[X64] [EgorBo] Small clean up in gtMarkAddrMode #474
Comments
Top method regressions8 (0.93 % of base) - System.Collections.Frozen.FrozenHashTable:CalcNumBuckets(System.ReadOnlySpan`1[int],ubyte):int ; Assembly listing for method System.Collections.Frozen.FrozenHashTable:CalcNumBuckets(System.ReadOnlySpan`1[int],ubyte):int (FullOpts)
; Emitting BLENDED_CODE for X64 with AVX512 - Unix
; FullOpts code
; optimized code
; rbp based frame
; fully interruptible
; No PGO data
; 0 inlinees with PGO data; 11 single block inlinees; 5 inlinees without PGO data
; Final local variable assignments
;
;* V00 arg0 [V00 ] ( 0, 0 ) struct (16) zero-ref multireg-arg ld-addr-op single-def <System.ReadOnlySpan`1[int]>
; V01 arg1 [V01,T35] ( 3, 3 ) ubyte -> rdx single-def
;* V02 loc0 [V02 ] ( 0, 0 ) struct (24) zero-ref ld-addr-op <System.Collections.Frozen.FrozenHashTable+<>c__DisplayClass10_0>
; V03 loc1 [V03,T28] ( 7, 12.50) ref -> r14 class-hnd <System.Collections.Generic.HashSet`1[int]>
; V04 loc2 [V04,T29] ( 9, 8.50) int -> r13
; V05 loc3 [V05,T36] ( 2, 5 ) int -> rdi
;* V06 loc4 [V06 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op <System.ReadOnlySpan`1[int]>
; V07 loc5 [V07,T37] ( 4, 3 ) int -> [rbp-0x2C] spill-single-def
; V08 loc6 [V08,T33] ( 6, 6.50) int -> [rbp-0x30]
; V09 loc7 [V09,T32] ( 8, 7.50) int -> [rbp-0x34]
; V10 loc8 [V10,T40] ( 3, 3 ) int -> [rbp-0x38]
;* V11 loc9 [V11 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op <System.ReadOnlySpan`1[int]>
; V12 loc10 [V12,T00] ( 10,656.50) int -> registers
; V13 loc11 [V13,T31] ( 2, 8 ) int -> rsi
; V14 loc12 [V14,T45] ( 2, 1 ) int -> rdi
; V15 loc13 [V15 ] ( 6, 40 ) struct (24) [rbp-0x50] do-not-enreg[XS] must-init addr-exposed ld-addr-op <System.Collections.Generic.HashSet`1+Enumerator[int]>
;# V16 OutArgs [V16 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;* V17 tmp1 [V17 ] ( 0, 0 ) int -> zero-ref
;* V18 tmp2 [V18 ] ( 0, 0 ) int -> zero-ref
; V19 tmp3 [V19,T46] ( 2, 1 ) int -> rcx
;* V20 tmp4 [V20 ] ( 0, 0 ) ref -> zero-ref class-hnd "impAppendStmt" <int[]>
; V21 tmp5 [V21,T39] ( 3, 3 ) ref -> r14 class-hnd exact single-def "NewObj constructor temp" <System.Collections.Generic.HashSet`1[int]>
;* V22 tmp6 [V22 ] ( 0, 0 ) int -> zero-ref "Inlining Arg"
; V23 tmp7 [V23 ] ( 1, 4 ) int -> [rbp-0x58] do-not-enreg[X] addr-exposed ld-addr-op "Inline ldloca(s) first use temp"
;* V24 tmp8 [V24 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[int]>
; V25 tmp9 [V25,T30] ( 3, 8 ) int -> rdx "Inline return value spill temp"
;* V26 tmp10 [V26 ] ( 0, 0 ) int -> zero-ref "Inlining Arg"
; V27 tmp11 [V27,T24] ( 3, 20 ) int -> rdx "Inlining Arg"
;* V28 tmp12 [V28 ] ( 0, 0 ) struct (24) zero-ref ld-addr-op "NewObj constructor temp" <System.Collections.Generic.HashSet`1+Enumerator[int]>
;* V29 tmp13 [V29,T22] ( 0, 0 ) ubyte -> zero-ref "Inline return value spill temp"
; V30 tmp14 [V30,T16] ( 2, 64 ) int -> rcx "Inlining Arg"
; V31 tmp15 [V31,T18] ( 3, 48 ) int -> rdx "Inline stloc first use temp"
; V32 tmp16 [V32,T17] ( 3, 48 ) byref -> rax "dup spill"
;* V33 tmp17 [V33 ] ( 0, 0 ) int -> zero-ref "Inline stloc first use temp"
;* V34 tmp18 [V34,T10] ( 0, 0 ) ubyte -> zero-ref "Inline return value spill temp"
;* V35 tmp19 [V35 ] ( 0, 0 ) int -> zero-ref "Inlining Arg"
; V36 tmp20 [V36,T02] ( 3,384 ) int -> rdx "Inline stloc first use temp"
; V37 tmp21 [V37,T09] ( 3,192 ) byref -> rdx "dup spill"
;* V38 tmp22 [V38 ] ( 0, 0 ) int -> zero-ref "Inline stloc first use temp"
; V39 tmp23 [V39,T07] ( 3,133 ) byref -> r15 single-def "field V00._reference (fldOffset=0x0)" P-INDEP
; V40 tmp24 [V40,T01] ( 9,265.50) int -> rbx single-def "field V00._length (fldOffset=0x8)" P-INDEP
; V41 tmp25 [V41,T08] ( 8,193 ) ref -> [rbp-0x78] spill-single-def "field V02.seenBuckets (fldOffset=0x0)" P-INDEP
