search

MihuBot / runtime-utils

0 stars 0 forks source link

[JitDiff X64] [xtqqczze] Refactor memory copies in `SafeBuffer` #548

Open MihuBot opened 3 months ago

MihuBot commented 3 months ago

Job completed in 14 minutes. https://github.com/dotnet/runtime/pull/105206

Diffs

Found 261 files with textual diffs.

Summary of Code Size diffs:
(Lower is better)

Total bytes of base: 39655437
Total bytes of diff: 39654923
Total bytes of delta: -514 (-0.00 % of base)
Total relative delta: -2.06
    diff is an improvement.
    relative diff is an improvement.

Top file improvements (bytes):
        -514 : System.Private.CoreLib.dasm (-0.01 % of base)

1 total files with Code Size differences (1 improved, 0 regressed), 258 unchanged.

Top method regressions (bytes):
          13 (4.42 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[System.Numerics.Vector`1[float]](ulong,System.Span`1[System.Numerics.Vector`1[float]]):this (FullOpts)
          13 (4.42 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[System.Numerics.Vector`1[float]](ulong,System.ReadOnlySpan`1[System.Numerics.Vector`1[float]]):this (FullOpts)

Top method improvements (bytes):
         -45 (-18.83 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[ubyte](ulong):ubyte:this (FullOpts)
         -44 (-18.57 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[int](ulong):int:this (FullOpts)
         -44 (-18.41 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[long](ulong):long:this (FullOpts)
         -44 (-18.41 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[short](ulong):short:this (FullOpts)
         -37 (-15.42 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[double](ulong):double:this (FullOpts)
         -35 (-13.16 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[System.Numerics.Vector`1[float]](ulong):System.Numerics.Vector`1[float]:this (FullOpts)
         -29 (-12.45 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[double](ulong,double):this (FullOpts)
         -25 (-10.78 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[long](ulong,long):this (FullOpts)
         -24 (-10.39 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[int](ulong,int):this (FullOpts)
         -24 (-10.39 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[ubyte](ulong,ubyte):this (FullOpts)
         -23 (-9.96 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[short](ulong,short):this (FullOpts)
         -23 (-10.09 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[System.Numerics.Vector`1[float]](ulong,System.Numerics.Vector`1[float]):this (FullOpts)
         -19 (-6.35 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[long](ulong,System.ReadOnlySpan`1[long]):this (FullOpts)
         -17 (-5.72 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[int](ulong,System.ReadOnlySpan`1[int]):this (FullOpts)
         -17 (-5.69 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[short](ulong,System.ReadOnlySpan`1[short]):this (FullOpts)
         -16 (-5.35 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[long](ulong,System.Span`1[long]):this (FullOpts)
         -16 (-5.39 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[ubyte](ulong,System.ReadOnlySpan`1[ubyte]):this (FullOpts)
         -15 (-5.02 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[double](ulong,System.ReadOnlySpan`1[double]):this (FullOpts)
         -14 (-4.71 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[int](ulong,System.Span`1[int]):this (FullOpts)
         -14 (-4.68 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[short](ulong,System.Span`1[short]):this (FullOpts)
         -13 (-4.38 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[ubyte](ulong,System.Span`1[ubyte]):this (FullOpts)
          -2 (-0.67 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[double](ulong,System.Span`1[double]):this (FullOpts)

Top method regressions (percentages):
          13 (4.42 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[System.Numerics.Vector`1[float]](ulong,System.Span`1[System.Numerics.Vector`1[float]]):this (FullOpts)
          13 (4.42 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[System.Numerics.Vector`1[float]](ulong,System.ReadOnlySpan`1[System.Numerics.Vector`1[float]]):this (FullOpts)

Top method improvements (percentages):
         -45 (-18.83 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[ubyte](ulong):ubyte:this (FullOpts)
         -44 (-18.57 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[int](ulong):int:this (FullOpts)
         -44 (-18.41 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[long](ulong):long:this (FullOpts)
         -44 (-18.41 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[short](ulong):short:this (FullOpts)
         -37 (-15.42 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[double](ulong):double:this (FullOpts)
         -35 (-13.16 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Read[System.Numerics.Vector`1[float]](ulong):System.Numerics.Vector`1[float]:this (FullOpts)
         -29 (-12.45 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[double](ulong,double):this (FullOpts)
         -25 (-10.78 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[long](ulong,long):this (FullOpts)
         -24 (-10.39 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[int](ulong,int):this (FullOpts)
         -24 (-10.39 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[ubyte](ulong,ubyte):this (FullOpts)
         -23 (-10.09 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[System.Numerics.Vector`1[float]](ulong,System.Numerics.Vector`1[float]):this (FullOpts)
         -23 (-9.96 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:Write[short](ulong,short):this (FullOpts)
         -19 (-6.35 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[long](ulong,System.ReadOnlySpan`1[long]):this (FullOpts)
         -17 (-5.72 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[int](ulong,System.ReadOnlySpan`1[int]):this (FullOpts)
         -17 (-5.69 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[short](ulong,System.ReadOnlySpan`1[short]):this (FullOpts)
         -16 (-5.39 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[ubyte](ulong,System.ReadOnlySpan`1[ubyte]):this (FullOpts)
         -16 (-5.35 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[long](ulong,System.Span`1[long]):this (FullOpts)
         -15 (-5.02 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:WriteSpan[double](ulong,System.ReadOnlySpan`1[double]):this (FullOpts)
         -14 (-4.71 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[int](ulong,System.Span`1[int]):this (FullOpts)
         -14 (-4.68 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[short](ulong,System.Span`1[short]):this (FullOpts)
         -13 (-4.38 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[ubyte](ulong,System.Span`1[ubyte]):this (FullOpts)
          -2 (-0.67 % of base) : System.Private.CoreLib.dasm - System.Runtime.InteropServices.SafeBuffer:ReadSpan[double](ulong,System.Span`1[double]):this (FullOpts)

24 total methods with Code Size differences (22 improved, 2 regressed), 231354 unchanged.

--------------------------------------------------------------------------------

Artifacts:

