Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Strange behavior in big table benchmark #24

Open
Kogia-sima opened this issue Jul 10, 2020 · 11 comments
Open

Strange behavior in big table benchmark #24

Kogia-sima opened this issue Jul 10, 2020 · 11 comments
Labels
help wanted Extra attention is needed Type: Performance Performance improvement

Comments

@Kogia-sima
Copy link
Member

Kogia-sima commented Jul 10, 2020 •
edited
Loading

(Currently this behaviour is only observed in Big table benchmark)

When I remove #[inline] from Buffer::reserve_internal() method, call to itoap::write_u64 function will become indirect call (via function address on register), which makes the program 30% slower.

This issue prevents removing #[inline] for reserve_internal() method, resulting in slightly larger binary.

Full Assembly Output 
<sailfish_research::BigTable as sailfish::TemplateOnce>::render_once_to_string:
 push    rbp
 push    r15
 push    r14
 push    r13
 push    r12
 push    rbx
 sub     rsp, 56
 mov     r12, rdx
 mov     r13, rsi
 mov     qword, ptr, [rsp, +, 48], rdi
 mov     rax, qword, ptr, [rcx]
 movdqu  xmm0, xmmword, ptr, [rcx, +, 8]
 pshufd  xmm0, xmm0, 78
 mov     qword, ptr, [rcx], 1
 xorps   xmm1, xmm1
 mov     qword, ptr, [rsp, +, 24], rcx
 movups  xmmword, ptr, [rcx, +, 8], xmm1
 mov     qword, ptr, [rsp], rax
 movdqu  xmmword, ptr, [rsp, +, 8], xmm0
 mov     rsi, qword, ptr, [rip, +, _ZN70_$LT$sailfish_research..BigTable$u20$as$u20$sailfish..TemplateOnce$GT$21render_once_to_string9SIZE_HINT17h2be0f08b3976f89cE]
 mov     rax, qword, ptr, [rsp, +, 8]
 mov     rcx, qword, ptr, [rsp, +, 16]
 lea     rdx, [rax, +, rsi]
 cmp     rdx, rcx
 ja      .LBB10_1
 lea     rdx, [rax, +, 8]
 cmp     rdx, rcx
 ja      .LBB10_4
.LBB10_6:
 mov     rcx, qword, ptr, [rsp]
 movabs  rdx, 738138905116767292
 mov     qword, ptr, [rcx, +, rax], rdx
 mov     rcx, qword, ptr, [rsp, +, 8]
 lea     rax, [rcx, +, 8]
 mov     qword, ptr, [rsp, +, 8], rax
 add     rcx, 12
 cmp     rcx, qword, ptr, [rsp, +, 16]
 ja      .LBB10_7
.LBB10_9:
 mov     rcx, qword, ptr, [rsp]
 mov     dword, ptr, [rcx, +, rax], 1047688252
 mov     rax, qword, ptr, [rsp, +, 8]
 add     rax, 4
 mov     qword, ptr, [rsp, +, 8], rax
 shl     r12, 3
 lea     rcx, [r12, +, 2*r12]
 test    rcx, rcx
 je      .LBB10_28
 add     rcx, r13
 mov     r12, qword, ptr, [rip, +, _ZN5itoap9write_u6417hc82e35bf780e3b54E@GOTPCREL]
 mov     qword, ptr, [rsp, +, 32], rcx
.LBB10_11:
 lea     rcx, [rax, +, 4]
 cmp     rcx, qword, ptr, [rsp, +, 16]
 ja      .LBB10_12
.LBB10_14:
 mov     rcx, qword, ptr, [rsp]
 mov     dword, ptr, [rcx, +, rax], 1046770748
 mov     rsi, qword, ptr, [rsp, +, 8]
 add     rsi, 4
 mov     qword, ptr, [rsp, +, 8], rsi
 mov     r14, qword, ptr, [r13, +, 16]
 test    r14, r14
 je      .LBB10_24
 mov     r15, qword, ptr, [r13]
 shl     r14, 3
 xor     ebp, ebp
.LBB10_16:
 mov     rbx, qword, ptr, [r15, +, rbp]
 lea     rax, [rsi, +, 20]
 cmp     rax, qword, ptr, [rsp, +, 16]
 ja      .LBB10_17
.LBB10_19:
 add     rsi, qword, ptr, [rsp]
 mov     rdi, rbx
 call    r12
 mov     rdx, qword, ptr, [rsp, +, 8]
 lea     rcx, [rdx, +, rax]
 mov     qword, ptr, [rsp, +, 8], rcx
 add     rax, rdx
 add     rax, 9
 cmp     rax, qword, ptr, [rsp, +, 16]
 ja      .LBB10_21
.LBB10_23:
 mov     rax, qword, ptr, [rsp]
 movzx   edx, byte, ptr, [rip, +, .Lanon.28d6b14a742276b51f6db9706e341907.9+8]
 mov     byte, ptr, [rax, +, rcx, +, 8], dl
 mov     rdx, qword, ptr, [rip, +, .Lanon.28d6b14a742276b51f6db9706e341907.9]
 mov     qword, ptr, [rax, +, rcx], rdx
 mov     rsi, qword, ptr, [rsp, +, 8]
 add     rsi, 9
 mov     qword, ptr, [rsp, +, 8], rsi
 add     rbp, 8
 cmp     r14, rbp
 jne     .LBB10_16
 jmp     .LBB10_24
.LBB10_17:
 mov     esi, 20
 mov     rdi, rsp
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rsi, qword, ptr, [rsp, +, 8]
 jmp     .LBB10_19
.LBB10_21:
 mov     esi, 9
 mov     rdi, rsp
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rcx, qword, ptr, [rsp, +, 8]
 jmp     .LBB10_23
.LBB10_24:
 lea     rax, [rsi, -, 4]
 mov     qword, ptr, [rsp, +, 8], rax
 add     rsi, 5
 cmp     rsi, qword, ptr, [rsp, +, 16]
 ja      .LBB10_25
.LBB10_27:
 add     r13, 24
 mov     rcx, qword, ptr, [rsp]
 mov     dl, byte, ptr, [rip, +, .Lanon.28d6b14a742276b51f6db9706e341907.10+8]
 mov     byte, ptr, [rcx, +, rax, +, 8], dl
 mov     rdx, qword, ptr, [rip, +, .Lanon.28d6b14a742276b51f6db9706e341907.10]
 mov     qword, ptr, [rcx, +, rax], rdx
 mov     rax, qword, ptr, [rsp, +, 8]
 add     rax, 9
 mov     qword, ptr, [rsp, +, 8], rax
 mov     rcx, qword, ptr, [rsp, +, 32]
 cmp     r13, rcx
 jne     .LBB10_11
 jmp     .LBB10_28
.LBB10_12:
 mov     esi, 4
 mov     rdi, rsp
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rax, qword, ptr, [rsp, +, 8]
 jmp     .LBB10_14
.LBB10_25:
 mov     esi, 9
 mov     rdi, rsp
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rax, qword, ptr, [rsp, +, 8]
 jmp     .LBB10_27
.LBB10_28:
 lea     rcx, [rax, -, 4]
 mov     qword, ptr, [rsp, +, 8], rcx
 add     rax, 5
 cmp     rax, qword, ptr, [rsp, +, 16]
 ja      .LBB10_29
.LBB10_31:
 mov     rax, qword, ptr, [rsp]
 mov     dl, byte, ptr, [rip, +, .Lanon.28d6b14a742276b51f6db9706e341907.11+8]
 mov     byte, ptr, [rax, +, rcx, +, 8], dl
 mov     rdx, qword, ptr, [rip, +, .Lanon.28d6b14a742276b51f6db9706e341907.11]
 mov     qword, ptr, [rax, +, rcx], rdx
 mov     rax, qword, ptr, [rsp, +, 8]
 lea     rcx, [rax, +, 9]
 mov     qword, ptr, [rsp, +, 8], rcx
 shr     rcx, 2
 add     rax, rcx
 add     rax, 9
 mov     rcx, qword, ptr, [rip, +, _ZN70_$LT$sailfish_research..BigTable$u20$as$u20$sailfish..TemplateOnce$GT$21render_once_to_string9SIZE_HINT17h2be0f08b3976f89cE]
 cmp     rcx, rax
 jae     .LBB10_33
 mov     qword, ptr, [rip, +, _ZN70_$LT$sailfish_research..BigTable$u20$as$u20$sailfish..TemplateOnce$GT$21render_once_to_string9SIZE_HINT17h2be0f08b3976f89cE], rax
.LBB10_33:
 mov     rbx, qword, ptr, [rsp]
 movdqu  xmm0, xmmword, ptr, [rsp, +, 8]
 pshufd  xmm0, xmm0, 78
 mov     rax, qword, ptr, [rsp, +, 24]
 mov     rdi, qword, ptr, [rax]
 test    rdi, rdi
 je      .LBB10_36
 mov     rsi, qword, ptr, [rax, +, 8]
 test    rsi, rsi
 je      .LBB10_36
 mov     edx, 1
 movdqa  xmmword, ptr, [rsp, +, 32], xmm0
 call    qword, ptr, [rip, +, __rust_dealloc@GOTPCREL]
 movdqa  xmm0, xmmword, ptr, [rsp, +, 32]
 mov     rax, qword, ptr, [rsp, +, 24]
.LBB10_36:
 mov     qword, ptr, [rax], rbx
 movdqu  xmmword, ptr, [rax, +, 8], xmm0
 mov     rax, qword, ptr, [rsp, +, 48]
 mov     qword, ptr, [rax], 0
 add     rsp, 56
 pop     rbx
 pop     r12
 pop     r13
 pop     r14
 pop     r15
 pop     rbp
 ret
.LBB10_1:
 mov     rdi, rsp
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rax, qword, ptr, [rsp, +, 8]
 mov     rcx, qword, ptr, [rsp, +, 16]
 lea     rdx, [rax, +, 8]
 cmp     rdx, rcx
 jbe     .LBB10_6
.LBB10_4:
 mov     rdi, rsp
 mov     esi, 8
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rax, qword, ptr, [rsp, +, 8]
 jmp     .LBB10_6
.LBB10_7:
 mov     rdi, rsp
 mov     esi, 4
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rax, qword, ptr, [rsp, +, 8]
 jmp     .LBB10_9
.LBB10_29:
 mov     rdi, rsp
 mov     esi, 9
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer6Buffer16reserve_internal17h3a4aa4ca459d2714E@GOTPCREL]
 mov     rcx, qword, ptr, [rsp, +, 8]
 jmp     .LBB10_31
.LBB10_39:
 jmp     .LBB10_40
.LBB10_38:
 jmp     .LBB10_40
.LBB10_37:
.LBB10_40:
 mov     rbx, rax
 mov     rdi, rsp
 call    core::ptr::drop_in_place
 mov     rdi, rbx
 call    _Unwind_Resume
 ud2
@Kogia-sima Kogia-sima added Type: Bug Bug or Bug fixes help wanted Extra attention is needed Type: Performance Performance improvement and removed Type: Bug Bug or Bug fixes labels Jul 10, 2020
@Kogia-sima

