier reprogramming—perhaps via Intel’s new foray into the space—would be a big step forward. That, combined with reduced cost, would allow a wider range of programmers to start working with hardware-level reprogrammable computers, and I believe compiler technology & tooling would quickly follow.
Perhaps someday soon we’ll see a narrowing of the gap (and even eventual unificiation) between software- and hardware-level programming, helping to reduce and/or eliminate the Von Neumann and fetch-decode-execute bottlenecks.