MihuBot commented 3 months ago

Top method regressions

13 (4.42 % of base) - System.Runtime.InteropServices.SafeBuffer:ReadSpan[System.Numerics.Vector`1[float]](ulong,System.Span`1[System.Numerics.Vector`1[float]]):this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:ReadSpan[System.Numerics.Vector`1[float]](ulong,System.Span`1[System.Numerics.Vector`1[float]]):this (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; 4 single block inlinees; 1 inlinees without PGO data +; 0 inlinees with PGO data; 4 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; ; V00 this [V00,T03] ( 8, 6.50) ref -> [rbp-0x30] this class-hnd EH-live single-def ; V01 arg1 [V01,T06] ( 3, 3 ) long -> r15 single-def ;* V02 arg2 [V02 ] ( 0, 0 ) struct (16) zero-ref multireg-arg ld-addr-op single-def ; V03 loc0 [V03,T09] ( 2, 2 ) int -> rax ; V04 loc1 [V04,T04] ( 3, 6 ) long -> r15 ; V05 loc2 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x28] do-not-enreg[X] addr-exposed ld-addr-op ; V06 loc3 [V06,T07] ( 2, 5 ) byref -> r14 single-def ; V07 loc4 [V07,T00] ( 5, 17 ) int -> rdi ;# V08 OutArgs [V08 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V09 tmp1 [V09 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" ;* V10 tmp2 [V10 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op "Inlining Arg" -;* V11 tmp3 [V11 ] ( 0, 0 ) byref -> zero-ref "Inlining Arg" -;* V12 tmp4 [V12 ] ( 0, 0 ) byref -> zero-ref "Inlining Arg" -; V13 tmp5 [V13,T08] ( 2, 2 ) byref -> r14 single-def "field V02._reference (fldOffset=0x0)" P-INDEP -; V14 tmp6 [V14,T02] ( 4, 7 ) int -> rbx single-def "field V02._length (fldOffset=0x8)" P-INDEP -;* V15 tmp7 [V15 ] ( 0, 0 ) byref -> zero-ref single-def "field V10._reference (fldOffset=0x0)" P-INDEP -;* V16 tmp8 [V16 ] ( 0, 0 ) int -> zero-ref "field V10._length (fldOffset=0x8)" P-INDEP -; V17 PSPSym [V17,T10] ( 1, 1 ) long -> [rbp-0x40] do-not-enreg[V] "PSPSym" -; V18 cse0 [V18,T05] ( 3, 6 ) long -> r13 "CSE #01: aggressive" -; V19 cse1 [V19,T01] ( 3, 12 ) long -> rsi "CSE #02: aggressive" +; V11 tmp3 [V11,T08] ( 2, 2 ) byref -> r14 single-def "field V02._reference (fldOffset=0x0)" P-INDEP +; V12 tmp4 [V12,T02] ( 4, 7 ) int -> rbx single-def "field V02._length (fldOffset=0x8)" P-INDEP +;* V13 tmp5 [V13 ] ( 0, 0 ) byref -> zero-ref single-def "field V10._reference (fldOffset=0x0)" P-INDEP +;* V14 tmp6 [V14 ] ( 0, 0 ) int -> zero-ref "field V10._length (fldOffset=0x8)" P-INDEP +; V15 PSPSym [V15,T10] ( 1, 1 ) long -> [rbp-0x40] do-not-enreg[V] "PSPSym" +; V16 cse0 [V16,T05] ( 3, 6 ) long -> r13 "CSE #01: aggressive" +; V17 cse1 [V17,T01] ( 3, 12 ) long -> rsi "CSE #02: aggressive" ; ; Lcl frame size = 32 G_M7477_IG01: push rbp push r15 push r14 push r13 push rbx sub rsp, 32 lea rbp, [rsp+0x40] mov qword ptr [rbp-0x40], rsp mov gword ptr [rbp-0x30], rdi mov r15, rsi mov r14, rdx mov ebx, ecx ;; size=33 bbWeight=1 PerfScore 8.50 G_M7477_IG02: cmp qword ptr [rdi+0x18], -1 je G_M7477_IG09 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:AlignedSizeOf[System.Numerics.Vector`1[float]]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:AlignedSizeOf[System.Numerics.Vector`1[float]]():uint mov rdi, gword ptr [rbp-0x30] add r15, qword ptr [rdi+0x08] mov r13d, eax mov edx, ebx imul rdx, r13 jo G_M7477_IG10 test rdx, rdx jl G_M7477_IG10 mov rsi, r15 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi mov dword ptr [rbp-0x28], esi ;; size=75 bbWeight=1 PerfScore 20.75 G_M7477_IG03: lea rsi, [rbp-0x28] mov rdi, gword ptr [rbp-0x30] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this xor edi, edi test ebx, ebx jle SHORT G_M7477_IG05 - align [0 bytes for IG04] - ;; size=26 bbWeight=1 PerfScore 6.25 + align [10 bytes for IG04] + ;; size=36 bbWeight=1 PerfScore 6.50 G_M7477_IG04: - movsxd rsi, edi + mov esi, edi mov rax, r13 imul rax, rsi - add rax, r15 + vmovups ymm0, ymmword ptr [rax+r15] shl rsi, 5 - add rsi, r14 - vmovdqu ymm0, ymmword ptr [rax] - vmovdqu ymmword ptr [rsi], ymm0 + vmovups ymmword ptr [r14+rsi], ymm0 inc edi cmp edi, ebx jl SHORT G_M7477_IG04 - ;; size=34 bbWeight=4 PerfScore 48.00 + ;; size=31 bbWeight=4 PerfScore 46.00 G_M7477_IG05: cmp byte ptr [rbp-0x28], 0 je SHORT G_M7477_IG07 ;; size=6 bbWeight=1 PerfScore 3.00 G_M7477_IG06: mov rdi, gword ptr [rbp-0x30] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 G_M7477_IG07: nop ;; size=1 bbWeight=1 PerfScore 0.25 G_M7477_IG08: + vzeroupper add rsp, 32 pop rbx pop r13 pop r14 pop r15 pop rbp ret - ;; size=13 bbWeight=1 PerfScore 3.75 + ;; size=16 bbWeight=1 PerfScore 4.75 G_M7477_IG09: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 G_M7477_IG10: call CORINFO_HELP_OVERFLOW int3 ;; size=6 bbWeight=0 PerfScore 0.00 G_M7477_IG11: push rbp push r15 push r14 push r13 push rbx sub rsp, 16 mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp lea rbp, [rbp+0x40] ;; size=23 bbWeight=0 PerfScore 0.00 G_M7477_IG12: cmp byte ptr [rbp-0x28], 0 je SHORT G_M7477_IG13 mov rdi, gword ptr [rbp-0x30] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 G_M7477_IG13: nop ;; size=1 bbWeight=0 PerfScore 0.00 G_M7477_IG14: + vzeroupper add rsp, 16 pop rbx pop r13 pop r14 pop r15 pop rbp ret - ;; size=13 bbWeight=0 PerfScore 0.00 + ;; size=16 bbWeight=0 PerfScore 0.00 -; Total bytes of code 294, prolog size 33, PerfScore 92.75, instruction count 92, allocated bytes for code 294 (MethodHash=6e36e2ca) for method System.Runtime.InteropServices.SafeBuffer:ReadSpan[System.Numerics.Vector`1[float]](ulong,System.Span`1[System.Numerics.Vector`1[float]]):this (FullOpts) +; Total bytes of code 307, prolog size 33, PerfScore 92.00, instruction count 92, allocated bytes for code 307 (MethodHash=6e36e2ca) for method System.Runtime.InteropServices.SafeBuffer:ReadSpan[System.Numerics.Vector`1[float]](ulong,System.Span`1[System.Numerics.Vector`1[float]]):this (FullOpts) ```