This comment has been minimized.

Sign in to view
This was referenced Jul 21, 2020
Closed
Closed
@Kogia-sima
Copy link
Member Author

I found that enhanced for-loop optimization could cause significant performance drop (~40%) in big-table benchmark. That makes me feel even more confused.

Previous compiler output

{
    __sf_rt::render_text!(__sf_buf, "<table>\n");
    {
        __sf_rt::render_text!(__sf_buf, "<tr>");
        for row in table {
            {
                __sf_rt::render_text!(__sf_buf, "<td>");
                for col in row {
                    __sf_rt::render_escaped!(__sf_buf, *col);
                    __sf_rt::render_text!(__sf_buf, "</td><td>");
                }
                unsafe {
                    __sf_buf._set_len(__sf_buf.len() - 4usize);
                }
            }
            __sf_rt::render_text!(__sf_buf, "</tr><tr>");
        }
        unsafe {
            __sf_buf._set_len(__sf_buf.len() - 4usize);
        }
    }
    __sf_rt::render_text!(__sf_buf, "\n</table>");
}

perf/join-after-optimize branch (~40% slower)

{
    __sf_rt::render_text!(__sf_buf, "<table>\n<tr><td>");
    for row in table {
        for col in row {
            __sf_rt::render_escaped!(__sf_buf, *col);
            __sf_rt::render_text!(__sf_buf, "</td><td>");
        }
        unsafe {
            __sf_buf._set_len(__sf_buf.len() - 4usize);
        }
        __sf_rt::render_text!(__sf_buf, "</tr><tr><td>");
    }
    unsafe {
        __sf_buf._set_len(__sf_buf.len() - 8usize);
    }
    __sf_rt::render_text!(__sf_buf, "\n</table>");
}
Full Assembly Output 
all::big_table:
 push    rbp
 push    r15
 push    r14
 push    r13
 push    r12
 push    rbx
 sub     rsp, 120
 mov     qword, ptr, [rsp, +, 88], rdi
 mov     r12d, 8
 mov     edi, 2400
 mov     esi, 8
 call    qword, ptr, [rip, +, __rust_alloc@GOTPCREL]
 test    rax, rax
 je      .LBB20_1
 mov     qword, ptr, [rsp, +, 40], rax
 vmovaps xmm0, xmmword, ptr, [rip, +, .LCPI20_0]
 vmovups xmmword, ptr, [rsp, +, 48], xmm0
 mov     r13d, 100
 mov     ebp, 4
 xor     r15d, r15d
 jmp     .LBB20_49
.LBB20_52:
 mov     rcx, qword, ptr, [rsp, +, 40]
 mov     r14, qword, ptr, [rsp, +, 32]
.LBB20_53:
 lea     rax, [r15, +, 2*r15]
 mov     rdx, qword, ptr, [rsp, +, 16]
 mov     qword, ptr, [rcx, +, 8*rax, +, 16], rdx
 vmovaps xmm0, xmmword, ptr, [rsp]
 vmovups xmmword, ptr, [rcx, +, 8*rax], xmm0
 mov     r15, r14
 cmp     r14d, 100
 je      .LBB20_3
.LBB20_49:
 mov     edi, 800
 mov     esi, 8
 call    qword, ptr, [rip, +, __rust_alloc@GOTPCREL]
 test    rax, rax
 je      .LBB20_50
 mov     rdi, rax
 lea     rax, [r15, +, 1]
 mov     qword, ptr, [rsp, +, 32], rax
 mov     qword, ptr, [rsp, +, 64], rdi
 vmovaps xmm0, xmmword, ptr, [rip, +, .LCPI20_0]
 vmovups xmmword, ptr, [rsp, +, 72], xmm0
 mov     esi, 100
 xor     r14d, r14d
 jmp     .LBB20_71
.LBB20_75:
 mov     rdx, rcx
 mov     rcx, rbx
 call    qword, ptr, [rip, +, __rust_realloc@GOTPCREL]
.LBB20_79:
 mov     rcx, rax
.LBB20_80:
 test    rcx, rcx
 je      .LBB20_84
 mov     qword, ptr, [rsp, +, 64], rcx
 shr     rbx, 3
 mov     rdi, rcx
 mov     rsi, rbx
.LBB20_82:
 mov     qword, ptr, [rdi, +, 8*r14], r14
 lea     rax, [r14, +, 1]
 mov     r14, rax
 cmp     rax, 100
 je      .LBB20_51
.LBB20_71:
 cmp     r14, rsi
 jne     .LBB20_82
 lea     rax, [rsi, +, 1]
 lea     rcx, [rsi, +, rsi]
 cmp     rcx, rax
 cmova   rax, rcx
 cmp     rax, 4
 cmovbe  rax, rbp
 xor     ecx, ecx
 mul     r12
 mov     rbx, rax
 seto    al
 setno   cl
 shl     rcx, 3
 test    rsi, rsi
 je      .LBB20_76
 test    al, al
 jne     .LBB20_83
 shl     rsi, 3
 test    rsi, rsi
 jne     .LBB20_75
 jmp     .LBB20_77
.LBB20_76:
 test    al, al
 jne     .LBB20_83
.LBB20_77:
 test    rbx, rbx
 je      .LBB20_80
 mov     rdi, rbx
 mov     rsi, rcx
 call    qword, ptr, [rip, +, __rust_alloc@GOTPCREL]
 jmp     .LBB20_79
.LBB20_51:
 mov     qword, ptr, [rsp, +, 80], 100
 mov     qword, ptr, [rsp, +, 72], rsi
 mov     rax, qword, ptr, [rsp, +, 64]
 mov     qword, ptr, [rsp], rax
 mov     rax, qword, ptr, [rsp, +, 72]
 mov     qword, ptr, [rsp, +, 8], rax
 mov     qword, ptr, [rsp, +, 16], 100
 cmp     r15, r13
 jne     .LBB20_52
 mov     rax, r13
 inc     rax
 mov     r14, qword, ptr, [rsp, +, 32]
 je      .LBB20_55
 mov     rcx, r13
 add     rcx, r13
 cmp     rcx, rax
 cmova   rax, rcx
 cmp     rax, 4
 cmovbe  rax, rbp
 xor     ecx, ecx
 mov     edx, 24
 mul     rdx
 mov     rbx, rax
 seto    al
 setno   cl
 shl     rcx, 3
 test    r13, r13
 je      .LBB20_63
 test    al, al
 jne     .LBB20_56
 mov     rdi, qword, ptr, [rsp, +, 40]
 test    rdi, rdi
 je      .LBB20_61
 shl     r13, 3
 lea     rsi, [2*r13]
 add     rsi, r13
 test    rsi, rsi
 je      .LBB20_61
 mov     rdx, rcx
 mov     rcx, rbx
 call    qword, ptr, [rip, +, __rust_realloc@GOTPCREL]
 jmp     .LBB20_67
.LBB20_63:
 test    al, al
 jne     .LBB20_64
.LBB20_61:
 test    rbx, rbx
 je      .LBB20_68
 mov     rdi, rbx
 mov     rsi, rcx
 call    qword, ptr, [rip, +, __rust_alloc@GOTPCREL]
.LBB20_67:
 mov     rcx, rax
.LBB20_68:
 test    rcx, rcx
 je      .LBB20_84
 mov     qword, ptr, [rsp, +, 40], rcx
 mov     rdx, rbx
 movabs  rax, -6148914691236517205
 mulx    r13, r13, rax
 shr     r13, 4
 jmp     .LBB20_53
.LBB20_3:
 mov     qword, ptr, [rsp, +, 56], r14
 mov     qword, ptr, [rsp, +, 48], r13
 call    qword, ptr, [rip, +, _ZN3std4time7Instant3now17hb624d5f02fff610dE@GOTPCREL]
 mov     qword, ptr, [rsp, +, 64], rax
 mov     qword, ptr, [rsp, +, 72], rdx
 mov     rax, qword, ptr, [rsp, +, 88]
 mov     rax, qword, ptr, [rax]
 mov     qword, ptr, [rsp, +, 104], rax
 test    rax, rax
 je      .LBB20_85
 xor     ebp, ebp
.LBB20_6:
 mov     qword, ptr, [rsp], 1
 lea     rax, [rsp, +, 8]
 vxorps  xmm0, xmm0, xmm0
 vmovups xmmword, ptr, [rax], xmm0
 mov     r15, qword, ptr, [rsp, +, 40]
 mov     r12, qword, ptr, [rip, +, _ZN56_$LT$all..BigTable$u20$as$u20$sailfish..TemplateOnce$GT$11render_once9SIZE_HINT17h7471bb5c60182121E]
 mov     rcx, qword, ptr, [rsp, +, 8]
 mov     rsi, qword, ptr, [rsp, +, 16]
 mov     rdi, rsi
 sub     rdi, rcx
 cmp     rdi, r12
 jae     .LBB20_7
 lea     rax, [rsi, +, rsi]
 add     r12, rsi
 cmp     rax, r12
 cmova   r12, rax
 mov     rdi, qword, ptr, [rsp]
 mov     rdx, r12
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer12safe_realloc17hefa6669a8d0a53bfE@GOTPCREL]
 mov     qword, ptr, [rsp], rax
 mov     qword, ptr, [rsp, +, 16], r12
 mov     rcx, qword, ptr, [rsp, +, 8]
 mov     rdi, r12
 sub     rdi, rcx
 cmp     rdi, 16
 jb      .LBB20_16
.LBB20_15:
 mov     rax, qword, ptr, [rsp]
 jmp     .LBB20_18
.LBB20_7:
 mov     r12, rsi
 cmp     rdi, 16
 jae     .LBB20_15
.LBB20_16:
 lea     rax, [r12, +, r12]
 lea     rbx, [r12, +, 16]
 cmp     rax, rbx
 cmova   rbx, rax
 mov     rdi, qword, ptr, [rsp]
 mov     rsi, r12
 mov     rdx, rbx
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer12safe_realloc17hefa6669a8d0a53bfE@GOTPCREL]
 mov     qword, ptr, [rsp], rax
 mov     qword, ptr, [rsp, +, 16], rbx
 mov     rcx, qword, ptr, [rsp, +, 8]
 mov     r12, rbx
.LBB20_18:
 mov     qword, ptr, [rsp, +, 96], rbp
 vmovups xmm0, xmmword, ptr, [rip, +, .L__unnamed_2]
 vmovups xmmword, ptr, [rax, +, rcx], xmm0
 add     rcx, 16
 mov     qword, ptr, [rsp, +, 8], rcx
 test    r14, r14
 je      .LBB20_19
 lea     rdx, [r14, +, 2*r14]
 mov     rsi, r15
 lea     rdx, [r15, +, 8*rdx]