; V42 tmp26 [V42,T11] ( 6,156 ) int -> [rbp-0x5C] spill-single-def "field V02.numBuckets (fldOffset=0x8)" P-INDEP
; V43 tmp27 [V43,T04] ( 10,312 ) int -> [rbp-0x60] "field V02.numCollisions (fldOffset=0xc)" P-INDEP
; V44 tmp28 [V44,T13] ( 5,146.50) int -> [rbp-0x64] "field V02.bestNumCollisions (fldOffset=0x10)" P-INDEP
; V45 tmp29 [V45,T26] ( 5, 13.50) byref -> r12 single-def "field V06._reference (fldOffset=0x0)" P-INDEP
;* V46 tmp30 [V46,T47] ( 0, 0 ) int -> zero-ref "field V06._length (fldOffset=0x8)" P-INDEP
;* V47 tmp31 [V47,T14] ( 0, 0 ) byref -> zero-ref "field V11._reference (fldOffset=0x0)" P-INDEP
;* V48 tmp32 [V48,T15] ( 0, 0 ) int -> zero-ref "field V11._length (fldOffset=0x8)" P-INDEP
; V49 tmp33 [V49 ] ( 2, 18 ) ref -> [rbp-0x50] do-not-enreg[X] addr-exposed "field V15._hashSet (fldOffset=0x0)" P-DEP
; V50 tmp34 [V50 ] ( 2, 18 ) int -> [rbp-0x48] do-not-enreg[X] addr-exposed "field V15._version (fldOffset=0x8)" P-DEP
; V51 tmp35 [V51 ] ( 2, 18 ) int -> [rbp-0x44] do-not-enreg[X] addr-exposed "field V15._index (fldOffset=0xc)" P-DEP
; V52 tmp36 [V52 ] ( 3, 34 ) int -> [rbp-0x40] do-not-enreg[X] addr-exposed "field V15._current (fldOffset=0x10)" P-DEP
; V53 tmp37 [V53,T41] ( 2, 2 ) byref -> r12 single-def "field V24._reference (fldOffset=0x0)" P-INDEP
;* V54 tmp38 [V54,T48] ( 0, 0 ) int -> zero-ref "field V24._length (fldOffset=0x8)" P-INDEP
; V55 tmp39 [V55,T34] ( 3, 6 ) ref -> rdi "field V28._hashSet (fldOffset=0x0)" P-INDEP
; V56 tmp40 [V56,T38] ( 2, 4 ) int -> rcx "field V28._version (fldOffset=0x8)" P-INDEP
;* V57 tmp41 [V57,T42] ( 0, 0 ) int -> zero-ref "field V28._index (fldOffset=0xc)" P-INDEP
;* V58 tmp42 [V58,T43] ( 0, 0 ) int -> zero-ref "field V28._current (fldOffset=0x10)" P-INDEP
-; V59 cse0 [V59,T05] ( 3,288 ) long -> r9 "CSE #09: aggressive"
-; V60 cse1 [V60,T20] ( 3, 40 ) long -> rax "CSE #04: moderate"
-; V61 cse2 [V61,T03] ( 3,384 ) int -> rdi "CSE #08: aggressive"
+; V59 cse0 [V59,T05] ( 3,288 ) long -> r9 "CSE #11: aggressive"
+; V60 cse1 [V60,T20] ( 3, 40 ) long -> rax "CSE #05: moderate"
+; V61 cse2 [V61,T03] ( 3,384 ) int -> rdi "CSE #09: aggressive"
; V62 cse3 [V62,T19] ( 3, 48 ) int -> rax "CSE #03: moderate"
-; V63 cse4 [V63,T06] ( 3,288 ) int -> r10 "CSE #12: aggressive"
-; V64 cse5 [V64,T21] ( 3, 40 ) int -> rdx "CSE #07: moderate"
-; V65 cse6 [V65,T44] ( 3, 1.50) ref -> [rbp-0x80] spill-single-def "CSE #13: conservative"
+; V63 cse4 [V63,T06] ( 3,288 ) int -> r10 "CSE #14: aggressive"
+; V64 cse5 [V64,T21] ( 3, 40 ) int -> rdx "CSE #08: moderate"
+; V65 cse6 [V65,T44] ( 3, 1.50) ref -> [rbp-0x80] spill-single-def "CSE #15: conservative"
; V66 cse7 [V66,T12] ( 4,152 ) int -> [rbp-0x68] spill-single-def "CSE #02: moderate"
; V67 rat0 [V67,T25] ( 6, 18 ) long -> rax "Widened IV V07"
; V68 rat1 [V68,T27] ( 6, 13 ) long -> rdx "Widened IV V09"
; V69 rat2 [V69,T23] ( 7, 21 ) long -> [rbp-0x70] "Widened IV V14"
;
; Lcl frame size = 88
G_M18874_IG01:
push rbp
push r15
push r14
push r13
push r12
push rbx
sub rsp, 88
lea rbp, [rsp+0x80]
vxorps xmm8, xmm8, xmm8
vmovdqa xmmword ptr [rbp-0x50], xmm8
xor eax, eax
mov qword ptr [rbp-0x40], rax
mov r15, rdi
mov ebx, esi
;; size=43 bbWeight=1 PerfScore 10.83
G_M18874_IG02:
xor r14, r14
mov r13d, ebx
test dl, dl
jne SHORT G_M18874_IG07
;; size=10 bbWeight=1 PerfScore 1.75
G_M18874_IG03:
mov rdi, 0xD1FFAB1E ; System.Collections.Generic.HashSet`1[int]
call CORINFO_HELP_NEWSFAST
mov r14, rax
test ebx, ebx
jle SHORT G_M18874_IG04
mov rdi, r14
mov esi, ebx
mov rax, 0xD1FFAB1E ; code for System.Collections.Generic.HashSet`1[int]:Initialize(int):int:this
call [rax]System.Collections.Generic.HashSet`1[int]:Initialize(int):int:this
;; size=39 bbWeight=0.50 PerfScore 3.25
G_M18874_IG04:
xor r12d, r12d
test ebx, ebx
jle SHORT G_M18874_IG06