13 (4.42 % of base) - System.Runtime.InteropServices.SafeBuffer:WriteSpan[System.Numerics.Vector`1[float]](ulong,System.ReadOnlySpan`1[System.Numerics.Vector`1[float]]):this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:WriteSpan[System.Numerics.Vector`1[float]](ulong,System.ReadOnlySpan`1[System.Numerics.Vector`1[float]]):this (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; 4 single block inlinees; 1 inlinees without PGO data +; 0 inlinees with PGO data; 4 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; ; V00 this [V00,T03] ( 8, 6.50) ref -> [rbp-0x30] this class-hnd EH-live single-def ; V01 arg1 [V01,T06] ( 3, 3 ) long -> r15 single-def ;* V02 arg2 [V02 ] ( 0, 0 ) struct (16) zero-ref multireg-arg ld-addr-op single-def ; V03 loc0 [V03,T09] ( 2, 2 ) int -> rax ; V04 loc1 [V04,T04] ( 3, 6 ) long -> r15 ; V05 loc2 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x28] do-not-enreg[X] addr-exposed ld-addr-op ; V06 loc3 [V06,T07] ( 2, 5 ) byref -> r14 single-def ; V07 loc4 [V07,T00] ( 5, 17 ) int -> rdi ;# V08 OutArgs [V08 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V09 tmp1 [V09 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" ;* V10 tmp2 [V10 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op "Inlining Arg" -;* V11 tmp3 [V11 ] ( 0, 0 ) byref -> zero-ref "Inlining Arg" -;* V12 tmp4 [V12 ] ( 0, 0 ) byref -> zero-ref "Inlining Arg" -; V13 tmp5 [V13,T08] ( 2, 2 ) byref -> r14 single-def "field V02._reference (fldOffset=0x0)" P-INDEP -; V14 tmp6 [V14,T02] ( 4, 7 ) int -> rbx single-def "field V02._length (fldOffset=0x8)" P-INDEP -;* V15 tmp7 [V15 ] ( 0, 0 ) byref -> zero-ref single-def "field V10._reference (fldOffset=0x0)" P-INDEP -;* V16 tmp8 [V16 ] ( 0, 0 ) int -> zero-ref "field V10._length (fldOffset=0x8)" P-INDEP -; V17 PSPSym [V17,T10] ( 1, 1 ) long -> [rbp-0x40] do-not-enreg[V] "PSPSym" -; V18 cse0 [V18,T05] ( 3, 6 ) long -> r13 "CSE #01: aggressive" -; V19 cse1 [V19,T01] ( 3, 12 ) long -> rsi "CSE #02: aggressive" +; V11 tmp3 [V11,T08] ( 2, 2 ) byref -> r14 single-def "field V02._reference (fldOffset=0x0)" P-INDEP +; V12 tmp4 [V12,T02] ( 4, 7 ) int -> rbx single-def "field V02._length (fldOffset=0x8)" P-INDEP +;* V13 tmp5 [V13 ] ( 0, 0 ) byref -> zero-ref single-def "field V10._reference (fldOffset=0x0)" P-INDEP +;* V14 tmp6 [V14 ] ( 0, 0 ) int -> zero-ref "field V10._length (fldOffset=0x8)" P-INDEP +; V15 PSPSym [V15,T10] ( 1, 1 ) long -> [rbp-0x40] do-not-enreg[V] "PSPSym" +; V16 cse0 [V16,T05] ( 3, 6 ) long -> r13 "CSE #01: aggressive" +; V17 cse1 [V17,T01] ( 3, 12 ) long -> rsi "CSE #02: aggressive" ; ; Lcl frame size = 32 G_M34876_IG01: push rbp push r15 push r14 push r13 push rbx sub rsp, 32 lea rbp, [rsp+0x40] mov qword ptr [rbp-0x40], rsp mov gword ptr [rbp-0x30], rdi mov r15, rsi mov r14, rdx mov ebx, ecx ;; size=33 bbWeight=1 PerfScore 8.50 G_M34876_IG02: cmp qword ptr [rdi+0x18], -1 je G_M34876_IG09 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:AlignedSizeOf[System.Numerics.Vector`1[float]]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:AlignedSizeOf[System.Numerics.Vector`1[float]]():uint mov rdi, gword ptr [rbp-0x30] add r15, qword ptr [rdi+0x08] mov r13d, eax mov edx, ebx imul rdx, r13 jo G_M34876_IG10 test rdx, rdx jl G_M34876_IG10 mov rsi, r15 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi mov dword ptr [rbp-0x28], esi ;; size=75 bbWeight=1 PerfScore 20.75 G_M34876_IG03: lea rsi, [rbp-0x28] mov rdi, gword ptr [rbp-0x30] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this xor edi, edi test ebx, ebx jle SHORT G_M34876_IG05 - align [0 bytes for IG04] - ;; size=26 bbWeight=1 PerfScore 6.25 + align [10 bytes for IG04] + ;; size=36 bbWeight=1 PerfScore 6.50 G_M34876_IG04: - movsxd rsi, edi - mov rax, r13 - imul rax, rsi - add rax, r15 - shl rsi, 5 - add rsi, r14 - vmovdqu ymm0, ymmword ptr [rsi] - vmovdqu ymmword ptr [rax], ymm0 + mov esi, edi + mov rax, rsi + shl rax, 5 + vmovups ymm0, ymmword ptr [r14+rax] + imul rsi, r13 + vmovups ymmword ptr [rsi+r15], ymm0 inc edi cmp edi, ebx jl SHORT G_M34876_IG04 - ;; size=34 bbWeight=4 PerfScore 48.00 + ;; size=31 bbWeight=4 PerfScore 46.00 G_M34876_IG05: cmp byte ptr [rbp-0x28], 0 je SHORT G_M34876_IG07 ;; size=6 bbWeight=1 PerfScore 3.00 G_M34876_IG06: mov rdi, gword ptr [rbp-0x30] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 G_M34876_IG07: nop ;; size=1 bbWeight=1 PerfScore 0.25 G_M34876_IG08: + vzeroupper add rsp, 32 pop rbx pop r13 pop r14 pop r15 pop rbp ret - ;; size=13 bbWeight=1 PerfScore 3.75 + ;; size=16 bbWeight=1 PerfScore 4.75 G_M34876_IG09: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 G_M34876_IG10: call CORINFO_HELP_OVERFLOW int3 ;; size=6 bbWeight=0 PerfScore 0.00 G_M34876_IG11: push rbp push r15 push r14 push r13 push rbx sub rsp, 16 mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp lea rbp, [rbp+0x40] ;; size=23 bbWeight=0 PerfScore 0.00 G_M34876_IG12: cmp byte ptr [rbp-0x28], 0 je SHORT G_M34876_IG13 mov rdi, gword ptr [rbp-0x30] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 G_M34876_IG13: nop ;; size=1 bbWeight=0 PerfScore 0.00 G_M34876_IG14: + vzeroupper add rsp, 16 pop rbx pop r13 pop r14 pop r15 pop rbp ret - ;; size=13 bbWeight=0 PerfScore 0.00 + ;; size=16 bbWeight=0 PerfScore 0.00 -; Total bytes of code 294, prolog size 33, PerfScore 92.75, instruction count 92, allocated bytes for code 294 (MethodHash=fab077c3) for method System.Runtime.InteropServices.SafeBuffer:WriteSpan[System.Numerics.Vector`1[float]](ulong,System.ReadOnlySpan`1[System.Numerics.Vector`1[float]]):this (FullOpts) +; Total bytes of code 307, prolog size 33, PerfScore 92.00, instruction count 92, allocated bytes for code 307 (MethodHash=fab077c3) for method System.Runtime.InteropServices.SafeBuffer:WriteSpan[System.Numerics.Vector`1[float]](ulong,System.ReadOnlySpan`1[System.Numerics.Vector`1[float]]):this (FullOpts) ```
MihuBot commented 3 months ago