 mov     qword, ptr, [rsp, +, 112], rdx
 jmp     .LBB20_23
.LBB20_41:
 mov     r15, r12
 mov     rsi, qword, ptr, [rsp, +, 32]
.LBB20_44:
 add     rsi, 24
 mov     rdx, qword, ptr, [rip, +, .L__unnamed_3+5]
 mov     qword, ptr, [rax, +, rcx, +, 5], rdx
 mov     rdx, qword, ptr, [rip, +, .L__unnamed_3]
 mov     qword, ptr, [rax, +, rcx], rdx
 add     rcx, 13
 mov     qword, ptr, [rsp, +, 8], rcx
 cmp     rsi, qword, ptr, [rsp, +, 112]
 je      .LBB20_20
.LBB20_23:
 mov     r13, qword, ptr, [rsi, +, 16]
 test    r13, r13
 mov     qword, ptr, [rsp, +, 32], rsi
 je      .LBB20_40
 mov     rbp, qword, ptr, [rsi]
 shl     r13, 3
 xor     r14d, r14d
 jmp     .LBB20_25
.LBB20_30:
 mov     rax, qword, ptr, [rsp]
.LBB20_47:
 movzx   edx, byte, ptr, [rip, +, .L__unnamed_4+8]
 mov     byte, ptr, [rax, +, rcx, +, 8], dl
 mov     rdx, qword, ptr, [rip, +, .L__unnamed_4]
 mov     qword, ptr, [rax, +, rcx], rdx
 add     rcx, 9
 mov     qword, ptr, [rsp, +, 8], rcx
 add     r14, 8
 cmp     r13, r14
 je      .LBB20_40
.LBB20_25:
 mov     r15, qword, ptr, [rbp, +, r14]
 lea     rdx, [rcx, +, 20]
 cmp     rdx, r12
 jbe     .LBB20_28
 lea     rcx, [r12, +, r12]
 lea     rbx, [r12, +, 20]
 cmp     rcx, rbx
 cmova   rbx, rcx
 mov     rdi, rax
 mov     rsi, r12
 mov     rdx, rbx
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer12safe_realloc17hefa6669a8d0a53bfE@GOTPCREL]
 mov     qword, ptr, [rsp], rax
 mov     qword, ptr, [rsp, +, 16], rbx
 mov     rcx, qword, ptr, [rsp, +, 8]
.LBB20_28:
 add     rax, rcx
 mov     rdi, r15
 mov     rsi, rax
 call    qword, ptr, [rip, +, _ZN5itoap9write_u6417hb594a3d8c0958d6cE@GOTPCREL]
 mov     rcx, rax
 add     rcx, qword, ptr, [rsp, +, 8]
 mov     qword, ptr, [rsp, +, 8], rcx
 mov     r12, qword, ptr, [rsp, +, 16]
 mov     rax, r12
 sub     rax, rcx
 cmp     rax, 9
 jae     .LBB20_30
 lea     rax, [r12, +, r12]
 lea     rbx, [r12, +, 9]
 cmp     rax, rbx
 cmova   rbx, rax
 mov     rdi, qword, ptr, [rsp]
 mov     rsi, r12
 mov     rdx, rbx
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer12safe_realloc17hefa6669a8d0a53bfE@GOTPCREL]
 mov     qword, ptr, [rsp], rax
 mov     qword, ptr, [rsp, +, 16], rbx
 mov     rcx, qword, ptr, [rsp, +, 8]
 mov     r12, rbx
 jmp     .LBB20_47
.LBB20_40:
 add     rcx, -4
 mov     qword, ptr, [rsp, +, 8], rcx
 mov     rdx, r12
 sub     rdx, rcx
 cmp     rdx, 12
 ja      .LBB20_41
 lea     rcx, [r12, +, r12]
 lea     r15, [r12, +, 13]
 cmp     rcx, r15
 cmova   r15, rcx
 mov     rdi, rax
 mov     rsi, r12
 mov     rdx, r15
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer12safe_realloc17hefa6669a8d0a53bfE@GOTPCREL]
 mov     rsi, qword, ptr, [rsp, +, 32]
 mov     qword, ptr, [rsp], rax
 mov     qword, ptr, [rsp, +, 16], r15
 mov     rcx, qword, ptr, [rsp, +, 8]
 mov     r12, r15
 jmp     .LBB20_44
.LBB20_19:
 mov     r15, r12
.LBB20_20:
 add     rcx, -8
 mov     qword, ptr, [rsp, +, 8], rcx
 mov     rdx, r15
 sub     rdx, rcx
 cmp     rdx, 8
 ja      .LBB20_21
 lea     rcx, [r15, +, r15]
 lea     rbx, [r15, +, 9]
 cmp     rcx, rbx
 cmova   rbx, rcx
 mov     rdi, rax
 mov     rsi, r15
 mov     rdx, rbx
 call    qword, ptr, [rip, +, _ZN8sailfish7runtime6buffer12safe_realloc17hefa6669a8d0a53bfE@GOTPCREL]
 mov     rbp, qword, ptr, [rsp, +, 96]
 mov     qword, ptr, [rsp], rax
 mov     qword, ptr, [rsp, +, 16], rbx
 mov     rcx, qword, ptr, [rsp, +, 8]
 jmp     .LBB20_33
.LBB20_21:
 mov     rbp, qword, ptr, [rsp, +, 96]
.LBB20_33:
 mov     dl, byte, ptr, [rip, +, .L__unnamed_5+8]
 mov     byte, ptr, [rax, +, rcx, +, 8], dl
 mov     rdx, qword, ptr, [rip, +, .L__unnamed_5]
 mov     qword, ptr, [rax, +, rcx], rdx
 lea     rax, [rcx, +, 9]
 mov     qword, ptr, [rsp, +, 8], rax
 shr     rax, 3
 add     rax, rcx
 add     rax, 84
 mov     rcx, qword, ptr, [rip, +, _ZN56_$LT$all..BigTable$u20$as$u20$sailfish..TemplateOnce$GT$11render_once9SIZE_HINT17h7471bb5c60182121E]
 cmp     rcx, rax
 jae     .LBB20_35
 mov     qword, ptr, [rip, +, _ZN56_$LT$all..BigTable$u20$as$u20$sailfish..TemplateOnce$GT$11render_once9SIZE_HINT17h7471bb5c60182121E], rax
.LBB20_35:
 inc     rbp
 mov     rax, qword, ptr, [rsp]
 vpermilps xmm0, xmmword, ptr, [rsp, +, 8], 78
 mov     qword, ptr, [rsp], rax
 vmovups xmmword, ptr, [rsp, +, 8], xmm0
 mov     rsi, qword, ptr, [rsp, +, 8]
 mov     rax, qword, ptr, [rsp, +, 16]
 mov     rdi, qword, ptr, [rsp]
 test    rdi, rdi
 je      .LBB20_38
 test    rsi, rsi
 je      .LBB20_38
 mov     edx, 1
 call    qword, ptr, [rip, +, __rust_dealloc@GOTPCREL]
.LBB20_38:
 cmp     rbp, qword, ptr, [rsp, +, 104]
 je      .LBB20_85
 mov     r14, qword, ptr, [rsp, +, 56]
 jmp     .LBB20_6
.LBB20_85:
 lea     rdi, [rsp, +, 64]
 call    qword, ptr, [rip, +, _ZN3std4time7Instant7elapsed17h0b88ba2cb63ee684E@GOTPCREL]
 mov     rcx, qword, ptr, [rsp, +, 88]
 mov     qword, ptr, [rcx, +, 8], rax
 mov     dword, ptr, [rcx, +, 16], edx
 mov     r14, qword, ptr, [rsp, +, 40]
 mov     rax, qword, ptr, [rsp, +, 56]
 test    rax, rax
 je      .LBB20_93
 lea     rax, [rax, +, 2*rax]
 lea     rbp, [r14, +, 8*rax]
 mov     r15, qword, ptr, [rip, +, __rust_dealloc@GOTPCREL]
 mov     rbx, r14
 jmp     .LBB20_88
.LBB20_92:
 add     rbx, 24
 cmp     rbx, rbp
 je      .LBB20_93
.LBB20_88:
 mov     rsi, qword, ptr, [rbx, +, 8]
 test    rsi, rsi
 je      .LBB20_92
 mov     rdi, qword, ptr, [rbx]
 test    rdi, rdi
 je      .LBB20_92
 shl     rsi, 3
 test    rsi, rsi
 je      .LBB20_92
 mov     edx, 8
 call    r15
 jmp     .LBB20_92
.LBB20_93:
 mov     rax, qword, ptr, [rsp, +, 48]
 test    rax, rax
 je      .LBB20_96
 shl     rax, 3
 lea     rsi, [rax, +, 2*rax]
 test    rsi, rsi
 je      .LBB20_96
 mov     edx, 8
 mov     rdi, r14
 call    qword, ptr, [rip, +, __rust_dealloc@GOTPCREL]
.LBB20_96:
 add     rsp, 120
 pop     rbx
 pop     r12
 pop     r13
 pop     r14
 pop     r15
 pop     rbp
 ret
.LBB20_84:
 mov     esi, 8
 mov     rdi, rbx
 call    qword, ptr, [rip, +, _ZN5alloc5alloc18handle_alloc_error17hb12bf6f6b6f1927aE@GOTPCREL]
 ud2
.LBB20_83:
 mov     qword, ptr, [rsp, +, 56], r15
 mov     qword, ptr, [rsp, +, 48], r13
 mov     qword, ptr, [rsp, +, 80], rsi
 mov     qword, ptr, [rsp, +, 72], rsi
 call    qword, ptr, [rip, +, _ZN5alloc7raw_vec17capacity_overflow17h35f3dcdf1e879e98E@GOTPCREL]
.LBB20_57:
 ud2
.LBB20_50:
 mov     edi, 800
 mov     esi, 8
 call    qword, ptr, [rip, +, _ZN5alloc5alloc18handle_alloc_error17hb12bf6f6b6f1927aE@GOTPCREL]
 ud2
.LBB20_55:
 mov     r13, -1
 jmp     .LBB20_56
.LBB20_64:
 xor     r13d, r13d
.LBB20_56:
 mov     qword, ptr, [rsp, +, 56], r13
 mov     qword, ptr, [rsp, +, 48], r13
 call    qword, ptr, [rip, +, _ZN5alloc7raw_vec17capacity_overflow17h35f3dcdf1e879e98E@GOTPCREL]
 jmp     .LBB20_57
.LBB20_1:
 mov     edi, 2400
 mov     esi, 8
 call    qword, ptr, [rip, +, _ZN5alloc5alloc18handle_alloc_error17hb12bf6f6b6f1927aE@GOTPCREL]
 ud2
.LBB20_100:
 mov     rbx, rax
 mov     rdi, rsp
 call    core::ptr::drop_in_place
 jmp     .LBB20_99
.LBB20_97:
 mov     rbx, rax
 lea     rdi, [rsp, +, 64]
 call    core::ptr::drop_in_place
 jmp     .LBB20_99
.LBB20_10:
 jmp     .LBB20_11
.LBB20_9:
 jmp     .LBB20_11
.LBB20_8:
.LBB20_11:
 mov     rbx, rax
 mov     rdi, rsp
 call    core::ptr::drop_in_place
.LBB20_99:
 lea     rdi, [rsp, +, 40]
 call    core::ptr::drop_in_place
 mov     rdi, rbx
 call    _Unwind_Resume
 ud2
.LBB20_98:
 mov     rbx, rax
 jmp     .LBB20_99