;; size=7 bbWeight=0.50 PerfScore 0.75
G_M18874_IG05:
mov edx, r12d
mov esi, dword ptr [r15+4*rdx]
lea rdx, [rbp-0x58]
mov rdi, r14
mov rax, 0xD1FFAB1E ; code for System.Collections.Generic.HashSet`1[int]:AddIfNotPresent(int,byref):ubyte:this
call [rax]System.Collections.Generic.HashSet`1[int]:AddIfNotPresent(int,byref):ubyte:this
inc r12d
cmp r12d, ebx
jl SHORT G_M18874_IG05
;; size=34 bbWeight=4 PerfScore 31.00
G_M18874_IG06:
mov r13d, dword ptr [r14+0x28]
sub r13d, dword ptr [r14+0x30]
;; size=8 bbWeight=0.50 PerfScore 2.50
G_M18874_IG07:
lea edi, [r13+r13]
mov r12, 0xD1FFAB1E ; static handle
xor eax, eax
jmp SHORT G_M18874_IG09
;; size=19 bbWeight=1 PerfScore 4.50
G_M18874_IG08:
inc eax
;; size=2 bbWeight=2 PerfScore 0.50
G_M18874_IG09:
cmp eax, 72
jae SHORT G_M18874_IG11
;; size=5 bbWeight=8 PerfScore 10.00
G_M18874_IG10:
cmp dword ptr [r12+4*rax], edi
jl SHORT G_M18874_IG08
;; size=6 bbWeight=4 PerfScore 16.00
G_M18874_IG11:
mov dword ptr [rbp-0x2C], eax
cmp eax, 72
jl SHORT G_M18874_IG14
;; size=8 bbWeight=1 PerfScore 2.25
G_M18874_IG12:
mov edi, r13d
mov rax, 0xD1FFAB1E ; code for System.Collections.HashHelpers:GetPrime(int):int
call [rax]System.Collections.HashHelpers:GetPrime(int):int
nop
;; size=16 bbWeight=0.50 PerfScore 1.88
G_M18874_IG13:
add rsp, 88
pop rbx
pop r12
pop r13
pop r14
pop r15
pop rbp
ret
;; size=15 bbWeight=0.50 PerfScore 2.12
G_M18874_IG14:
mov ecx, 3
mov esi, 16
cmp r13d, 0x3E8
cmovl ecx, esi
imul ecx, r13d
mov edx, eax
jmp SHORT G_M18874_IG16
;; size=28 bbWeight=0.50 PerfScore 2.62
G_M18874_IG15:
inc edx
;; size=2 bbWeight=2 PerfScore 0.50
G_M18874_IG16:
cmp edx, 72
jae SHORT G_M18874_IG17
cmp dword ptr [r12+4*rdx], ecx
jl SHORT G_M18874_IG15
;; size=11 bbWeight=4 PerfScore 21.00
G_M18874_IG17:
cmp edx, 72
jge G_M18874_IG22
lea ecx, [rdx-0x01]
cmp ecx, 72
jae G_M18874_IG46
mov dword ptr [rbp-0x34], edx
lea esi, [rdx-0x01]
mov ecx, dword ptr [r12+4*rsi]
;; size=31 bbWeight=0.50 PerfScore 3.25
G_M18874_IG18:
mov esi, ecx
sar esi, 31
and esi, 31
mov dword ptr [rbp-0x30], ecx
add esi, ecx
sar esi, 5
inc esi
mov rdi, 0xD1FFAB1E ; const ptr
mov r8, gword ptr [rdi]
mov gword ptr [rbp-0x80], r8
mov rdi, r8
mov r9, 0xD1FFAB1E ; code for System.Buffers.SharedArrayPool`1[int]:Rent(int):int[]:this
call [r9]System.Buffers.SharedArrayPool`1[int]:Rent(int):int[]:this
mov gword ptr [rbp-0x78], rax
mov ecx, dword ptr [rbp-0x30]
mov dword ptr [rbp-0x38], ecx
mov dword ptr [rbp-0x64], r13d
mov edi, dword ptr [rbp-0x2C]
mov r9d, edi
mov r10d, dword ptr [rbp-0x34]
cmp r9d, r10d
jge SHORT G_M18874_IG21
;; size=80 bbWeight=0.50 PerfScore 8.62
G_M18874_IG19:
cmp r9d, 72
jae G_M18874_IG46
mov qword ptr [rbp-0x70], r9
mov r11d, dword ptr [r12+4*r9]
mov dword ptr [rbp-0x5C], r11d
mov esi, dword ptr [rax+0x08]
mov dword ptr [rbp-0x68], esi
mov edx, esi
cmp r11d, edx
jle SHORT G_M18874_IG23
;; size=35 bbWeight=4 PerfScore 39.00
G_M18874_IG20:
jmp SHORT G_M18874_IG24
;; size=2 bbWeight=2 PerfScore 4.00
G_M18874_IG21:
mov r11, rax
jmp G_M18874_IG36
align [0 bytes for IG30]
;; size=8 bbWeight=0.25 PerfScore 0.56
G_M18874_IG22:
mov dword ptr [rbp-0x34], edx
jmp G_M18874_IG18
;; size=8 bbWeight=0.25 PerfScore 0.75
G_M18874_IG23:
mov edx, r11d
;; size=3 bbWeight=2 PerfScore 0.50
G_M18874_IG24:
mov rdi, rax
xor esi, esi
mov rcx, 0xD1FFAB1E ; code for System.Array:Clear(System.Array,int,int)
call [rcx]System.Array:Clear(System.Array,int,int)
xor eax, eax
mov dword ptr [rbp-0x60], eax
test r14, r14
je SHORT G_M18874_IG29
;; size=27 bbWeight=4 PerfScore 25.00
G_M18874_IG25:
cmp r13d, ebx
je SHORT G_M18874_IG29
mov rdi, r14
mov ecx, dword ptr [rdi+0x34]
mov gword ptr [rbp-0x50], rdi
mov dword ptr [rbp-0x48], ecx
xor edi, edi
mov dword ptr [rbp-0x44], edi
;; size=23 bbWeight=2 PerfScore 13.50
G_M18874_IG26:
mov dword ptr [rbp-0x40], edi