Top method improvements

-45 (-18.83 % of base) - System.Runtime.InteropServices.SafeBuffer:Read[ubyte](ulong):ubyte:this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Read[ubyte](ulong):ubyte:this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x28] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x18] this class-hnd EH-live single-def ; V01 arg1 [V01,T01] ( 3, 3 ) long -> rbx single-def -; V02 loc0 [V02,T02] ( 3, 3 ) int -> r15 -; V03 loc1 [V03,T03] ( 3, 3 ) long -> rbx -; V04 loc2 [V04 ] ( 3, 3 ) ubyte -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op -; V05 loc3 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x20] do-not-enreg[X] addr-exposed ld-addr-op +; V02 loc0 [V02,T03] ( 2, 2 ) int -> rax +; V03 loc1 [V03,T02] ( 3, 3 ) long -> rbx +; V04 loc2 [V04 ] ( 4, 3 ) ubyte -> [rbp-0x10] do-not-enreg[X] addr-exposed ld-addr-op +; V05 loc3 [V05,T04] ( 2, 2 ) ubyte -> rbx ;# V06 OutArgs [V06 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V07 tmp1 [V07 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" -; V08 PSPSym [V08,T04] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" +; V08 PSPSym [V08,T05] ( 1, 1 ) long -> [rbp-0x20] do-not-enreg[V] "PSPSym" ; -; Lcl frame size = 32 +; Lcl frame size = 24 G_M48575_IG01: push rbp - push r15 push rbx - sub rsp, 32 - lea rbp, [rsp+0x30] - mov qword ptr [rbp-0x30], rsp - mov gword ptr [rbp-0x28], rdi + sub rsp, 24 + lea rbp, [rsp+0x20] + mov qword ptr [rbp-0x20], rsp + mov gword ptr [rbp-0x18], rdi mov rbx, rsi - ;; size=24 bbWeight=1 PerfScore 6.00 + ;; size=22 bbWeight=1 PerfScore 5.00 G_M48575_IG02: cmp qword ptr [rdi+0x18], -1 - je G_M48575_IG08 + je SHORT G_M48575_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[ubyte]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[ubyte]():uint - mov r15d, eax - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x18] add rbx, qword ptr [rdi+0x08] - mov edx, r15d + mov edx, eax mov rsi, rbx mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this - mov byte ptr [rbp-0x18], 0 xor esi, esi - mov dword ptr [rbp-0x20], esi - ;; size=61 bbWeight=1 PerfScore 17.50 + mov dword ptr [rbp-0x10], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M48575_IG03: - lea rsi, [rbp-0x20] - mov rdi, gword ptr [rbp-0x28] + lea rsi, [rbp-0x10] + mov rdi, gword ptr [rbp-0x18] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r15d - lea rdi, [rbp-0x18] - mov rsi, rbx - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 + movzx rbx, byte ptr [rbx] + ;; size=23 bbWeight=1 PerfScore 6.75 G_M48575_IG04: - cmp byte ptr [rbp-0x20], 0 + cmp byte ptr [rbp-0x10], 0 je SHORT G_M48575_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 G_M48575_IG05: - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 G_M48575_IG06: - movzx rax, byte ptr [rbp-0x18] - ;; size=4 bbWeight=1 PerfScore 1.00 + mov eax, ebx + ;; size=2 bbWeight=1 PerfScore 0.25 G_M48575_IG07: - add rsp, 32 + add rsp, 24 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=1 PerfScore 2.75 + ;; size=7 bbWeight=1 PerfScore 2.25 G_M48575_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 G_M48575_IG09: push rbp - push r15 push rbx - sub rsp, 16 + push rax mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp - lea rbp, [rbp+0x30] - ;; size=19 bbWeight=0 PerfScore 0.00 + lea rbp, [rbp+0x20] + ;; size=14 bbWeight=0 PerfScore 0.00 G_M48575_IG10: - cmp byte ptr [rbp-0x20], 0 + cmp byte ptr [rbp-0x10], 0 je SHORT G_M48575_IG11 - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 G_M48575_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 G_M48575_IG12: - add rsp, 16 + add rsp, 8 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=0 PerfScore 0.00 + ;; size=7 bbWeight=0 PerfScore 0.00 -; Total bytes of code 239, prolog size 24, PerfScore 41.75, instruction count 68, allocated bytes for code 239 (MethodHash=0df04240) for method System.Runtime.InteropServices.SafeBuffer:Read[ubyte](ulong):ubyte:this (FullOpts) +; Total bytes of code 194, prolog size 22, PerfScore 35.75, instruction count 57, allocated bytes for code 194 (MethodHash=0df04240) for method System.Runtime.InteropServices.SafeBuffer:Read[ubyte](ulong):ubyte:this (FullOpts) ```
-44 (-18.57 % of base) - System.Runtime.InteropServices.SafeBuffer:Read[int](ulong):int:this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Read[int](ulong):int:this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x28] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x18] this class-hnd EH-live single-def ; V01 arg1 [V01,T01] ( 3, 3 ) long -> rbx single-def -; V02 loc0 [V02,T02] ( 3, 3 ) int -> r15 -; V03 loc1 [V03,T03] ( 3, 3 ) long -> rbx -; V04 loc2 [V04 ] ( 3, 3 ) int -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op -; V05 loc3 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x20] do-not-enreg[X] addr-exposed ld-addr-op +; V02 loc0 [V02,T03] ( 2, 2 ) int -> rax +; V03 loc1 [V03,T02] ( 3, 3 ) long -> rbx +; V04 loc2 [V04 ] ( 4, 3 ) ubyte -> [rbp-0x10] do-not-enreg[X] addr-exposed ld-addr-op +; V05 loc3 [V05,T04] ( 2, 2 ) int -> rbx ;# V06 OutArgs [V06 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V07 tmp1 [V07 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" -; V08 PSPSym [V08,T04] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" +; V08 PSPSym [V08,T05] ( 1, 1 ) long -> [rbp-0x20] do-not-enreg[V] "PSPSym" ; -; Lcl frame size = 32 +; Lcl frame size = 24 G_M1215_IG01: push rbp - push r15 push rbx - sub rsp, 32 - lea rbp, [rsp+0x30] - mov qword ptr [rbp-0x30], rsp - mov gword ptr [rbp-0x28], rdi + sub rsp, 24 + lea rbp, [rsp+0x20] + mov qword ptr [rbp-0x20], rsp + mov gword ptr [rbp-0x18], rdi mov rbx, rsi - ;; size=24 bbWeight=1 PerfScore 6.00 + ;; size=22 bbWeight=1 PerfScore 5.00 G_M1215_IG02: cmp qword ptr [rdi+0x18], -1 - je G_M1215_IG09 + je SHORT G_M1215_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[int]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[int]():uint - mov r15d, eax - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x18] add rbx, qword ptr [rdi+0x08] - mov edx, r15d + mov edx, eax mov rsi, rbx mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi - mov dword ptr [rbp-0x18], esi - ;; size=57 bbWeight=1 PerfScore 16.50 + mov dword ptr [rbp-0x10], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M1215_IG03: - mov dword ptr [rbp-0x20], esi - ;; size=3 bbWeight=1 PerfScore 1.00 -G_M1215_IG04: - lea rsi, [rbp-0x20] - mov rdi, gword ptr [rbp-0x28] + lea rsi, [rbp-0x10] + mov rdi, gword ptr [rbp-0x18] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r15d - lea rdi, [rbp-0x18] - mov rsi, rbx - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 -G_M1215_IG05: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M1215_IG07 + mov ebx, dword ptr [rbx] + ;; size=22 bbWeight=1 PerfScore 6.75 +G_M1215_IG04: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M1215_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 -G_M1215_IG06: - mov rdi, gword ptr [rbp-0x28] +G_M1215_IG05: + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 +G_M1215_IG06: + mov eax, ebx + ;; size=2 bbWeight=1 PerfScore 0.25 G_M1215_IG07: - mov eax, dword ptr [rbp-0x18] - ;; size=3 