@Kogia-sima Kogia-sima changed the title Strange behaviour about Buffer::reserve_internal Strange behavior in big table benchmark Dec 18, 2020
@zzau13
Copy link
Contributor

zzau13 commented Dec 18, 2020

.LBB10_16: is a loop starts. r12 set to qword, ptr, [rip, +, _ZN5itoap9write_u6417hc82e35bf780e3b54E@GOTPCREL] to save n additions rip + .... and memory gets.

@Kogia-sima
Copy link
Member Author

Kogia-sima commented Dec 19, 2020 •
edited
Loading

@botika This isn't that simple. Here are the objdump results for master and perf/join-after-optimize branch

master branch 
00000000000167f0 <<all::BigTable as sailfish::TemplateOnce>::render_once>:
   167f0:	55                   	push   rbp
   167f1:	41 57                	push   r15
   167f3:	41 56                	push   r14
   167f5:	41 55                	push   r13
   167f7:	41 54                	push   r12
   167f9:	53                   	push   rbx
   167fa:	48 83 ec 38          	sub    rsp,0x38
   167fe:	48 c7 44 24 08 01 00 	mov    QWORD PTR [rsp+0x8],0x1
   16805:	00 00 
   16807:	c5 f8 57 c0          	vxorps xmm0,xmm0,xmm0
   1680b:	c5 f8 11 44 24 10    	vmovups XMMWORD PTR [rsp+0x10],xmm0
   16811:	49 89 d7             	mov    r15,rdx
   16814:	48 89 f5             	mov    rbp,rsi
   16817:	49 89 fe             	mov    r14,rdi
   1681a:	4c 8b 25 87 0e 04 00 	mov    r12,QWORD PTR [rip+0x40e87]        # 576a8 <<all::BigTable as sailfish::TemplateOnce>::render_once::SIZE_HINT>
   16821:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16826:	48 8b 74 24 18       	mov    rsi,QWORD PTR [rsp+0x18]
   1682b:	48 89 f0             	mov    rax,rsi
   1682e:	48 29 c8             	sub    rax,rcx
   16831:	4c 39 e0             	cmp    rax,r12
   16834:	73 3e                	jae    16874 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x84>
   16836:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   1683a:	49 01 f4             	add    r12,rsi
   1683d:	4c 39 e0             	cmp    rax,r12
   16840:	4c 0f 47 e0          	cmova  r12,rax
   16844:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   16849:	4c 89 e2             	mov    rdx,r12
   1684c:	ff 15 ae f6 03 00    	call   QWORD PTR [rip+0x3f6ae]        # 55f00 <_DYNAMIC+0x240>
   16852:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16857:	4c 89 64 24 18       	mov    QWORD PTR [rsp+0x18],r12
   1685c:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16861:	4c 89 e0             	mov    rax,r12
   16864:	48 29 c8             	sub    rax,rcx
   16867:	48 83 f8 08          	cmp    rax,0x8
   1686b:	72 10                	jb     1687d <<all::BigTable as sailfish::TemplateOnce>::render_once+0x8d>
   1686d:	48 8b 44 24 08       	mov    rax,QWORD PTR [rsp+0x8]
   16872:	eb 39                	jmp    168ad <<all::BigTable as sailfish::TemplateOnce>::render_once+0xbd>
   16874:	49 89 f4             	mov    r12,rsi
   16877:	48 83 f8 08          	cmp    rax,0x8
   1687b:	73 f0                	jae    1686d <<all::BigTable as sailfish::TemplateOnce>::render_once+0x7d>
   1687d:	4b 8d 04 24          	lea    rax,[r12+r12*1]
   16881:	49 8d 5c 24 08       	lea    rbx,[r12+0x8]
   16886:	48 39 d8             	cmp    rax,rbx
   16889:	48 0f 47 d8          	cmova  rbx,rax
   1688d:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   16892:	4c 89 e6             	mov    rsi,r12
   16895:	48 89 da             	mov    rdx,rbx
   16898:	ff 15 62 f6 03 00    	call   QWORD PTR [rip+0x3f662]        # 55f00 <_DYNAMIC+0x240>
   1689e:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   168a3:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   168a8:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   168ad:	48 ba 3c 74 61 62 6c 	movabs rdx,0xa3e656c6261743c
   168b4:	65 3e 0a 
   168b7:	48 89 14 08          	mov    QWORD PTR [rax+rcx*1],rdx
   168bb:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   168c0:	48 8b 74 24 18       	mov    rsi,QWORD PTR [rsp+0x18]
   168c5:	48 83 c1 08          	add    rcx,0x8
   168c9:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   168ce:	48 89 f0             	mov    rax,rsi
   168d1:	48 29 c8             	sub    rax,rcx
   168d4:	48 83 f8 04          	cmp    rax,0x4
   168d8:	73 2e                	jae    16908 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x118>
   168da:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   168de:	48 8d 5e 04          	lea    rbx,[rsi+0x4]
   168e2:	48 39 d8             	cmp    rax,rbx
   168e5:	48 0f 47 d8          	cmova  rbx,rax
   168e9:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   168ee:	48 89 da             	mov    rdx,rbx
   168f1:	ff 15 09 f6 03 00    	call   QWORD PTR [rip+0x3f609]        # 55f00 <_DYNAMIC+0x240>
   168f7:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   168fc:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   16901:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16906:	eb 05                	jmp    1690d <<all::BigTable as sailfish::TemplateOnce>::render_once+0x11d>
   16908:	48 8b 44 24 08       	mov    rax,QWORD PTR [rsp+0x8]
   1690d:	4c 89 74 24 20       	mov    QWORD PTR [rsp+0x20],r14
   16912:	c7 04 08 3c 74 72 3e 	mov    DWORD PTR [rax+rcx*1],0x3e72743c
   16919:	48 8b 44 24 10       	mov    rax,QWORD PTR [rsp+0x10]
   1691e:	48 8b 74 24 18       	mov    rsi,QWORD PTR [rsp+0x18]
   16923:	48 83 c0 04          	add    rax,0x4
   16927:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   1692c:	49 c1 e7 03          	shl    r15,0x3
   16930:	4b 8d 0c 7f          	lea    rcx,[r15+r15*2]
   16934:	48 85 c9             	test   rcx,rcx
   16937:	0f 84 20 02 00 00    	je     16b5d <<all::BigTable as sailfish::TemplateOnce>::render_once+0x36d>
   1693d:	48 01 e9             	add    rcx,rbp
   16940:	48 89 4c 24 28       	mov    QWORD PTR [rsp+0x28],rcx
   16945:	e9 40 00 00 00       	jmp    1698a <<all::BigTable as sailfish::TemplateOnce>::render_once+0x19a>
   1694a:	66 0f 1f 44 00 00    	nop    WORD PTR [rax+rax*1+0x0]
   16950:	48 8b 4c 24 08       	mov    rcx,QWORD PTR [rsp+0x8]
   16955:	48 8b 6c 24 30       	mov    rbp,QWORD PTR [rsp+0x30]
   1695a:	48 8b 7c 24 28       	mov    rdi,QWORD PTR [rsp+0x28]
   1695f:	48 83 c5 18          	add    rbp,0x18
   16963:	8a 15 5a 1d ff ff    	mov    dl,BYTE PTR [rip+0xffffffffffff1d5a]        # 86c3 <GCC_except_table28+0x3af>
   16969:	88 54 01 08          	mov    BYTE PTR [rcx+rax*1+0x8],dl
   1696d:	48 8b 15 47 1d ff ff 	mov    rdx,QWORD PTR [rip+0xffffffffffff1d47]        # 86bb <GCC_except_table28+0x3a7>
   16974:	48 89 14 01          	mov    QWORD PTR [rcx+rax*1],rdx
   16978:	48 83 c0 09          	add    rax,0x9
   1697c:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   16981:	48 39 fd             	cmp    rbp,rdi
   16984:	0f 84 d3 01 00 00    	je     16b5d <<all::BigTable as sailfish::TemplateOnce>::render_once+0x36d>
   1698a:	48 89 f1             	mov    rcx,rsi
   1698d:	48 29 c1             	sub    rcx,rax
   16990:	48 83 f9 04          	cmp    rcx,0x4
   16994:	73 3a                	jae    169d0 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x1e0>
   16996:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   1699a:	48 8d 5e 04          	lea    rbx,[rsi+0x4]
   1699e:	48 39 d8             	cmp    rax,rbx
   169a1:	48 0f 47 d8          	cmova  rbx,rax
   169a5:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   169aa:	48 89 da             	mov    rdx,rbx
   169ad:	ff 15 4d f5 03 00    	call   QWORD PTR [rip+0x3f54d]        # 55f00 <_DYNAMIC+0x240>
   169b3:	48 89 c1             	mov    rcx,rax
   169b6:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   169bb:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   169c0:	48 8b 44 24 10       	mov    rax,QWORD PTR [rsp+0x10]
   169c5:	e9 0b 00 00 00       	jmp    169d5 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x1e5>
   169ca:	66 0f 1f 44 00 00    	nop    WORD PTR [rax+rax*1+0x0]
