Extending the idea of fastcat, this is an attempt to implement cat with higher throughput by using splice(2) to avoid copying between kernel space and user space.
$ make$ make installJust like cat:
$ fcat *.cTested with 10GB file generated with dd.
$ dd if=/dev/random of=10gb.txt count=1024 bs=10485760The pipe speed can be checked with pv.
GNU cat
$ cat 10gb.txt | pv -r > /dev/null
[ 500MiB/s]fcat
$ fcat 10gb.txt | pv -r > /dev/null
[ 780MiB/s]Interestingly, the speed of fcat fluctuates more than cat. And sometimes it is slower than cat.