;; size=3 bbWeight=2 PerfScore 2.00
G_M18874_IG27:
lea rdi, [rbp-0x50]
mov rcx, 0xD1FFAB1E ; code for System.Collections.Generic.HashSet`1+Enumerator[int]:MoveNext():ubyte:this
call [rcx]System.Collections.Generic.HashSet`1+Enumerator[int]:MoveNext():ubyte:this
test eax, eax
jne G_M18874_IG43
;; size=24 bbWeight=16 PerfScore 80.00
G_M18874_IG28:
mov r11, gword ptr [rbp-0x78]
jmp SHORT G_M18874_IG33
;; size=6 bbWeight=8 PerfScore 24.00
G_M18874_IG29:
xor edx, edx
xor ecx, ecx
test ebx, ebx
jle G_M18874_IG39
;; size=12 bbWeight=128 PerfScore 224.00
G_M18874_IG30:
mov edx, ecx
mov eax, dword ptr [r15+4*rdx]
mov r8d, dword ptr [rbp-0x5C]
xor edx, edx
div edx:eax, r8d
mov edi, edx
shr edi, 5
mov eax, dword ptr [rbp-0x68]
cmp edi, eax
jae G_M18874_IG46
mov r9d, edi
+ shl r9, 2
mov edi, 1
shlx r10d, edi, edx
mov r11, gword ptr [rbp-0x78]
- test dword ptr [r11+4*r9+0x10], r10d
+ test dword ptr [r11+r9+0x10], r10d
je G_M18874_IG40
- ;; size=59 bbWeight=128 PerfScore 4800.00
+ ;; size=63 bbWeight=128 PerfScore 4864.00
G_M18874_IG31:
mov edi, dword ptr [rbp-0x60]
inc edi
mov edx, dword ptr [rbp-0x64]
cmp edi, edx
jl G_M18874_IG41
;; size=16 bbWeight=64 PerfScore 224.00
G_M18874_IG32:
mov dword ptr [rbp-0x60], edi
mov dword ptr [rbp-0x64], edx
;; size=6 bbWeight=32 PerfScore 64.00
G_M18874_IG33:
mov eax, dword ptr [rbp-0x60]
mov edx, dword ptr [rbp-0x64]
cmp eax, edx
jge SHORT G_M18874_IG35
;; size=10 bbWeight=64 PerfScore 208.00
G_M18874_IG34:
mov r8d, dword ptr [rbp-0x5C]
mov dword ptr [rbp-0x38], r8d
vxorps xmm0, xmm0, xmm0
vcvtsi2sd xmm0, xmm0, eax
vxorps xmm1, xmm1, xmm1
vcvtsi2sd xmm1, xmm1, r13d
vdivsd xmm0, xmm0, xmm1
vmovsd xmm1, qword ptr [reloc @RWD00]
vucomisd xmm1, xmm0
jae SHORT G_M18874_IG36
mov edx, eax
;; size=45 bbWeight=2 PerfScore 65.83
G_M18874_IG35:
mov rdi, qword ptr [rbp-0x70]
inc edi
mov eax, dword ptr [rbp-0x34]
cmp edi, eax
mov dword ptr [rbp-0x64], edx
mov dword ptr [rbp-0x34], eax
mov r9, rdi
jl SHORT G_M18874_IG38
;; size=22 bbWeight=4 PerfScore 23.00
G_M18874_IG36:
mov rdi, gword ptr [rbp-0x80]
mov rsi, r11
xor edx, edx
mov rax, 0xD1FFAB1E ; code for System.Buffers.SharedArrayPool`1[int]:Return(int[],ubyte):this
call [rax]System.Buffers.SharedArrayPool`1[int]:Return(int[],ubyte):this
mov eax, dword ptr [rbp-0x38]
;; size=24 bbWeight=0.50 PerfScore 2.88
G_M18874_IG37:
add rsp, 88
pop rbx
pop r12
pop r13
pop r14
pop r15
pop rbp
ret
;; size=15 bbWeight=0.50 PerfScore 2.12
G_M18874_IG38:
mov rax, r11
jmp G_M18874_IG19
;; size=8 bbWeight=2 PerfScore 4.50
G_M18874_IG39:
mov r11, gword ptr [rbp-0x78]
jmp G_M18874_IG33
;; size=9 bbWeight=64 PerfScore 192.00
G_M18874_IG40:
- lea rdx, bword ptr [r11+4*r9+0x10]
+ lea rdx, bword ptr [r11+r9+0x10]
or dword ptr [rdx], r10d
mov edx, dword ptr [rbp-0x64]
mov edi, dword ptr [rbp-0x60]
;; size=14 bbWeight=32 PerfScore 192.00
G_M18874_IG41:
inc ecx
cmp ecx, ebx
mov dword ptr [rbp-0x60], edi
mov dword ptr [rbp-0x64], edx
jge G_M18874_IG33
;; size=16 bbWeight=128 PerfScore 448.00
G_M18874_IG42:
jmp G_M18874_IG30
;; size=5 bbWeight=64 PerfScore 128.00
G_M18874_IG43:
mov ecx, dword ptr [rbp-0x40]
mov r8d, dword ptr [rbp-0x5C]
mov eax, ecx
xor edx, edx
div edx:eax, r8d
mov eax, edx
shr eax, 5
mov ecx, dword ptr [rbp-0x68]
cmp eax, ecx
jae SHORT G_M18874_IG46
+ shl rax, 2
mov edi, 1
shlx edx, edi, edx
mov r11, gword ptr [rbp-0x78]
- test dword ptr [r11+4*rax+0x10], edx
+ test dword ptr [r11+rax+0x10], edx
je SHORT G_M18874_IG45
mov edi, dword ptr [rbp-0x60]
inc edi
mov esi, dword ptr [rbp-0x64]
cmp edi, esi
mov dword ptr [rbp-0x60], edi
mov dword ptr [rbp-0x64], esi
jge G_M18874_IG33
- ;; size=69 bbWeight=16 PerfScore 668.00
+ ;; size=73 bbWeight=16 PerfScore 676.00
G_M18874_IG44:
jmp G_M18874_IG27
;; size=5 bbWeight=8 PerfScore 16.00
G_M18874_IG45:
- lea rax, bword ptr [r11+4*rax+0x10]