   169d0:	48 8b 4c 24 08       	mov    rcx,QWORD PTR [rsp+0x8]
   169d5:	c7 04 01 3c 74 64 3e 	mov    DWORD PTR [rcx+rax*1],0x3e64743c
   169dc:	48 8b 44 24 10       	mov    rax,QWORD PTR [rsp+0x10]
   169e1:	48 8b 74 24 18       	mov    rsi,QWORD PTR [rsp+0x18]
   169e6:	48 83 c0 04          	add    rax,0x4
   169ea:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   169ef:	4c 8b 65 10          	mov    r12,QWORD PTR [rbp+0x10]
   169f3:	4d 85 e4             	test   r12,r12
   169f6:	48 89 6c 24 30       	mov    QWORD PTR [rsp+0x30],rbp
   169fb:	0f 84 ff 00 00 00    	je     16b00 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x310>
   16a01:	4c 8b 6d 00          	mov    r13,QWORD PTR [rbp+0x0]
   16a05:	49 c1 e4 03          	shl    r12,0x3
   16a09:	45 31 f6             	xor    r14d,r14d
   16a0c:	48 8b 2d ed f4 03 00 	mov    rbp,QWORD PTR [rip+0x3f4ed]        # 55f00 <_DYNAMIC+0x240>
   16a13:	e9 39 00 00 00       	jmp    16a51 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x261>
   16a18:	0f 1f 84 00 00 00 00 	nop    DWORD PTR [rax+rax*1+0x0]
   16a1f:	00 
   16a20:	48 8b 4c 24 08       	mov    rcx,QWORD PTR [rsp+0x8]
   16a25:	0f b6 15 8e 1c ff ff 	movzx  edx,BYTE PTR [rip+0xffffffffffff1c8e]        # 86ba <GCC_except_table28+0x3a6>
   16a2c:	88 54 01 08          	mov    BYTE PTR [rcx+rax*1+0x8],dl
   16a30:	48 8b 15 7b 1c ff ff 	mov    rdx,QWORD PTR [rip+0xffffffffffff1c7b]        # 86b2 <GCC_except_table28+0x39e>
   16a37:	48 89 14 01          	mov    QWORD PTR [rcx+rax*1],rdx
   16a3b:	48 83 c0 09          	add    rax,0x9
   16a3f:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   16a44:	49 83 c6 08          	add    r14,0x8
   16a48:	4d 39 f4             	cmp    r12,r14
   16a4b:	0f 84 b6 00 00 00    	je     16b07 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x317>
   16a51:	4f 8b 7c 35 00       	mov    r15,QWORD PTR [r13+r14*1+0x0]
   16a56:	48 8d 48 14          	lea    rcx,[rax+0x14]
   16a5a:	48 39 f1             	cmp    rcx,rsi
   16a5d:	76 31                	jbe    16a90 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x2a0>
   16a5f:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   16a63:	48 8d 5e 14          	lea    rbx,[rsi+0x14]
   16a67:	48 39 d8             	cmp    rax,rbx
   16a6a:	48 0f 47 d8          	cmova  rbx,rax
   16a6e:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   16a73:	48 89 da             	mov    rdx,rbx
   16a76:	ff d5                	call   rbp
   16a78:	48 89 c6             	mov    rsi,rax
   16a7b:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16a80:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   16a85:	48 8b 44 24 10       	mov    rax,QWORD PTR [rsp+0x10]
   16a8a:	e9 06 00 00 00       	jmp    16a95 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x2a5>
   16a8f:	90                   	nop
   16a90:	48 8b 74 24 08       	mov    rsi,QWORD PTR [rsp+0x8]
   16a95:	48 01 c6             	add    rsi,rax
   16a98:	4c 89 ff             	mov    rdi,r15
   16a9b:	ff 15 67 f4 03 00    	call   QWORD PTR [rip+0x3f467]        # 55f08 <_DYNAMIC+0x248>
   16aa1:	48 03 44 24 10       	add    rax,QWORD PTR [rsp+0x10]
   16aa6:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   16aab:	48 8b 74 24 18       	mov    rsi,QWORD PTR [rsp+0x18]
   16ab0:	48 89 f1             	mov    rcx,rsi
   16ab3:	48 29 c1             	sub    rcx,rax
   16ab6:	48 83 f9 09          	cmp    rcx,0x9
   16aba:	0f 83 60 ff ff ff    	jae    16a20 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x230>
   16ac0:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   16ac4:	48 8d 5e 09          	lea    rbx,[rsi+0x9]
   16ac8:	48 39 d8             	cmp    rax,rbx
   16acb:	48 0f 47 d8          	cmova  rbx,rax
   16acf:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   16ad4:	48 89 da             	mov    rdx,rbx
   16ad7:	ff d5                	call   rbp
   16ad9:	48 89 c1             	mov    rcx,rax
   16adc:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16ae1:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   16ae6:	48 8b 44 24 10       	mov    rax,QWORD PTR [rsp+0x10]
   16aeb:	48 89 de             	mov    rsi,rbx
   16aee:	e9 32 ff ff ff       	jmp    16a25 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x235>
   16af3:	66 66 66 66 2e 0f 1f 	data16 data16 data16 nop WORD PTR cs:[rax+rax*1+0x0]
   16afa:	84 00 00 00 00 00 
   16b00:	48 8b 2d f9 f3 03 00 	mov    rbp,QWORD PTR [rip+0x3f3f9]        # 55f00 <_DYNAMIC+0x240>
   16b07:	48 83 c0 fc          	add    rax,0xfffffffffffffffc
   16b0b:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   16b10:	48 89 f1             	mov    rcx,rsi
   16b13:	48 29 c1             	sub    rcx,rax
   16b16:	48 83 f9 09          	cmp    rcx,0x9
   16b1a:	0f 83 30 fe ff ff    	jae    16950 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x160>
   16b20:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   16b24:	48 8d 5e 09          	lea    rbx,[rsi+0x9]
   16b28:	48 39 d8             	cmp    rax,rbx
   16b2b:	48 0f 47 d8          	cmova  rbx,rax
   16b2f:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   16b34:	48 89 da             	mov    rdx,rbx
   16b37:	ff d5                	call   rbp
   16b39:	48 8b 6c 24 30       	mov    rbp,QWORD PTR [rsp+0x30]
   16b3e:	48 8b 7c 24 28       	mov    rdi,QWORD PTR [rsp+0x28]
   16b43:	48 89 c1             	mov    rcx,rax
   16b46:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16b4b:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   16b50:	48 8b 44 24 10       	mov    rax,QWORD PTR [rsp+0x10]
   16b55:	48 89 de             	mov    rsi,rbx
   16b58:	e9 02 fe ff ff       	jmp    1695f <<all::BigTable as sailfish::TemplateOnce>::render_once+0x16f>
   16b5d:	48 83 c0 fc          	add    rax,0xfffffffffffffffc
   16b61:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   16b66:	48 89 f1             	mov    rcx,rsi
   16b69:	48 29 c1             	sub    rcx,rax
   16b6c:	48 83 f9 09          	cmp    rcx,0x9
   16b70:	73 36                	jae    16ba8 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x3b8>
   16b72:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   16b76:	48 8d 5e 09          	lea    rbx,[rsi+0x9]
   16b7a:	48 39 d8             	cmp    rax,rbx
   16b7d:	48 0f 47 d8          	cmova  rbx,rax
   16b81:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   16b86:	48 89 da             	mov    rdx,rbx
   16b89:	ff 15 71 f3 03 00    	call   QWORD PTR [rip+0x3f371]        # 55f00 <_DYNAMIC+0x240>
   16b8f:	48 8b 74 24 20       	mov    rsi,QWORD PTR [rsp+0x20]
   16b94:	48 89 c1             	mov    rcx,rax
   16b97:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16b9c:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   16ba1:	48 8b 44 24 10       	mov    rax,QWORD PTR [rsp+0x10]
   16ba6:	eb 0a                	jmp    16bb2 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x3c2>
   16ba8:	48 8b 4c 24 08       	mov    rcx,QWORD PTR [rsp+0x8]
   16bad:	48 8b 74 24 20       	mov    rsi,QWORD PTR [rsp+0x20]
   16bb2:	8a 15 14 1b ff ff    	mov    dl,BYTE PTR [rip+0xffffffffffff1b14]        # 86cc <GCC_except_table28+0x3b8>
   16bb8:	88 54 01 08          	mov    BYTE PTR [rcx+rax*1+0x8],dl
   16bbc:	48 8b 15 01 1b ff ff 	mov    rdx,QWORD PTR [rip+0xffffffffffff1b01]        # 86c4 <GCC_except_table28+0x3b0>
   16bc3:	48 89 14 01          	mov    QWORD PTR [rcx+rax*1],rdx
   16bc7:	48 8d 48 09          	lea    rcx,[rax+0x9]
   16bcb:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   16bd0:	48 c1 e9 03          	shr    rcx,0x3
   16bd4:	48 01 c8             	add    rax,rcx
   16bd7:	48 83 c0 54          	add    rax,0x54
   16bdb:	48 8b 0d c6 0a 04 00 	mov    rcx,QWORD PTR [rip+0x40ac6]        # 576a8 <<all::BigTable as sailfish::TemplateOnce>::render_once::SIZE_HINT>
   16be2:	48 39 c1             	cmp    rcx,rax
   16be5:	73 07                	jae    16bee <<all::BigTable as sailfish::TemplateOnce>::render_once+0x3fe>
   16be7:	48 89 05 ba 0a 04 00 	mov    QWORD PTR [rip+0x40aba],rax        # 576a8 <<all::BigTable as sailfish::TemplateOnce>::render_once::SIZE_HINT>
   16bee:	48 8b 44 24 08       	mov    rax,QWORD PTR [rsp+0x8]
   16bf3:	c4 e3 79 04 44 24 10 	vpermilps xmm0,XMMWORD PTR [rsp+0x10],0x4e
   16bfa:	4e 
   16bfb:	48 89 46 08          	mov    QWORD PTR [rsi+0x8],rax
   16bff:	c5 f8 11 46 10       	vmovups XMMWORD PTR [rsi+0x10],xmm0
   16c04:	48 c7 06 00 00 00 00 	mov    QWORD PTR [rsi],0x0
   16c0b:	48 89 f0             	mov    rax,rsi
   16c0e:	48 83 c4 38          	add    rsp,0x38
   16c12:	5b                   	pop    rbx
   16c13:	41 5c                	pop    r12
   16c15:	41 5d                	pop    r13
   16c17:	41 5e                	pop    r14
   16c19:	41 5f                	pop    r15
   16c1b:	5d                   	pop    rbp
   16c1c:	c3                   	ret    
   16c1d:	eb 02                	jmp    16c21 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x431>
   16c1f:	eb 00                	jmp    16c21 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x431>
   16c21:	48 89 c3             	mov    rbx,rax
   16c24:	48 8d 7c 24 08       	lea    rdi,[rsp+0x8]
   16c29:	e8 72 f5 ff ff       	call   161a0 <core::ptr::drop_in_place>
   16c2e:	48 89 df             	mov    rdi,rbx
   16c31:	e8 1a b7 03 00       	call   52350 <_Unwind_Resume@plt>
   16c36:	0f 0b                	ud2    
   16c38:	cc                   	int3   
   16c39:	cc                   	int3   
   16c3a:	cc                   	int3   
   16c3b:	cc                   	int3   
   16c3c:	cc                   	int3   
   16c3d:	cc                   	int3   
   16c3e:	cc                   	int3   
   16c3f:	cc                   	int3
perf/join-after-optimize branch 
00000000000167d0 <<all::BigTable as sailfish::TemplateOnce>::render_once>:
   167d0:	55                   	push   rbp
   167d1:	41 57                	push   r15
   167d3:	41 56                	push   r14
   167d5:	41 55                	push   r13
   167d7:	41 54                	push   r12
   167d9:	53                   	push   rbx
   167da:	48 83 ec 38          	sub    rsp,0x38
   167de:	48 c7 44 24 08 01 00 	mov    QWORD PTR [rsp+0x8],0x1
   167e5:	00 00 
   167e7:	c5 f8 57 c0          	vxorps xmm0,xmm0,xmm0
   167eb:	c5 f8 11 44 24 10    	vmovups XMMWORD PTR [rsp+0x10],xmm0
   167f1:	49 89 d4             	mov    r12,rdx
   167f4:	48 89 f5             	mov    rbp,rsi
   167f7:	48 89 7c 24 28       	mov    QWORD PTR [rsp+0x28],rdi
   167fc:	4c 8b 2d 45 0d 04 00 	mov    r13,QWORD PTR [rip+0x40d45]        # 57548 <<all::BigTable as sailfish::TemplateOnce>::render_once::SIZE_HINT>
   16803:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16808:	48 8b 74 24 18       	mov    rsi,QWORD PTR [rsp+0x18]
   1680d:	48 89 f0             	mov    rax,rsi
   16810:	48 29 c8             	sub    rax,rcx
   16813:	4c 39 e8             	cmp    rax,r13
   16816:	73 3e                	jae    16856 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x86>
   16818:	48 8d 04 36          	lea    rax,[rsi+rsi*1]
   1681c:	49 01 f5             	add    r13,rsi
   1681f:	4c 39 e8             	cmp    rax,r13
   16822:	4c 0f 47 e8          	cmova  r13,rax
   16826:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   1682b:	4c 89 ea             	mov    rdx,r13
   1682e:	ff 15 6c f5 03 00    	call   QWORD PTR [rip+0x3f56c]        # 55da0 <_DYNAMIC+0x240>
   16834:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16839:	4c 89 6c 24 18       	mov    QWORD PTR [rsp+0x18],r13
   1683e:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16843:	4c 89 e8             	mov    rax,r13
   16846:	48 29 c8             	sub    rax,rcx
   16849:	48 83 f8 10          	cmp    rax,0x10
   1684d:	72 10                	jb     1685f <<all::BigTable as sailfish::TemplateOnce>::render_once+0x8f>
   1684f:	48 8b 44 24 08       	mov    rax,QWORD PTR [rsp+0x8]
   16854:	eb 3d                	jmp    16893 <<all::BigTable as sailfish::TemplateOnce>::render_once+0xc3>
   16856:	49 89 f5             	mov    r13,rsi
   16859:	48 83 f8 10          	cmp    rax,0x10
   1685d:	73 f0                	jae    1684f <<all::BigTable as sailfish::TemplateOnce>::render_once+0x7f>
   1685f:	4c 89 e8             	mov    rax,r13
   16862:	4c 01 e8             	add    rax,r13
   16865:	49 8d 5d 10          	lea    rbx,[r13+0x10]
   16869:	48 39 d8             	cmp    rax,rbx
   1686c:	48 0f 47 d8          	cmova  rbx,rax
   16870:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   16875:	4c 89 ee             	mov    rsi,r13
   16878:	48 89 da             	mov    rdx,rbx
   1687b:	ff 15 1f f5 03 00    	call   QWORD PTR [rip+0x3f51f]        # 55da0 <_DYNAMIC+0x240>
   16881:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16886:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   1688b:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16890:	49 89 dd             	mov    r13,rbx
   16893:	c5 f8 10 05 05 1c ff 	vmovups xmm0,XMMWORD PTR [rip+0xffffffffffff1c05]        # 84a0 <GCC_except_table28+0x190>
   1689a:	ff 
   1689b:	c5 f8 11 04 08       	vmovups XMMWORD PTR [rax+rcx*1],xmm0
   168a0:	48 83 c1 10          	add    rcx,0x10
   168a4:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   168a9:	49 c1 e4 03          	shl    r12,0x3
   168ad:	4b 8d 14 64          	lea    rdx,[r12+r12*2]
   168b1:	48 85 d2             	test   rdx,rdx
   168b4:	0f 84 ab 01 00 00    	je     16a65 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x295>
   168ba:	48 01 ea             	add    rdx,rbp
   168bd:	48 89 54 24 30       	mov    QWORD PTR [rsp+0x30],rdx
   168c2:	e9 43 00 00 00       	jmp    1690a <<all::BigTable as sailfish::TemplateOnce>::render_once+0x13a>
   168c7:	66 0f 1f 84 00 00 00 	nop    WORD PTR [rax+rax*1+0x0]
   168ce:	00 00 
   168d0:	4d 89 ec             	mov    r12,r13
   168d3:	48 8b 6c 24 20       	mov    rbp,QWORD PTR [rsp+0x20]
   168d8:	48 83 c5 18          	add    rbp,0x18
   168dc:	48 8b 15 cd 1d ff ff 	mov    rdx,QWORD PTR [rip+0xffffffffffff1dcd]        # 86b0 <GCC_except_table28+0x3a0>
   168e3:	48 89 54 08 05       	mov    QWORD PTR [rax+rcx*1+0x5],rdx
   168e8:	48 8b 15 bc 1d ff ff 	mov    rdx,QWORD PTR [rip+0xffffffffffff1dbc]        # 86ab <GCC_except_table28+0x39b>
   168ef:	48 89 14 08          	mov    QWORD PTR [rax+rcx*1],rdx
   168f3:	48 83 c1 0d          	add    rcx,0xd
   168f7:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   168fc:	48 8b 54 24 30       	mov    rdx,QWORD PTR [rsp+0x30]
   16901:	48 39 d5             	cmp    rbp,rdx
   16904:	0f 84 5e 01 00 00    	je     16a68 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x298>
   1690a:	4c 8b 7d 10          	mov    r15,QWORD PTR [rbp+0x10]
   1690e:	4d 85 ff             	test   r15,r15
   16911:	48 89 6c 24 20       	mov    QWORD PTR [rsp+0x20],rbp
   16916:	0f 84 f4 00 00 00    	je     16a10 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x240>
   1691c:	48 8b 6d 00          	mov    rbp,QWORD PTR [rbp+0x0]
   16920:	49 c1 e7 03          	shl    r15,0x3
   16924:	45 31 f6             	xor    r14d,r14d
   16927:	e9 35 00 00 00       	jmp    16961 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x191>
   1692c:	0f 1f 40 00          	nop    DWORD PTR [rax+0x0]
   16930:	48 8b 44 24 08       	mov    rax,QWORD PTR [rsp+0x8]
   16935:	0f b6 15 6e 1d ff ff 	movzx  edx,BYTE PTR [rip+0xffffffffffff1d6e]        # 86aa <GCC_except_table28+0x39a>
   1693c:	88 54 08 08          	mov    BYTE PTR [rax+rcx*1+0x8],dl
   16940:	48 8b 15 5b 1d ff ff 	mov    rdx,QWORD PTR [rip+0xffffffffffff1d5b]        # 86a2 <GCC_except_table28+0x392>
   16947:	48 89 14 08          	mov    QWORD PTR [rax+rcx*1],rdx
   1694b:	48 83 c1 09          	add    rcx,0x9
   1694f:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   16954:	49 83 c6 08          	add    r14,0x8