bbWeight=1 PerfScore 1.00 -G_M1215_IG08: - add rsp, 32 + add rsp, 24 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=1 PerfScore 2.75 -G_M1215_IG09: + ;; size=7 bbWeight=1 PerfScore 2.25 +G_M1215_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 -G_M1215_IG10: +G_M1215_IG09: push rbp - push r15 push rbx - sub rsp, 16 + push rax mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp - lea rbp, [rbp+0x30] - ;; size=19 bbWeight=0 PerfScore 0.00 -G_M1215_IG11: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M1215_IG12 - mov rdi, gword ptr [rbp-0x28] + lea rbp, [rbp+0x20] + ;; size=14 bbWeight=0 PerfScore 0.00 +G_M1215_IG10: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M1215_IG11 + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 -G_M1215_IG12: +G_M1215_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 -G_M1215_IG13: - add rsp, 16 +G_M1215_IG12: + add rsp, 8 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=0 PerfScore 0.00 + ;; size=7 bbWeight=0 PerfScore 0.00 -; Total bytes of code 237, prolog size 24, PerfScore 41.75, instruction count 68, allocated bytes for code 237 (MethodHash=84fefb40) for method System.Runtime.InteropServices.SafeBuffer:Read[int](ulong):int:this (FullOpts) +; Total bytes of code 193, prolog size 22, PerfScore 35.75, instruction count 57, allocated bytes for code 193 (MethodHash=84fefb40) for method System.Runtime.InteropServices.SafeBuffer:Read[int](ulong):int:this (FullOpts) ```
-44 (-18.41 % of base) - System.Runtime.InteropServices.SafeBuffer:Read[long](ulong):long:this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Read[long](ulong):long:this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x28] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x18] this class-hnd EH-live single-def ; V01 arg1 [V01,T01] ( 3, 3 ) long -> rbx single-def -; V02 loc0 [V02,T02] ( 3, 3 ) int -> r15 -; V03 loc1 [V03,T03] ( 3, 3 ) long -> rbx -; V04 loc2 [V04 ] ( 3, 3 ) long -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op -; V05 loc3 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x20] do-not-enreg[X] addr-exposed ld-addr-op +; V02 loc0 [V02,T03] ( 2, 2 ) int -> rax +; V03 loc1 [V03,T02] ( 3, 3 ) long -> rbx +; V04 loc2 [V04 ] ( 4, 3 ) ubyte -> [rbp-0x10] do-not-enreg[X] addr-exposed ld-addr-op +; V05 loc3 [V05,T04] ( 2, 2 ) long -> rbx ;# V06 OutArgs [V06 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V07 tmp1 [V07 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" -; V08 PSPSym [V08,T04] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" +; V08 PSPSym [V08,T05] ( 1, 1 ) long -> [rbp-0x20] do-not-enreg[V] "PSPSym" ; -; Lcl frame size = 32 +; Lcl frame size = 24 G_M44191_IG01: push rbp - push r15 push rbx - sub rsp, 32 - lea rbp, [rsp+0x30] - mov qword ptr [rbp-0x30], rsp - mov gword ptr [rbp-0x28], rdi + sub rsp, 24 + lea rbp, [rsp+0x20] + mov qword ptr [rbp-0x20], rsp + mov gword ptr [rbp-0x18], rdi mov rbx, rsi - ;; size=24 bbWeight=1 PerfScore 6.00 + ;; size=22 bbWeight=1 PerfScore 5.00 G_M44191_IG02: cmp qword ptr [rdi+0x18], -1 - je G_M44191_IG09 + je SHORT G_M44191_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[long]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[long]():uint - mov r15d, eax - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x18] add rbx, qword ptr [rdi+0x08] - mov edx, r15d + mov edx, eax mov rsi, rbx mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi - mov qword ptr [rbp-0x18], rsi - ;; size=58 bbWeight=1 PerfScore 16.50 + mov dword ptr [rbp-0x10], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M44191_IG03: - mov dword ptr [rbp-0x20], esi - ;; size=3 bbWeight=1 PerfScore 1.00 -G_M44191_IG04: - lea rsi, [rbp-0x20] - mov rdi, gword ptr [rbp-0x28] + lea rsi, [rbp-0x10] + mov rdi, gword ptr [rbp-0x18] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r15d - lea rdi, [rbp-0x18] - mov rsi, rbx - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 -G_M44191_IG05: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M44191_IG07 + mov rbx, qword ptr [rbx] + ;; size=23 bbWeight=1 PerfScore 6.75 +G_M44191_IG04: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M44191_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 -G_M44191_IG06: - mov rdi, gword ptr [rbp-0x28] +G_M44191_IG05: + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 +G_M44191_IG06: + mov rax, rbx + ;; size=3 bbWeight=1 PerfScore 0.25 G_M44191_IG07: - mov rax, qword ptr [rbp-0x18] - ;; size=4 bbWeight=1 PerfScore 1.00 -G_M44191_IG08: - add rsp, 32 + add rsp, 24 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=1 PerfScore 2.75 -G_M44191_IG09: + ;; size=7 bbWeight=1 PerfScore 2.25 +G_M44191_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 -G_M44191_IG10: +G_M44191_IG09: push rbp - push r15 push rbx - sub rsp, 16 + push rax mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp - lea rbp, [rbp+0x30] - ;; size=19 bbWeight=0 PerfScore 0.00 -G_M44191_IG11: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M44191_IG12 - mov rdi, gword ptr [rbp-0x28] + lea rbp, [rbp+0x20] + ;; size=14 bbWeight=0 PerfScore 0.00 +G_M44191_IG10: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M44191_IG11 + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 -G_M44191_IG12: +G_M44191_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 -G_M44191_IG13: - add rsp, 16 +G_M44191_IG12: + add rsp, 8 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=0 PerfScore 0.00 + ;; size=7 bbWeight=0 PerfScore 0.00 -; Total bytes of code 239, prolog size 24, PerfScore 41.75, instruction count 68, allocated bytes for code 239 (MethodHash=99115360) for method System.Runtime.InteropServices.SafeBuffer:Read[long](ulong):long:this (FullOpts) +; Total bytes of code 195, prolog size 22, PerfScore 35.75, instruction count 57, allocated bytes for code 195 (MethodHash=99115360) for method System.Runtime.InteropServices.SafeBuffer:Read[long](ulong):long:this (FullOpts) ```
-44 (-18.41 % of base) - System.Runtime.InteropServices.SafeBuffer:Read[short](ulong):short:this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Read[short](ulong):short:this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x28] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x18] this class-hnd EH-live single-def ; V01 arg1 [V01,T01] ( 3, 3 ) long -> rbx single-def -; V02 loc0 [V02,T02] ( 3, 3 ) int -> r15 -; V03 loc1 [V03,T03] ( 3, 3 ) long -> rbx -; V04 loc2 [V04 ] ( 3, 3 ) short -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op -; V05 loc3 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x20] do-not-enreg[X] addr-exposed ld-addr-op +; V02 loc0 [V02,T03] ( 2, 2 ) int -> rax +; V03 loc1 [V03,T02] ( 3, 3 ) long -> rbx +; V04 loc2 [V04 ] ( 4, 3 ) ubyte -> [rbp-0x10] do-not-enreg[X] addr-exposed ld-addr-op +; V05 loc3 [V05,T04] ( 2, 2 ) short -> rbx ;# V06 OutArgs [V06 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V07 tmp1 [V07 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" -; V08 PSPSym [V08,T04] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" +; V08 PSPSym [V08,T05] ( 1, 1 ) long -> [rbp-0x20] do-not-enreg[V] "PSPSym" ; -; Lcl frame size = 32 +; Lcl frame size = 24 G_M14975_IG01: push rbp - push r15 push rbx - sub rsp, 32 - lea rbp, [rsp+0x30] - mov qword ptr [rbp-0x30], rsp - mov gword ptr [rbp-0x28], rdi + sub rsp, 24 + lea rbp, [rsp+0x20] + mov qword ptr [rbp-0x20], rsp + mov gword ptr [rbp-0x18], rdi mov rbx, rsi - ;; size=24 bbWeight=1 PerfScore 6.00 + ;; size=22 bbWeight=1 PerfScore 5.00 G_M14975_IG02: cmp qword ptr [rdi+0x18], -1 - je G_M14975_IG09 + je SHORT G_M14975_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[short]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[short]():uint - mov r15d, eax - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x18] add rbx, qword ptr [rdi+0x08] - mov edx, r15d + mov edx, eax mov rsi, rbx mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi - mov dword ptr [rbp-0x18], esi - ;; size=57 bbWeight=1 PerfScore 16.50 + mov dword ptr [rbp-0x10], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M14975_IG03: - mov dword ptr [rbp-0x20], esi - ;; size=3 bbWeight=1 PerfScore 1.00 -G_M14975_IG04: - lea rsi, [rbp-0x20] - mov rdi, gword ptr [rbp-0x28] + lea rsi, [rbp-0x10] + mov rdi, gword ptr [rbp-0x18] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r15d - lea rdi, [rbp-0x18] - mov rsi, rbx - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 -G_M14975_IG05: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M14975_IG07 + movsx rbx, word ptr [rbx] + ;; size=24 bbWeight=1 PerfScore 8.75 +G_M14975_IG04: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M14975_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 -G_M14975_IG06: - mov rdi, gword ptr [rbp-0x28] +G_M14975_IG05: + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 +G_M14975_IG06: + mov eax, ebx + ;; size=2 bbWeight=1 PerfScore 0.25 G_M14975_IG07: - movsx rax, word ptr [rbp-0x18] - ;; size=5 bbWeight=1 PerfScore 3.00 -G_M14975_IG08: - add rsp, 32 + add rsp, 24 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=1 PerfScore 2.75 -G_M14975_IG09: + ;; size=7 bbWeight=1 PerfScore 2.25 +G_M14975_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 -G_M14975_IG10: +G_M14975_IG09: push rbp - push r15 push rbx - sub rsp, 16 + push rax mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp - lea rbp, [rbp+0x30] - ;; size=19 bbWeight=0 PerfScore 0.00 -G_M14975_IG11: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M14975_IG12 - mov rdi, gword ptr [rbp-0x28] + lea rbp, [rbp+0x20] + ;; size=14 bbWeight=0 PerfScore 0.00 +G_M14975_IG10: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M14975_IG11 + mov rdi, gword ptr [rbp-0x18] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 -G_M14975_IG12: +G_M14975_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 -G_M14975_IG13: - add rsp, 16 +G_M14975_IG12: + add rsp, 8 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=0 PerfScore 0.00 + ;; size=7 bbWeight=0 PerfScore 0.00 -; Total bytes of code 239, prolog size 24, PerfScore 43.75, instruction count 68, allocated bytes for code 239 (MethodHash=291bc580) for method System.Runtime.InteropServices.SafeBuffer:Read[short](ulong):short:this (FullOpts) +; Total bytes of code 195, prolog size 22, PerfScore 37.75, instruction count 57, allocated bytes for code 195 (MethodHash=291bc580) for method System.Runtime.InteropServices.SafeBuffer:Read[short](ulong):short:this (FullOpts) ```
-37 (-15.42 % of base) - System.Runtime.InteropServices.SafeBuffer:Read[double](ulong):double:this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Read[double](ulong):double:this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x28] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x20] this class-hnd EH-live single-def ; V01 arg1 [V01,T01] ( 3, 3 ) long -> rbx single-def -; V02 loc0 [V02,T02] ( 3, 3 ) int -> r15 -; V03 loc1 [V03,T03] ( 3, 3 ) long -> rbx -; V04 loc2 [V04 ] ( 3, 3 ) double -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op -; V05 loc3 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x20] do-not-enreg[X] addr-exposed ld-addr-op +; V02 loc0 [V02,T03] ( 2, 2 ) int -> rax +; V03 loc1 [V03,T02] ( 3, 3 ) long -> rbx +; V04 loc2 [V04 ] ( 4, 3 ) ubyte -> [rbp-0x10] do-not-enreg[X] addr-exposed ld-addr-op +; V05 loc3 [V05,T05] ( 2, 2 ) double -> [rbp-0x18] spill-single-def ;# V06 OutArgs [V06 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V07 tmp1 [V07 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" ; V08 PSPSym [V08,T04] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" ; -; Lcl frame size = 32 +; Lcl frame size = 40 G_M25535_IG01: push rbp - push r15 push rbx - sub rsp, 32 + sub rsp, 40 lea rbp, [rsp+0x30] mov qword ptr [rbp-0x30], rsp - mov gword ptr [rbp-0x28], rdi + mov gword ptr [rbp-0x20], rdi mov rbx, rsi - ;; size=24 bbWeight=1 PerfScore 6.00 + ;; size=22 bbWeight=1 PerfScore 5.00 G_M25535_IG02: cmp qword ptr [rdi+0x18], -1 - je G_M25535_IG09 + je SHORT G_M25535_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[double]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[double]():uint - mov r15d, eax - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x20] add rbx, qword ptr [rdi+0x08] - mov edx, r15d + mov edx, eax mov rsi, rbx mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi - mov qword ptr [rbp-0x18], rsi - ;; size=58 bbWeight=1 PerfScore 16.50 + mov dword ptr [rbp-0x10], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M25535_IG03: - mov dword ptr [rbp-0x20], esi - ;; size=3 bbWeight=1 PerfScore 1.00 -G_M25535_IG04: - lea rsi, [rbp-0x20] - mov rdi, gword ptr [rbp-0x28] + lea rsi, [rbp-0x10] + mov rdi, gword ptr [rbp-0x20] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r15d - lea rdi, [rbp-0x18] - mov rsi, rbx - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 -G_M25535_IG05: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M25535_IG07 + vmovsd xmm0, qword ptr [rbx] + vmovsd qword ptr [rbp-0x18], xmm0 + ;; size=29 bbWeight=1 PerfScore 9.75 +G_M25535_IG04: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M25535_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 -G_M25535_IG06: - mov rdi, gword ptr [rbp-0x28] +G_M25535_IG05: + mov rdi, gword ptr [rbp-0x20] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 -G_M25535_IG07: +G_M25535_IG06: vmovsd xmm0, qword ptr [rbp-0x18] ;; size=5 bbWeight=1 PerfScore 3.00 -G_M25535_IG08: - add rsp, 32 +G_M25535_IG07: + add rsp, 40 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=1 PerfScore 2.75 -G_M25535_IG09: + ;; size=7 bbWeight=1 PerfScore 2.25 +G_M25535_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 -G_M25535_IG10: +G_M25535_IG09: push rbp - push r15 push rbx - sub rsp, 16 + push rax mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp lea rbp, [rbp+0x30] - ;; size=19 bbWeight=0 PerfScore 0.00 -G_M25535_IG11: - cmp byte ptr [rbp-0x20], 0 - je SHORT G_M25535_IG12 - mov rdi, gword ptr [rbp-0x28] + ;; size=14 bbWeight=0 PerfScore 0.00 +G_M25535_IG10: + cmp byte ptr [rbp-0x10], 0 + je SHORT G_M25535_IG11 + mov rdi, gword ptr [rbp-0x20] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 -G_M25535_IG12: +G_M25535_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 -G_M25535_IG13: - add rsp, 16 +G_M25535_IG12: + add rsp, 8 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=0 PerfScore 0.00 + ;; size=7 bbWeight=0 PerfScore 0.00 -; Total bytes of code 240, prolog size 24, PerfScore 43.75, instruction count 68, allocated bytes for code 240 (MethodHash=da8b9c40) for method System.Runtime.InteropServices.SafeBuffer:Read[double](ulong):double:this (FullOpts) +; Total bytes of code 203, prolog size 22, PerfScore 41.50, instruction count 58, allocated bytes for code 203 (MethodHash=da8b9c40) for method System.Runtime.InteropServices.SafeBuffer:Read[double](ulong):double:this (FullOpts) ```