+ lea rax, bword ptr [r11+rax+0x10]
or dword ptr [rax], edx
jmp G_M18874_IG27
;; size=12 bbWeight=8 PerfScore 48.00
G_M18874_IG46:
call CORINFO_HELP_RNGCHKFAIL
int3
;; size=6 bbWeight=0 PerfScore 0.00
RWD00 dq 3FA999999999999Ah ; 0.05
-; Total bytes of code 856, prolog size 43, PerfScore 7622.98, instruction count 254, allocated bytes for code 856 (MethodHash=a5deb645) for method System.Collections.Frozen.FrozenHashTable:CalcNumBuckets(System.ReadOnlySpan`1[int],ubyte):int (FullOpts)
+; Total bytes of code 864, prolog size 43, PerfScore 7694.98, instruction count 256, allocated bytes for code 864 (MethodHash=a5deb645) for method System.Collections.Frozen.FrozenHashTable:CalcNumBuckets(System.ReadOnlySpan`1[int],ubyte):int (FullOpts)Larger list of diffs: https://gist.github.com/MihuBot/169adbca348f12695d62216234968695 |
Top method improvements-7 (-4.24 % of base) - System.Number+Grisu3:GetCachedPowerForBinaryExponentRange(int,int,byref):System.Number+DiyFp ; Assembly listing for method System.Number+Grisu3:GetCachedPowerForBinaryExponentRange(int,int,byref):System.Number+DiyFp (FullOpts)
; Emitting BLENDED_CODE for X64 with AVX512 - Unix
; FullOpts code
; optimized code
; rsp based frame
; partially interruptible
; No PGO data
; 0 inlinees with PGO data; 4 single block inlinees; 0 inlinees without PGO data
; Final local variable assignments
;
; V00 arg0 [V00,T02] ( 3, 3 ) int -> rdi single-def
;* V01 arg1 [V01 ] ( 0, 0 ) int -> zero-ref single-def
; V02 arg2 [V02,T01] ( 3, 3 ) byref -> rdx single-def
;* V03 loc0 [V03 ] ( 0, 0 ) double -> zero-ref
-; V04 loc1 [V04,T06] ( 3, 3 ) int -> rax
+; V04 loc1 [V04,T05] ( 3, 3 ) int -> rax
;* V05 loc2 [V05 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op <System.ReadOnlySpan`1[short]>
;* V06 loc3 [V06 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op <System.ReadOnlySpan`1[ulong]>
;# V07 OutArgs [V07 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
-; V08 tmp1 [V08,T04] ( 2, 4 ) long -> rdx "impAppendStmt"
+; V08 tmp1 [V08,T03] ( 2, 4 ) long -> rax "impAppendStmt"
;* V09 tmp2 [V09 ] ( 0, 0 ) struct (16) zero-ref multireg-ret ld-addr-op "NewObj constructor temp" <System.Number+DiyFp>
;* V10 tmp3 [V10 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[short]>
;* V11 tmp4 [V11 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[ulong]>
;* V12 tmp5 [V12 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[short]>
-; V13 tmp6 [V13,T05] ( 2, 4 ) int -> rax "Inlining Arg"
+; V13 tmp6 [V13,T04] ( 2, 4 ) int -> rdx "Inlining Arg"
;* V14 tmp7 [V14 ] ( 0, 0 ) byref -> zero-ref "field V05._reference (fldOffset=0x0)" P-INDEP
;* V15 tmp8 [V15 ] ( 0, 0 ) int -> zero-ref "field V05._length (fldOffset=0x8)" P-INDEP
;* V16 tmp9 [V16 ] ( 0, 0 ) byref -> zero-ref single-def "field V06._reference (fldOffset=0x0)" P-INDEP
;* V17 tmp10 [V17 ] ( 0, 0 ) int -> zero-ref "field V06._length (fldOffset=0x8)" P-INDEP
-; V18 tmp11 [V18,T07] ( 2, 2 ) long -> registers "field V09.f (fldOffset=0x0)" P-INDEP
-; V19 tmp12 [V19,T08] ( 2, 2 ) int -> [rsp+0x04] spill-single-def "field V09.e (fldOffset=0x8)" P-INDEP
+; V18 tmp11 [V18,T08] ( 2, 2 ) long -> rax "field V09.f (fldOffset=0x0)" P-INDEP
+; V19 tmp12 [V19,T09] ( 2, 2 ) int -> rdx "field V09.e (fldOffset=0x8)" P-INDEP
;* V20 tmp13 [V20 ] ( 0, 0 ) byref -> zero-ref single-def "field V10._reference (fldOffset=0x0)" P-INDEP
;* V21 tmp14 [V21 ] ( 0, 0 ) int -> zero-ref "field V10._length (fldOffset=0x8)" P-INDEP
;* V22 tmp15 [V22 ] ( 0, 0 ) byref -> zero-ref single-def "field V11._reference (fldOffset=0x0)" P-INDEP
;* V23 tmp16 [V23 ] ( 0, 0 ) int -> zero-ref "field V11._length (fldOffset=0x8)" P-INDEP
;* V24 tmp17 [V24 ] ( 0, 0 ) byref -> zero-ref single-def "field V12._reference (fldOffset=0x0)" P-INDEP
