‘find’ is essentially a recursively visiting-evaluation loop. It is supposed to print every result matches the expression immediately (or alternatingly to those sub-expressions, you may say), therefore it has the best performance.
Am I correct in saying that what you want to do is deferring ‘results’, actions or side-effects to the end of searching? This is not what ‘find’ good at, and it implies a buffer to keep the intermediate results. In case you’re avoiding to simply do the ‘find’ twice, from a performance point of view, I’d like to suggest tagging those results, sorting them later (ie. piping to ‘sort’, which does has a buffer internally) and finally cutting out tags: find /usr -name sh -printf 'a:%p\n' -o -name gcc -printf 'b:%p\n' b:/usr/bin/gcc a:/usr/bin/sh b:/usr/lib/gcc a:/usr/lib/klibc/bin/sh b:/usr/share/bash-completion/completions/gcc a:/usr/share/bash-completion/completions/sh b:/usr/share/doc/gcc b:/usr/share/doc/gcc-9-base/gcc find /usr -name sh -printf 'a:%p\n' -o -name gcc -printf 'b:%p\n' | sort -s -t':' -k1,1 | cut -d':' -f2- /usr/bin/sh /usr/lib/klibc/bin/sh /usr/share/bash-completion/completions/sh /usr/bin/gcc /usr/lib/gcc /usr/share/bash-completion/completions/gcc /usr/share/doc/gcc /usr/share/doc/gcc-9-base/gcc