-35 (-13.16 % of base) - System.Runtime.InteropServices.SafeBuffer:Read[System.Numerics.Vector`1[float]](ulong):System.Numerics.Vector`1[float]:this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Read[System.Numerics.Vector`1[float]](ulong):System.Numerics.Vector`1[float]:this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x60] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x58] this class-hnd EH-live single-def ; V01 RetBuf [V01,T01] ( 4, 4 ) byref -> rbx single-def ; V02 arg1 [V02,T02] ( 3, 3 ) long -> r15 single-def -; V03 loc0 [V03,T03] ( 3, 3 ) int -> r14 -; V04 loc1 [V04,T04] ( 3, 3 ) long -> r15 -; V05 loc2 [V05 ] ( 3, 3 ) simd32 -> [rbp-0x50] do-not-enreg[XS] addr-exposed ld-addr-op -; V06 loc3 [V06 ] ( 4, 3 ) ubyte -> [rbp-0x58] do-not-enreg[X] addr-exposed ld-addr-op +; V03 loc0 [V03,T04] ( 2, 2 ) int -> rax +; V04 loc1 [V04,T03] ( 3, 3 ) long -> r15 +; V05 loc2 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op +; V06 loc3 [V06,T06] ( 2, 2 ) simd32 -> [rbp-0x50] spill-single-def ;# V07 OutArgs [V07 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V08 tmp1 [V08 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" -; V09 PSPSym [V09,T05] ( 1, 1 ) long -> [rbp-0x70] do-not-enreg[V] "PSPSym" +; V09 PSPSym [V09,T05] ( 1, 1 ) long -> [rbp-0x60] do-not-enreg[V] "PSPSym" ; -; Lcl frame size = 88 +; Lcl frame size = 80 G_M13951_IG01: push rbp push r15 - push r14 push rbx - sub rsp, 88 - lea rbp, [rsp+0x70] - mov qword ptr [rbp-0x70], rsp - mov gword ptr [rbp-0x60], rdi + sub rsp, 80 + lea rbp, [rsp+0x60] + mov qword ptr [rbp-0x60], rsp + mov gword ptr [rbp-0x58], rdi mov rbx, rsi mov r15, rdx - ;; size=29 bbWeight=1 PerfScore 7.25 + ;; size=27 bbWeight=1 PerfScore 6.25 G_M13951_IG02: cmp qword ptr [rdi+0x18], -1 - je G_M13951_IG08 + je SHORT G_M13951_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[System.Numerics.Vector`1[float]]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[System.Numerics.Vector`1[float]]():uint - mov r14d, eax - mov rdi, gword ptr [rbp-0x60] + mov rdi, gword ptr [rbp-0x58] add r15, qword ptr [rdi+0x08] - mov edx, r14d + mov edx, eax mov rsi, r15 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this - vxorps ymm0, ymm0, ymm0 - vmovups ymmword ptr [rbp-0x50], ymm0 xor esi, esi - mov dword ptr [rbp-0x58], esi - ;; size=66 bbWeight=1 PerfScore 17.83 + mov dword ptr [rbp-0x18], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M13951_IG03: - lea rsi, [rbp-0x58] - mov rdi, gword ptr [rbp-0x60] + lea rsi, [rbp-0x18] + mov rdi, gword ptr [rbp-0x58] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r14d - lea rdi, [rbp-0x50] - mov rsi, r15 - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 + vmovups ymm0, ymmword ptr [r15] + vmovups ymmword ptr [rbp-0x50], ymm0 + ;; size=30 bbWeight=1 PerfScore 10.75 G_M13951_IG04: - cmp byte ptr [rbp-0x58], 0 + cmp byte ptr [rbp-0x18], 0 je SHORT G_M13951_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 G_M13951_IG05: - mov rdi, gword ptr [rbp-0x60] + mov rdi, gword ptr [rbp-0x58] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 G_M13951_IG06: vmovups ymm0, ymmword ptr [rbp-0x50] vmovups ymmword ptr [rbx], ymm0 mov rax, rbx ;; size=12 bbWeight=1 PerfScore 6.25 G_M13951_IG07: vzeroupper - add rsp, 88 + add rsp, 80 pop rbx - pop r14 pop r15 pop rbp ret - ;; size=14 bbWeight=1 PerfScore 4.25 + ;; size=12 bbWeight=1 PerfScore 3.75 G_M13951_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 G_M13951_IG09: push rbp push r15 - push r14 push rbx - push rax + sub rsp, 16 mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp - lea rbp, [rbp+0x70] - ;; size=18 bbWeight=0 PerfScore 0.00 + lea rbp, [rbp+0x60] + ;; size=19 bbWeight=0 PerfScore 0.00 G_M13951_IG10: - cmp byte ptr [rbp-0x58], 0 + cmp byte ptr [rbp-0x18], 0 je SHORT G_M13951_IG11 - mov rdi, gword ptr [rbp-0x60] + mov rdi, gword ptr [rbp-0x58] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 G_M13951_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 G_M13951_IG12: vzeroupper - add rsp, 8 + add rsp, 16 pop rbx - pop r14 pop r15 pop rbp ret - ;; size=14 bbWeight=0 PerfScore 0.00 + ;; size=12 bbWeight=0 PerfScore 0.00 -; Total bytes of code 266, prolog size 29, PerfScore 50.08, instruction count 78, allocated bytes for code 266 (MethodHash=990ac980) for method System.Runtime.InteropServices.SafeBuffer:Read[System.Numerics.Vector`1[float]](ulong):System.Numerics.Vector`1[float]:this (FullOpts) +; Total bytes of code 231, prolog size 27, PerfScore 48.50, instruction count 67, allocated bytes for code 231 (MethodHash=990ac980) for method System.Runtime.InteropServices.SafeBuffer:Read[System.Numerics.Vector`1[float]](ulong):System.Numerics.Vector`1[float]:this (FullOpts) ```
-29 (-12.45 % of base) - System.Runtime.InteropServices.SafeBuffer:Write[double](ulong,double):this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Write[double](ulong,double):this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x28] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x20] this class-hnd EH-live single-def ; V01 arg1 [V01,T01] ( 3, 3 ) long -> rbx single-def -; V02 arg2 [V02 ] ( 3, 3 ) double -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op single-def -; V03 loc0 [V03,T02] ( 3, 3 ) int -> r15 -; V04 loc1 [V04,T03] ( 3, 3 ) long -> rbx -; V05 loc2 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x20] do-not-enreg[X] addr-exposed ld-addr-op +; V02 arg2 [V02,T05] ( 3, 3 ) double -> [rbp-0x10] single-def +; V03 loc0 [V03,T03] ( 2, 2 ) int -> rax +; V04 loc1 [V04,T02] ( 3, 3 ) long -> rbx +; V05 loc2 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op ;# V06 OutArgs [V06 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V07 tmp1 [V07 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" ; V08 PSPSym [V08,T04] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" ; -; Lcl frame size = 32 +; Lcl frame size = 40 G_M55238_IG01: push rbp - push r15 push rbx - sub rsp, 32 + sub rsp, 40 lea rbp, [rsp+0x30] mov qword ptr [rbp-0x30], rsp - mov gword ptr [rbp-0x28], rdi - vmovsd qword ptr [rbp-0x18], xmm0 + mov gword ptr [rbp-0x20], rdi + vmovsd qword ptr [rbp-0x10], xmm0 mov rbx, rsi - ;; size=29 bbWeight=1 PerfScore 7.00 + ;; size=27 bbWeight=1 PerfScore 6.00 G_M55238_IG02: cmp qword ptr [rdi+0x18], -1 je SHORT G_M55238_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[double]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[double]():uint - mov r15d, eax - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x20] add rbx, qword ptr [rdi+0x08] - mov edx, r15d + mov edx, eax mov