;* V25 tmp18 [V25 ] ( 0, 0 ) int -> zero-ref "field V12._length (fldOffset=0x8)" P-INDEP
-; V26 cse0 [V26,T03] ( 4, 4 ) long -> rax "CSE #01: aggressive"
-; V27 rat0 [V27,T09] ( 3, 6 ) double -> mm0 "ReplaceWithLclVar is creating a new local variable"
-; V28 rat1 [V28,T10] ( 3, 6 ) simd16 -> mm0 "ReplaceWithLclVar is creating a new local variable"
-; V29 rat2 [V29,T00] ( 3, 6 ) int -> rax "ReplaceWithLclVar is creating a new local variable"
+; V26 cse0 [V26,T06] ( 3, 3 ) long -> rcx "CSE #02: aggressive"
+; V27 cse1 [V27,T07] ( 3, 3 ) long -> rax "CSE #01: aggressive"
+; V28 rat0 [V28,T10] ( 3, 6 ) double -> mm0 "ReplaceWithLclVar is creating a new local variable"
+; V29 rat1 [V29,T11] ( 3, 6 ) simd16 -> mm0 "ReplaceWithLclVar is creating a new local variable"
+; V30 rat2 [V30,T00] ( 3, 6 ) int -> rax "ReplaceWithLclVar is creating a new local variable"
;
; Lcl frame size = 8
G_M18819_IG01:
push rax
;; size=1 bbWeight=1 PerfScore 1.00
G_M18819_IG02:
add edi, 63
vxorps xmm0, xmm0, xmm0
vcvtsi2sd xmm0, xmm0, edi
vmulsd xmm0, xmm0, qword ptr [reloc @RWD00]
vroundsd xmm0, xmm0, xmm0, 10
vfixupimmsd xmm0, xmm0, xmmword ptr [reloc @RWD16], 0
vcmppd k1, xmm0, xmmword ptr [reloc @RWD32], 13
vcvttsd2si eax, xmm0
vpbroadcastd xmm0, eax
vpblendmd xmm0 {k1}, xmm0, xmmword ptr [reloc @RWD48]
vmovd eax, xmm0
add eax, 347
mov ecx, eax
sar ecx, 31
and ecx, 7
add eax, ecx
sar eax, 3
inc eax
cmp eax, 87
jae SHORT G_M18819_IG04
- mov rcx, 0xD1FFAB1E ; static handle
- movsx rcx, word ptr [rcx+2*rax]
- mov dword ptr [rdx], ecx
+ lea rcx, [rax+rax]
+ mov rdi, 0xD1FFAB1E ; static handle
+ movsx rdi, word ptr [rcx+rdi]
+ mov dword ptr [rdx], edi
mov rdx, 0xD1FFAB1E ; static handle
- mov rdx, qword ptr [rdx+8*rax]
- mov rcx, 0xD1FFAB1E ; static handle
- movsx rax, word ptr [rcx+2*rax]
- mov dword ptr [rsp+0x04], eax
- mov rax, rdx
- mov edx, dword ptr [rsp+0x04]
- ;; size=153 bbWeight=1 PerfScore 55.08
+ mov rax, qword ptr [rdx+8*rax]
+ mov rdx, 0xD1FFAB1E ; static handle
+ movsx rdx, word ptr [rcx+rdx]
+ ;; size=146 bbWeight=1 PerfScore 53.33
G_M18819_IG03:
add rsp, 8
ret
;; size=5 bbWeight=1 PerfScore 1.25
G_M18819_IG04:
call CORINFO_HELP_RNGCHKFAIL
int3
;; size=6 bbWeight=0 PerfScore 0.00
RWD00 dq 3FD34413509F79FFh ; 0.301029996
RWD08 dd 00000000h, 00000000h
RWD16 dq 0000000000000088h, 0000000000000000h
RWD32 dq 41DFFFFFFFC00000h, 41DFFFFFFFC00000h
RWD48 dq 7FFFFFFF7FFFFFFFh, 7FFFFFFF7FFFFFFFh
-; Total bytes of code 165, prolog size 1, PerfScore 57.33, instruction count 35, allocated bytes for code 166 (MethodHash=763bb67c) for method System.Number+Grisu3:GetCachedPowerForBinaryExponentRange(int,int,byref):System.Number+DiyFp (FullOpts)
+; Total bytes of code 158, prolog size 1, PerfScore 55.58, instruction count 33, allocated bytes for code 159 (MethodHash=763bb67c) for method System.Number+Grisu3:GetCachedPowerForBinaryExponentRange(int,int,byref):System.Number+DiyFp (FullOpts)-1 (-0.26 % of base) - System.Text.EncodingTable:InternalGetCodePageDataItem(int,int):System.Text.CodePageDataItem ; Assembly listing for method System.Text.EncodingTable:InternalGetCodePageDataItem(int,int):System.Text.CodePageDataItem (FullOpts)
; Emitting BLENDED_CODE for X64 with AVX512 - Unix
; FullOpts code
; optimized code
; rbp based frame
; partially interruptible
; No PGO data
; 0 inlinees with PGO data; 7 single block inlinees; 2 inlinees without PGO data
; Final local variable assignments
;
-; V00 arg0 [V00,T03] ( 3, 3 ) int -> r15 single-def
-; V01 arg1 [V01,T01] ( 8, 8 ) int -> rbx single-def
+; V00 arg0 [V00,T03] ( 3, 3 ) int -> rbx single-def
+; V01 arg1 [V01,T02] ( 6, 6 ) int -> rsi single-def
;* V02 loc0 [V02 ] ( 0, 0 ) ref -> zero-ref class-hnd exact single-def <System.String>
;* V03 loc1 [V03 ] ( 0, 0 ) ref -> zero-ref class-hnd exact single-def <System.String>
