Török Edwin wrote:
> Do you really need to use Int64 for that though? Won't the 63-bit
> version do?

I'm running 32-bit.

> > I am unable to reproduce your results. Here, the time falls from 24s
> > to 19.5s (using ocamlopt 3.12.0 on Intel x86) which is still 26×
> > slower than HLVM.

Sorry, I'm actually using an Opteron x86 (logged in from an Intel x86!).

> Do you still have 'idiv' in the compiled code? See my attached
> assembly, and compare it with yours please.
> I was doing the test on 64-bit, with ocamlopt 3.11.2 and 3.12.0.

I get what appear to be calls to C code:

camlCollatz__collatzLen_1030:
        subl    $8, %esp
.L103:
        movl    %eax, 4(%esp)
        movl    %ebx, 0(%esp)
        pushl   $camlCollatz__10
        pushl   %ebx
        movl    $caml_equal, %eax
        call    caml_c_call
.L104:
        addl    $8, %esp
        cmpl    $1, %eax
        je      .L102
        movl    4(%esp), %eax
        addl    $8, %esp
        ret
        .align  16
.L102:
        pushl   $camlCollatz__8
        movl    4(%esp), %eax
        pushl   %eax
        movl    $caml_int64_and, %eax
        call    caml_c_call
.L105:
        addl    $8, %esp
        pushl   $camlCollatz__9
        pushl   %eax
        movl    $caml_equal, %eax
        call    caml_c_call
.L106:
        addl    $8, %esp
        cmpl    $1, %eax
        je      .L101
        pushl   $3
        movl    4(%esp), %eax
        pushl   %eax
        movl    $caml_int64_shift_right, %eax
        call    caml_c_call
.L107:
        addl    $8, %esp
        movl    %eax, %ebx
        jmp     .L100
        .align  16
.L101:
        movl    0(%esp), %eax
        pushl   %eax
        pushl   $camlCollatz__6
        movl    $caml_int64_mul, %eax
        call    caml_c_call
.L108:
        addl    $8, %esp
        pushl   $camlCollatz__7
        pushl   %eax
        movl    $caml_int64_add, %eax
        call    caml_c_call
.L109:
        addl    $8, %esp
        movl    %eax, %ebx
.L100:
        movl    4(%esp), %eax
        addl    $2, %eax
        jmp     .L103

> FWIW the original code took 2.8 seconds here, so only 4x slower (this
> is an AMD Phenom II x6 1090T CPU). It probably depends how fast/slow
> the 'idiv' is on your CPU.

The performance of idiv is irrelevant here. The bottleneck may be those C calls but I don't understand why they are being generated.

Cheers,
Jon.