> That is indeed very surprising. I think the fact that Go isn't dead-slow can be attributed mainly to the sheer speed of CPUs.

Well, I suspect that in any reasonably complex C/C++ application (e.g. firefox) compilers do pass params via stack and probably they do so rather often. Even on x86_64.

/usr/lib/firefox/firefox objdump:

        a8b8:       48 83 ec 40             sub    $0x40,%rsp
        a8bc:       48 89 d3                mov    %rdx,%rbx
        a8bf:       48 89 f5                mov    %rsi,%rbp
        a8c2:       49 89 ff                mov    %rdi,%r15
        a8c5:       64 48 8b 04 25 28 00    mov    %fs:0x28,%rax
        a8cc:       00 00 
        a8ce:       48 89 44 24 38          mov    %rax,0x38(%rsp)
        a8d3:       44 8b 76 10             mov    0x10(%rsi),%r14d
        a8d7:       44 8b 62 10             mov    0x10(%rdx),%r12d
        a8db:       48 89 74 24 20          mov    %rsi,0x20(%rsp)
        a8e0:       48 89 54 24 28          mov    %rdx,0x28(%rsp)
        a8e5:       c7 44 24 30 02 00 00    movl   $0x2,0x30(%rsp)
        a8ec:       00 
        a8ed:       48 8d 7c 24 20          lea    0x20(%rsp),%rdi
        a8f2:       e8 39 fd ff ff          callq  a630 <_ZN7mozilla11Compression3LZ417decompressPartialEPKcmPcmPm@@Base+0x2c0>
    --
      etc.

/usr/lib/firefox/libxul.so objdump:

      81cf5e:       48 89 44 24 30          mov    %rax,0x30(%rsp)
      81cf63:       4c 89 7c 24 58          mov    %r15,0x58(%rsp)
      81cf68:       4d 89 cf                mov    %r9,%r15
      81cf6b:       f2 44 0f 11 64 24 60    movsd  %xmm12,0x60(%rsp)
      81cf72:       e8 a9 00 00 00          callq  81d020 <mont_mulf_noconv@@xul66+0xa30>
    --
      etc.