rsi, rbx mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi - mov dword ptr [rbp-0x20], esi - ;; size=53 bbWeight=1 PerfScore 16.50 + mov dword ptr [rbp-0x18], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M55238_IG03: - lea rsi, [rbp-0x20] - mov rdi, gword ptr [rbp-0x28] + lea rsi, [rbp-0x18] + mov rdi, gword ptr [rbp-0x20] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r15d - lea rsi, [rbp-0x18] - mov rdi, rbx - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 + vmovsd xmm0, qword ptr [rbp-0x10] + vmovsd qword ptr [rbx], xmm0 + ;; size=29 bbWeight=1 PerfScore 9.75 G_M55238_IG04: - cmp byte ptr [rbp-0x20], 0 + cmp byte ptr [rbp-0x18], 0 je SHORT G_M55238_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 G_M55238_IG05: - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x20] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 G_M55238_IG06: nop ;; size=1 bbWeight=1 PerfScore 0.25 G_M55238_IG07: - add rsp, 32 + add rsp, 40 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=1 PerfScore 2.75 + ;; size=7 bbWeight=1 PerfScore 2.25 G_M55238_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 G_M55238_IG09: push rbp - push r15 push rbx - sub rsp, 16 + push rax mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp lea rbp, [rbp+0x30] - ;; size=19 bbWeight=0 PerfScore 0.00 + ;; size=14 bbWeight=0 PerfScore 0.00 G_M55238_IG10: - cmp byte ptr [rbp-0x20], 0 + cmp byte ptr [rbp-0x18], 0 je SHORT G_M55238_IG11 - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x20] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 G_M55238_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 G_M55238_IG12: - add rsp, 16 + add rsp, 8 pop rbx - pop r15 pop rbp ret - ;; size=9 bbWeight=0 PerfScore 0.00 + ;; size=7 bbWeight=0 PerfScore 0.00 -; Total bytes of code 233, prolog size 29, PerfScore 41.00, instruction count 68, allocated bytes for code 233 (MethodHash=67b02839) for method System.Runtime.InteropServices.SafeBuffer:Write[double](ulong,double):this (FullOpts) +; Total bytes of code 204, prolog size 27, PerfScore 39.75, instruction count 59, allocated bytes for code 204 (MethodHash=67b02839) for method System.Runtime.InteropServices.SafeBuffer:Write[double](ulong,double):this (FullOpts) ```
-25 (-10.78 % of base) - System.Runtime.InteropServices.SafeBuffer:Write[long](ulong,long):this ```diff ; Assembly listing for method System.Runtime.InteropServices.SafeBuffer:Write[long](ulong,long):this (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; 3 single block inlinees; 0 inlinees without PGO data ; Final local variable assignments ; -; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x28] this class-hnd EH-live single-def +; V00 this [V00,T00] ( 8, 6.50) ref -> [rbp-0x20] this class-hnd EH-live single-def ; V01 arg1 [V01,T01] ( 3, 3 ) long -> rbx single-def -; V02 arg2 [V02 ] ( 3, 3 ) long -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op single-def -; V03 loc0 [V03,T02] ( 3, 3 ) int -> r15 +; V02 arg2 [V02,T02] ( 3, 3 ) long -> r15 single-def +; V03 loc0 [V03,T04] ( 2, 2 ) int -> rax ; V04 loc1 [V04,T03] ( 3, 3 ) long -> rbx -; V05 loc2 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x20] do-not-enreg[X] addr-exposed ld-addr-op +; V05 loc2 [V05 ] ( 4, 3 ) ubyte -> [rbp-0x18] do-not-enreg[X] addr-exposed ld-addr-op ;# V06 OutArgs [V06 ] ( 1, 1 ) struct ( 0) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace" ;* V07 tmp1 [V07 ] ( 0, 0 ) long -> zero-ref "impAppendStmt" -; V08 PSPSym [V08,T04] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" +; V08 PSPSym [V08,T05] ( 1, 1 ) long -> [rbp-0x30] do-not-enreg[V] "PSPSym" ; ; Lcl frame size = 32 G_M39494_IG01: push rbp push r15 push rbx sub rsp, 32 lea rbp, [rsp+0x30] mov qword ptr [rbp-0x30], rsp - mov gword ptr [rbp-0x28], rdi - mov qword ptr [rbp-0x18], rdx + mov gword ptr [rbp-0x20], rdi mov rbx, rsi - ;; size=28 bbWeight=1 PerfScore 7.00 + mov r15, rdx + ;; size=27 bbWeight=1 PerfScore 6.25 G_M39494_IG02: cmp qword ptr [rdi+0x18], -1 je SHORT G_M39494_IG08 mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SizeOf[long]():uint call [rax]System.Runtime.InteropServices.SafeBuffer:SizeOf[long]():uint - mov r15d, eax - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x20] add rbx, qword ptr [rdi+0x08] - mov edx, r15d + mov edx, eax mov rsi, rbx mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this call [rax]System.Runtime.InteropServices.SafeBuffer:SpaceCheck(ulong,ulong):this xor esi, esi - mov dword ptr [rbp-0x20], esi - ;; size=53 bbWeight=1 PerfScore 16.50 + mov dword ptr [rbp-0x18], esi + ;; size=49 bbWeight=1 PerfScore 16.25 G_M39494_IG03: - lea rsi, [rbp-0x20] - mov rdi, gword ptr [rbp-0x28] + lea rsi, [rbp-0x18] + mov rdi, gword ptr [rbp-0x20] mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this call [rax]System.Runtime.InteropServices.SafeHandle:DangerousAddRef(byref):this - mov edx, r15d - lea rsi, [rbp-0x18] - mov rdi, rbx - mov rax, 0xD1FFAB1E ; code for System.SpanHelpers:Memmove(byref,byref,ulong) - call [rax]System.SpanHelpers:Memmove(byref,byref,ulong) - nop - ;; size=43 bbWeight=1 PerfScore 9.25 + mov qword ptr [rbx], r15 + ;; size=23 bbWeight=1 PerfScore 5.75 G_M39494_IG04: - cmp byte ptr [rbp-0x20], 0 + cmp byte ptr [rbp-0x18], 0 je SHORT G_M39494_IG06 ;; size=6 bbWeight=1 PerfScore 3.00 G_M39494_IG05: - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x20] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=18 bbWeight=0.50 PerfScore 2.25 G_M39494_IG06: nop ;; size=1 bbWeight=1 PerfScore 0.25 G_M39494_IG07: add rsp, 32 pop rbx pop r15 pop rbp ret ;; size=9 bbWeight=1 PerfScore 2.75 G_M39494_IG08: mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException call [rax]System.Runtime.InteropServices.SafeBuffer:NotInitialized():System.InvalidOperationException mov rdi, rax call CORINFO_HELP_THROW int3 ;; size=21 bbWeight=0 PerfScore 0.00 G_M39494_IG09: push rbp push r15 push rbx sub rsp, 16 mov rbp, qword ptr [rdi] mov qword ptr [rsp], rbp lea rbp, [rbp+0x30] ;; size=19 bbWeight=0 PerfScore 0.00 G_M39494_IG10: - cmp byte ptr [rbp-0x20], 0 + cmp byte ptr [rbp-0x18], 0 je SHORT G_M39494_IG11 - mov rdi, gword ptr [rbp-0x28] + mov rdi, gword ptr [rbp-0x20] xor esi, esi mov rax, 0xD1FFAB1E ; code for System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this call [rax]System.Runtime.InteropServices.SafeHandle:InternalRelease(ubyte):this ;; size=24 bbWeight=0 PerfScore 0.00 G_M39494_IG11: nop ;; size=1 bbWeight=0 PerfScore 0.00 G_M39494_IG12: add rsp, 16 pop rbx pop r15 pop rbp ret ;; size=9 bbWeight=0 PerfScore 0.00 -; Total bytes of code 232, prolog size 28, PerfScore 41.00, instruction count 68, allocated bytes for code 232 (MethodHash=59f965b9) for method System.Runtime.InteropServices.SafeBuffer:Write[long](ulong,long):this (FullOpts) +; Total bytes of code 207, prolog size 27, PerfScore 36.50, instruction count 62, allocated bytes for code 207 (MethodHash=59f965b9) for method System.Runtime.InteropServices.SafeBuffer:Write[long](ulong,long):this (FullOpts) ```

Larger list of diffs: https://gist.github.com/MihuBot/b3639a64f40ce798f1a8105e2ee156e3

MihuBot commented 3 months ago

@xtqqczze