   16958:	4d 39 f7             	cmp    r15,r14
   1695b:	0f 84 af 00 00 00    	je     16a10 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x240>
   16961:	4e 8b 64 35 00       	mov    r12,QWORD PTR [rbp+r14*1+0x0]
   16966:	48 8d 51 14          	lea    rdx,[rcx+0x14]
   1696a:	4c 39 ea             	cmp    rdx,r13
   1696d:	76 2f                	jbe    1699e <<all::BigTable as sailfish::TemplateOnce>::render_once+0x1ce>
   1696f:	4c 89 e9             	mov    rcx,r13
   16972:	4c 01 e9             	add    rcx,r13
   16975:	49 8d 5d 14          	lea    rbx,[r13+0x14]
   16979:	48 39 d9             	cmp    rcx,rbx
   1697c:	48 0f 47 d9          	cmova  rbx,rcx
   16980:	48 89 c7             	mov    rdi,rax
   16983:	4c 89 ee             	mov    rsi,r13
   16986:	48 89 da             	mov    rdx,rbx
   16989:	ff 15 11 f4 03 00    	call   QWORD PTR [rip+0x3f411]        # 55da0 <_DYNAMIC+0x240>
   1698f:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16994:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   16999:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   1699e:	48 01 c8             	add    rax,rcx
   169a1:	4c 89 e7             	mov    rdi,r12
   169a4:	48 89 c6             	mov    rsi,rax
   169a7:	ff 15 fb f3 03 00    	call   QWORD PTR [rip+0x3f3fb]        # 55da8 <_DYNAMIC+0x248>
   169ad:	48 89 c1             	mov    rcx,rax
   169b0:	48 03 4c 24 10       	add    rcx,QWORD PTR [rsp+0x10]
   169b5:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   169ba:	4c 8b 6c 24 18       	mov    r13,QWORD PTR [rsp+0x18]
   169bf:	4c 89 e8             	mov    rax,r13
   169c2:	48 29 c8             	sub    rax,rcx
   169c5:	48 83 f8 09          	cmp    rax,0x9
   169c9:	0f 83 61 ff ff ff    	jae    16930 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x160>
   169cf:	4c 89 e8             	mov    rax,r13
   169d2:	4c 01 e8             	add    rax,r13
   169d5:	49 8d 5d 09          	lea    rbx,[r13+0x9]
   169d9:	48 39 d8             	cmp    rax,rbx
   169dc:	48 0f 47 d8          	cmova  rbx,rax
   169e0:	48 8b 7c 24 08       	mov    rdi,QWORD PTR [rsp+0x8]
   169e5:	4c 89 ee             	mov    rsi,r13
   169e8:	48 89 da             	mov    rdx,rbx
   169eb:	ff 15 af f3 03 00    	call   QWORD PTR [rip+0x3f3af]        # 55da0 <_DYNAMIC+0x240>
   169f1:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   169f6:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   169fb:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16a00:	49 89 dd             	mov    r13,rbx
   16a03:	e9 2d ff ff ff       	jmp    16935 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x165>
   16a08:	0f 1f 84 00 00 00 00 	nop    DWORD PTR [rax+rax*1+0x0]
   16a0f:	00 
   16a10:	48 83 c1 fc          	add    rcx,0xfffffffffffffffc
   16a14:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   16a19:	4c 89 ea             	mov    rdx,r13
   16a1c:	48 29 ca             	sub    rdx,rcx
   16a1f:	48 83 fa 0c          	cmp    rdx,0xc
   16a23:	0f 87 a7 fe ff ff    	ja     168d0 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x100>
   16a29:	4c 89 e9             	mov    rcx,r13
   16a2c:	4c 01 e9             	add    rcx,r13
   16a2f:	4d 8d 65 0d          	lea    r12,[r13+0xd]
   16a33:	4c 39 e1             	cmp    rcx,r12
   16a36:	4c 0f 47 e1          	cmova  r12,rcx
   16a3a:	48 89 c7             	mov    rdi,rax
   16a3d:	4c 89 ee             	mov    rsi,r13
   16a40:	4c 89 e2             	mov    rdx,r12
   16a43:	ff 15 57 f3 03 00    	call   QWORD PTR [rip+0x3f357]        # 55da0 <_DYNAMIC+0x240>
   16a49:	48 8b 6c 24 20       	mov    rbp,QWORD PTR [rsp+0x20]
   16a4e:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16a53:	4c 89 64 24 18       	mov    QWORD PTR [rsp+0x18],r12
   16a58:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16a5d:	4d 89 e5             	mov    r13,r12
   16a60:	e9 73 fe ff ff       	jmp    168d8 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x108>
   16a65:	4d 89 ec             	mov    r12,r13
   16a68:	48 83 c1 f8          	add    rcx,0xfffffffffffffff8
   16a6c:	48 89 4c 24 10       	mov    QWORD PTR [rsp+0x10],rcx
   16a71:	4c 89 e2             	mov    rdx,r12
   16a74:	48 29 ca             	sub    rdx,rcx
   16a77:	48 83 fa 08          	cmp    rdx,0x8
   16a7b:	77 2e                	ja     16aab <<all::BigTable as sailfish::TemplateOnce>::render_once+0x2db>
   16a7d:	4b 8d 0c 24          	lea    rcx,[r12+r12*1]
   16a81:	49 8d 5c 24 09       	lea    rbx,[r12+0x9]
   16a86:	48 39 d9             	cmp    rcx,rbx
   16a89:	48 0f 47 d9          	cmova  rbx,rcx
   16a8d:	48 89 c7             	mov    rdi,rax
   16a90:	4c 89 e6             	mov    rsi,r12
   16a93:	48 89 da             	mov    rdx,rbx
   16a96:	ff 15 04 f3 03 00    	call   QWORD PTR [rip+0x3f304]        # 55da0 <_DYNAMIC+0x240>
   16a9c:	48 89 44 24 08       	mov    QWORD PTR [rsp+0x8],rax
   16aa1:	48 89 5c 24 18       	mov    QWORD PTR [rsp+0x18],rbx
   16aa6:	48 8b 4c 24 10       	mov    rcx,QWORD PTR [rsp+0x10]
   16aab:	8a 15 0f 1c ff ff    	mov    dl,BYTE PTR [rip+0xffffffffffff1c0f]        # 86c0 <GCC_except_table28+0x3b0>
   16ab1:	88 54 08 08          	mov    BYTE PTR [rax+rcx*1+0x8],dl
   16ab5:	48 8b 15 fc 1b ff ff 	mov    rdx,QWORD PTR [rip+0xffffffffffff1bfc]        # 86b8 <GCC_except_table28+0x3a8>
   16abc:	48 89 14 08          	mov    QWORD PTR [rax+rcx*1],rdx
   16ac0:	48 8d 41 09          	lea    rax,[rcx+0x9]
   16ac4:	48 89 44 24 10       	mov    QWORD PTR [rsp+0x10],rax
   16ac9:	48 c1 e8 03          	shr    rax,0x3
   16acd:	48 01 c8             	add    rax,rcx
   16ad0:	48 83 c0 54          	add    rax,0x54
   16ad4:	48 8b 0d 6d 0a 04 00 	mov    rcx,QWORD PTR [rip+0x40a6d]        # 57548 <<all::BigTable as sailfish::TemplateOnce>::render_once::SIZE_HINT>
   16adb:	48 39 c1             	cmp    rcx,rax
   16ade:	73 07                	jae    16ae7 <<all::BigTable as sailfish::TemplateOnce>::render_once+0x317>
   16ae0:	48 89 05 61 0a 04 00 	mov    QWORD PTR [rip+0x40a61],rax        # 57548 <<all::BigTable as sailfish::TemplateOnce>::render_once::SIZE_HINT>
   16ae7:	48 8b 4c 24 08       	mov    rcx,QWORD PTR [rsp+0x8]
   16aec:	c4 e3 79 04 44 24 10 	vpermilps xmm0,XMMWORD PTR [rsp+0x10],0x4e
   16af3:	4e 
   16af4:	48 8b 44 24 28       	mov    rax,QWORD PTR [rsp+0x28]
   16af9:	48 89 48 08          	mov    QWORD PTR [rax+0x8],rcx
   16afd:	c5 f8 11 40 10       	vmovups XMMWORD PTR [rax+0x10],xmm0
   16b02:	48 c7 00 00 00 00 00 	mov    QWORD PTR [rax],0x0
   16b09:	48 83 c4 38          	add    rsp,0x38
   16b0d:	5b                   	pop    rbx
   16b0e:	41 5c                	pop    r12
   16b10:	41 5d                	pop    r13
   16b12:	41 5e                	pop    r14
   16b14:	41 5f                	pop    r15
   16b16:	5d                   	pop    rbp
   16b17:	c3                   	ret    
   16b18:	eb 02                	jmp    16b1c <<all::BigTable as sailfish::TemplateOnce>::render_once+0x34c>
   16b1a:	eb 00                	jmp    16b1c <<all::BigTable as sailfish::TemplateOnce>::render_once+0x34c>
   16b1c:	48 89 c3             	mov    rbx,rax
   16b1f:	48 8d 7c 24 08       	lea    rdi,[rsp+0x8]
   16b24:	e8 57 f6 ff ff       	call   16180 <core::ptr::drop_in_place>
   16b29:	48 89 df             	mov    rdi,rbx
   16b2c:	e8 bf b6 03 00       	call   521f0 <_Unwind_Resume@plt>
   16b31:	0f 0b                	ud2    
   16b33:	cc                   	int3   
   16b34:	cc                   	int3   
   16b35:	cc                   	int3   
   16b36:	cc                   	int3   
   16b37:	cc                   	int3   
   16b38:	cc                   	int3   
   16b39:	cc                   	int3   
   16b3a:	cc                   	int3   
   16b3b:	cc                   	int3   
   16b3c:	cc                   	int3   
   16b3d:	cc                   	int3   
   16b3e:	cc                   	int3   
   16b3f:	cc                   	int3