;* V04 loc2 [V04 ] ( 0, 0 ) ref -> zero-ref class-hnd exact single-def <System.String>
-; V05 loc3 [V05,T10] ( 2, 2 ) ref -> r15 class-hnd exact single-def <System.String>
-; V06 loc4 [V06,T11] ( 2, 2 ) int -> rbx
+; V05 loc3 [V05,T11] ( 2, 2 ) ref -> rbx class-hnd exact single-def <System.String>
+; V06 loc4 [V06,T12] ( 2, 2 ) int -> r15
;* V07 loc5 [V07 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op <System.ReadOnlySpan`1[int]>
-; V08 loc6 [V08,T07] ( 5, 3.50) int -> r13
+; V08 loc6 [V08,T08] ( 5, 3.50) int -> r13
;* V09 loc7 [V09 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op <System.ReadOnlySpan`1[uint]>
;# V10 OutArgs [V10 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
-; V11 tmp1 [V11,T06] ( 2, 4 ) int -> r14 "impAppendStmt"
+; V11 tmp1 [V11,T07] ( 2, 4 ) int -> r14 "impAppendStmt"
; V12 tmp2 [V12,T00] ( 8, 16 ) ref -> r12 class-hnd exact single-def "NewObj constructor temp" <System.Text.CodePageDataItem>
;* V13 tmp3 [V13 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[int]>
;* V14 tmp4 [V14 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[int]>
;* V15 tmp5 [V15 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[int]>
; V16 tmp6 [V16,T04] ( 6, 4.50) ref -> r13 class-hnd "Inline return value spill temp" <System.String>
-; V17 tmp7 [V17,T02] ( 7, 9 ) int -> r12 "Inlining Arg"
-; V18 tmp8 [V18,T12] ( 2, 1 ) long -> [rbp-0x30] spill-single-def "Inline stloc first use temp"
+; V17 tmp7 [V17,T01] ( 7, 9 ) int -> r12 "Inlining Arg"
+; V18 tmp8 [V18,T13] ( 2, 1 ) long -> [rbp-0x30] spill-single-def "Inline stloc first use temp"
; V19 tmp9 [V19,T05] ( 4, 4 ) ref -> r12 class-hnd exact single-def "dup spill" <System.String>
;* V20 tmp10 [V20 ] ( 0, 0 ) byref -> zero-ref "Inlining Arg"
;* V21 tmp11 [V21 ] ( 0, 0 ) byref -> zero-ref "Inlining Arg"
;* V22 tmp12 [V22 ] ( 0, 0 ) struct (16) zero-ref "ReadOnlySpan<T> for CreateSpan<T>" <System.ReadOnlySpan`1[uint]>
;* V23 tmp13 [V23 ] ( 0, 0 ) byref -> zero-ref "field V07._reference (fldOffset=0x0)" P-INDEP
;* V24 tmp14 [V24 ] ( 0, 0 ) int -> zero-ref "field V07._length (fldOffset=0x8)" P-INDEP
;* V25 tmp15 [V25 ] ( 0, 0 ) byref -> zero-ref single-def "field V09._reference (fldOffset=0x0)" P-INDEP
;* V26 tmp16 [V26 ] ( 0, 0 ) int -> zero-ref "field V09._length (fldOffset=0x8)" P-INDEP
;* V27 tmp17 [V27 ] ( 0, 0 ) byref -> zero-ref single-def "field V13._reference (fldOffset=0x0)" P-INDEP
;* V28 tmp18 [V28 ] ( 0, 0 ) int -> zero-ref "field V13._length (fldOffset=0x8)" P-INDEP
;* V29 tmp19 [V29 ] ( 0, 0 ) byref -> zero-ref single-def "field V14._reference (fldOffset=0x0)" P-INDEP
;* V30 tmp20 [V30 ] ( 0, 0 ) int -> zero-ref "field V14._length (fldOffset=0x8)" P-INDEP
;* V31 tmp21 [V31 ] ( 0, 0 ) byref -> zero-ref single-def "field V15._reference (fldOffset=0x0)" P-INDEP
;* V32 tmp22 [V32 ] ( 0, 0 ) int -> zero-ref "field V15._length (fldOffset=0x8)" P-INDEP
;* V33 tmp23 [V33 ] ( 0, 0 ) byref -> zero-ref single-def "field V22._reference (fldOffset=0x0)" P-INDEP
;* V34 tmp24 [V34 ] ( 0, 0 ) int -> zero-ref "field V22._length (fldOffset=0x8)" P-INDEP
-; V35 tmp25 [V35,T13] ( 2, 0 ) ref -> rdi single-def "argument with side effect"
-; V36 cse0 [V36,T08] ( 3, 3 ) int -> rdi "CSE #02: moderate"
-; V37 cse1 [V37,T09] ( 3, 3 ) long -> rax "CSE #03: moderate"
+; V35 tmp25 [V35,T14] ( 2, 0 ) ref -> rdi single-def "argument with side effect"
+; V36 cse0 [V36,T06] ( 4, 4 ) long -> r15 "CSE #02: moderate"
+; V37 cse1 [V37,T09] ( 3, 3 ) int -> rsi "CSE #03: moderate"
+; V38 cse2 [V38,T10] ( 3, 3 ) long -> rdi "CSE #04: moderate"
;
; Lcl frame size = 8
G_M21659_IG01:
push rbp
push r15
push r14
push r13
push r12
push rbx
push rax
lea rbp, [rsp+0x30]
- mov r15d, edi
- mov ebx, esi
- ;; size=21 bbWeight=1 PerfScore 8.00
+ mov ebx, edi
+ ;; size=18 bbWeight=1 PerfScore 7.75
