Removing the crc16 check shows interesting results;
SPP parallel, pcsend 332 KB/s, pcrecv 135 KB/s
PS2 (bidirectional) parallel, pcsend 339 KB/s, pcrecv 255 KB/s
EPP (hardware handshaking) parallel, pcsend 598 KB/s, pcrecv 366 KB/s
Only a slight bit of increase for SPP/PS2, but a decent increase for EPP.
Maybe I'll try crc8 to see how it does - do you guys think that would be better (only have 256 values) than some sort of a simple addition of each byte into a uint16_t would be?
I want to keep the checksum to 2 bytes.
© 2020 Microchip Technology Inc.