I see that both branch call write_u64 function via pointer offset from the content of rip register (at 16a9b and 169a7 respectively).

@Kogia-sima
Copy link
Member Author

Note: no significant difference in Teams benchmark

@zzau13
Copy link
Contributor

zzau13 commented Dec 19, 2020

Yes, that is so easy, ask in Rust or in llvm.
Why is writeu64 put in r12 and called from this registry?

@zzau13
Copy link
Contributor

zzau13 commented Dec 19, 2020

Raise the benchmarks that the world sees, if so democracy works.

@Kogia-sima
Copy link
Member Author

There is some concern about a licence in big table benchmark. We need another example

https://choosealicense.com/no-permission/

@zzau13
Copy link
Contributor

zzau13 commented Dec 20, 2020

I do not understand what you mean that is not compatible with the License. Can't use the big table with MIT or APACHE License? I did not understand you.

On the other hand, I have made a new lexer for you that can tokenize any block language (handlebars, jinja, ..). I am now with the AST. And with that the i18n at compile time. After that goes the html5 parser and webpack. And once here the DOM is ready to build anything.
Yarte will do the same as this and it is faster and I think it is impossible to do it faster. It has a compilation-time evaluator v_eval, meta-programming, compilation-time errors, derive, proc_macro, nothing has a cost and it is extensible to infinity.

We could now have the hot module replace implemented instead of reinventing the wheel. And it would be from the team, neither yours nor mine. I don't understand it.
Nothing feel the veneer. I've already done everything I could. goodbye! 😸

@Kogia-sima
Copy link
Member Author

Big table benchmark is distributed under No Lincense. That's why I removed the benchmark from this repository.

https://github.com/djc/template-benchmarks-rs

@zzau13
Copy link
Contributor

zzau13 commented Dec 21, 2020

https://pythonhosted.org/wheezy.template/examples.html Yes of course ... the benchmarks too, everything, the ifs, the else minus the design patterns that don't seem to do them. Good grief if Euler raised his head.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
help wanted Extra attention is needed Type: Performance Performance improvement
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@zzau13 @Kogia-sima