G_M21659_IG02:
- cmp ebx, 8
+ cmp esi, 8
jae G_M21659_IG09
- mov edi, ebx
- mov rax, 0xD1FFAB1E ; static handle
- mov r14d, dword ptr [rax+4*rdi]
- cmp ebx, 9
+ mov r15d, esi
+ shl r15, 2
+ mov rdi, 0xD1FFAB1E ; static handle
+ mov r14d, dword ptr [r15+rdi]
+ cmp esi, 9
jae G_M21659_IG09
- mov edi, ebx
- mov rax, 0xD1FFAB1E ; static handle
- mov r13d, dword ptr [rax+4*rdi]
- lea edi, [rbx+0x01]
- cmp edi, 9
+ mov rdi, 0xD1FFAB1E ; static handle
+ mov r13d, dword ptr [r15+rdi]
+ inc esi
+ cmp esi, 9
jae G_M21659_IG09
- mov r12d, dword ptr [rax+4*rdi]
+ mov eax, esi
+ mov r12d, dword ptr [rdi+4*rax]
sub r12d, r13d
mov edi, r13d
mov eax, r12d
add rdi, rax
cmp rdi, 56
ja G_M21659_IG08
test r12d, r12d
jne SHORT G_M21659_IG04
- ;; size=93 bbWeight=1 PerfScore 14.75
+ ;; size=97 bbWeight=1 PerfScore 15.00
G_M21659_IG03:
mov r13, 0xD1FFAB1E ; ''
jmp SHORT G_M21659_IG06
;; size=12 bbWeight=0.50 PerfScore 1.12
G_M21659_IG04:
cmp r12d, 56
jne SHORT G_M21659_IG05
mov r13, 0xD1FFAB1E ; 'utf-16utf-16BEutf-32utf-32BEus-asciiiso-8859-1utf-7utf-8'
jmp SHORT G_M21659_IG06
;; size=18 bbWeight=0.50 PerfScore 1.75
G_M21659_IG05:
mov eax, r12d
mov qword ptr [rbp-0x30], rax
mov edi, r12d
call System.String:FastAllocateString(int):System.String
mov r12, rax
mov rsi, 0xD1FFAB1E
mov edi, r13d
lea rsi, bword ptr [rsi+2*rdi]
cmp byte ptr [r12], r12b
lea rdi, bword ptr [r12+0x0C]
mov r13, qword ptr [rbp-0x30]
lea rdx, [r13+r13]
mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong)
call [rax]System.SpanHelpers:Memmove(byref,byref,ulong)
mov r13, r12
;; size=68 bbWeight=0.50 PerfScore 6.88
G_M21659_IG06:
- mov edi, r15d
+ mov edi, ebx
mov rax, 0xD1FFAB1E ; code for System.Text.EncodingTable:GetDisplayName(int):System.String
call [rax]System.Text.EncodingTable:GetDisplayName(int):System.String
- mov r15, rax
- mov edi, ebx
- mov rax, 0xD1FFAB1E ; static handle
- mov ebx, dword ptr [rax+4*rdi]
+ mov rbx, rax
+ mov rdi, 0xD1FFAB1E ; static handle
+ mov r15d, dword ptr [r15+rdi]
mov rdi, 0xD1FFAB1E ; System.Text.CodePageDataItem
call CORINFO_HELP_NEWSFAST
mov r12, rax
mov dword ptr [r12+0x28], r14d
lea rdi, bword ptr [r12+0x08]
mov rsi, r13
call CORINFO_HELP_ASSIGN_REF
lea rdi, bword ptr [r12+0x10]
mov rsi, r13
call CORINFO_HELP_ASSIGN_REF
lea rdi, bword ptr [r12+0x18]
mov rsi, r13
call CORINFO_HELP_ASSIGN_REF
lea rdi, bword ptr [r12+0x20]
- mov rsi, r15
+ mov rsi, rbx
call CORINFO_HELP_ASSIGN_REF
- mov dword ptr [r12+0x2C], ebx
+ mov dword ptr [r12+0x2C], r15d
mov rax, r12
- ;; size=116 bbWeight=1 PerfScore 17.00
+ ;; size=114 bbWeight=1 PerfScore 16.75
G_M21659_IG07:
add rsp, 8
pop rbx
pop r12
pop r13
pop r14
pop r15
pop rbp
ret
;; size=15 bbWeight=1 PerfScore 4.25
G_M21659_IG08:
mov edi, 0x18B8D
mov rsi, 0xD1FFAB1E
call CORINFO_HELP_STRCNS
mov rdi, rax
mov esi, r13d
mov edx, r12d
mov rax, 0xD1FFAB1E ; code for System.String:ThrowSubstringArgumentOutOfRange(int,int):this
call [rax]System.String:ThrowSubstringArgumentOutOfRange(int,int):this
int3
;; size=42 bbWeight=0 PerfScore 0.00
G_M21659_IG09:
call CORINFO_HELP_RNGCHKFAIL
int3
;; size=6 bbWeight=0 PerfScore 0.00
-; Total bytes of code 391, prolog size 16, PerfScore 53.75, instruction count 97, allocated bytes for code 391 (MethodHash=8edfab64) for method System.Text.EncodingTable:InternalGetCodePageDataItem(int,int):System.Text.CodePageDataItem (FullOpts)
+; Total bytes of code 390, prolog size 16, PerfScore 53.50, instruction count 96, allocated bytes for code 390 (MethodHash=8edfab64) for method System.Text.EncodingTable:InternalGetCodePageDataItem(int,int):System.Text.CodePageDataItem (FullOpts) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Job completed in 14 minutes.
dotnet/runtime#104387
Diffs
Artifacts:
The text was updated successfully, but these errors were encountered: