[JitDiff X64] [Copilot] Replace ExtractMostSignificantBits+BitOp patterns wi ...

[MihuBot]

Job completed in 5 minutes 43 seconds (remote runner delay: 0 seconds).
dotnet/runtime#126841

Main commit: dotnet/runtime@33be435
PR commit: dotnet/runtime@9b3f39f

Diffs

Diffs

Found 85 files with textual diffs.

Summary of Code Size diffs:
(Lower is better)

Total bytes of base: 45705409
Total bytes of diff: 45705232
Total bytes of delta: -177 (-0.00 % of base)
Total relative delta: 4.69
    diff is an improvement.
    relative diff is a regression.


Total byte diff includes -499 bytes from reconciling methods
	Base had   24 unique methods,      499 unique bytes
	Diff had    0 unique methods,        0 unique bytes

Top file regressions (bytes):
         322 : System.Private.CoreLib.dasm (0.00 % of base)

Top file improvements (bytes):
        -499 : System.Numerics.Tensors.dasm (-0.04 % of base)

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

Top method regressions (bytes):
         103 (13.41 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:NonPackedIndexOfValueType[short,System.SpanHelpers+DontNegate`1[short]](byref,short,int):int (FullOpts)
          84 (14.07 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:IndexOfNullByte(ptr):int (FullOpts)
           9 (40.91 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector128`1[byte]):int (FullOpts)
           9 (36.00 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector256`1[byte]):int (FullOpts)
           9 (23.68 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector512`1[byte]):int (FullOpts)
           9 (34.62 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector128`1[double]):int (FullOpts)
           9 (31.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector256`1[double]):int (FullOpts)
           9 (21.95 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector512`1[double]):int (FullOpts)
           9 (34.62 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts)
           9 (31.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector256`1[int]):int (FullOpts)
           9 (21.95 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector512`1[int]):int (FullOpts)
           9 (34.62 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[long](byref,byref,System.Runtime.Intrinsics.Vector128`1[long]):int (FullOpts)
           9 (31.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[long](byref,byref,System.Runtime.Intrinsics.Vector256`1[long]):int (FullOpts)
           9 (21.95 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[long](byref,byref,System.Runtime.Intrinsics.Vector512`1[long]):int (FullOpts)
           9 (29.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[short](byref,byref,System.Runtime.Intrinsics.Vector128`1[short]):int (FullOpts)
           9 (26.47 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[short](byref,byref,System.Runtime.Intrinsics.Vector256`1[short]):int (FullOpts)
           9 (22.50 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[short](byref,byref,System.Runtime.Intrinsics.Vector512`1[short]):int (FullOpts)

Top method improvements (bytes):
         -31 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[byte](System.Runtime.Intrinsics.Vector512`1[byte]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[double](System.Runtime.Intrinsics.Vector512`1[double]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[int](System.Runtime.Intrinsics.Vector512`1[int]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[long](System.Runtime.Intrinsics.Vector512`1[long]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[short](System.Runtime.Intrinsics.Vector512`1[short]):int (FullOpts) (1 base, 0 diff methods)
         -24 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[short](System.Runtime.Intrinsics.Vector256`1[short]):int (FullOpts) (1 base, 0 diff methods)
         -21 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[short](System.Runtime.Intrinsics.Vector128`1[short]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.__Canon](System.Runtime.Intrinsics.Vector128`1[System.__Canon]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.__Canon](System.Runtime.Intrinsics.Vector256`1[System.__Canon]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.__Canon](System.Runtime.Intrinsics.Vector512`1[System.__Canon]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Nullable`1[int]](System.Runtime.Intrinsics.Vector128`1[System.Nullable`1[int]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Nullable`1[int]](System.Runtime.Intrinsics.Vector256`1[System.Nullable`1[int]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Nullable`1[int]](System.Runtime.Intrinsics.Vector512`1[System.Nullable`1[int]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Numerics.Vector`1[float]](System.Runtime.Intrinsics.Vector128`1[System.Numerics.Vector`1[float]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Numerics.Vector`1[float]](System.Runtime.Intrinsics.Vector256`1[System.Numerics.Vector`1[float]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Numerics.Vector`1[float]](System.Runtime.Intrinsics.Vector512`1[System.Numerics.Vector`1[float]]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[byte](System.Runtime.Intrinsics.Vector256`1[byte]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[double](System.Runtime.Intrinsics.Vector256`1[double]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[int](System.Runtime.Intrinsics.Vector256`1[int]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[long](System.Runtime.Intrinsics.Vector256`1[long]):int (FullOpts) (1 base, 0 diff methods)
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[byte](System.Runtime.Intrinsics.Vector128`1[byte]):int (FullOpts) (1 base, 0 diff methods)
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[double](System.Runtime.Intrinsics.Vector128`1[double]):int (FullOpts) (1 base, 0 diff methods)
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[int](System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts) (1 base, 0 diff methods)
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[long](System.Runtime.Intrinsics.Vector128`1[long]):int (FullOpts) (1 base, 0 diff methods)

Top method regressions (percentages):
           9 (40.91 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector128`1[byte]):int (FullOpts)
           9 (36.00 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector256`1[byte]):int (FullOpts)
           9 (34.62 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector128`1[double]):int (FullOpts)
           9 (34.62 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts)
           9 (34.62 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[long](byref,byref,System.Runtime.Intrinsics.Vector128`1[long]):int (FullOpts)
           9 (31.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector256`1[double]):int (FullOpts)
           9 (31.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector256`1[int]):int (FullOpts)
           9 (31.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[long](byref,byref,System.Runtime.Intrinsics.Vector256`1[long]):int (FullOpts)
           9 (29.03 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[short](byref,byref,System.Runtime.Intrinsics.Vector128`1[short]):int (FullOpts)
           9 (26.47 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[short](byref,byref,System.Runtime.Intrinsics.Vector256`1[short]):int (FullOpts)
           9 (23.68 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector512`1[byte]):int (FullOpts)
           9 (22.50 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[short](byref,byref,System.Runtime.Intrinsics.Vector512`1[short]):int (FullOpts)
           9 (21.95 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector512`1[double]):int (FullOpts)
           9 (21.95 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector512`1[int]):int (FullOpts)
           9 (21.95 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:ComputeFirstIndex[long](byref,byref,System.Runtime.Intrinsics.Vector512`1[long]):int (FullOpts)
          84 (14.07 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:IndexOfNullByte(ptr):int (FullOpts)
         103 (13.41 % of base) : System.Private.CoreLib.dasm - System.SpanHelpers:NonPackedIndexOfValueType[short,System.SpanHelpers+DontNegate`1[short]](byref,short,int):int (FullOpts)

Top method improvements (percentages):
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[byte](System.Runtime.Intrinsics.Vector128`1[byte]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[byte](System.Runtime.Intrinsics.Vector256`1[byte]):int (FullOpts) (1 base, 0 diff methods)
         -31 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[byte](System.Runtime.Intrinsics.Vector512`1[byte]):int (FullOpts) (1 base, 0 diff methods)
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[double](System.Runtime.Intrinsics.Vector128`1[double]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[double](System.Runtime.Intrinsics.Vector256`1[double]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[double](System.Runtime.Intrinsics.Vector512`1[double]):int (FullOpts) (1 base, 0 diff methods)
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[int](System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[int](System.Runtime.Intrinsics.Vector256`1[int]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[int](System.Runtime.Intrinsics.Vector512`1[int]):int (FullOpts) (1 base, 0 diff methods)
         -15 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[long](System.Runtime.Intrinsics.Vector128`1[long]):int (FullOpts) (1 base, 0 diff methods)
         -18 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[long](System.Runtime.Intrinsics.Vector256`1[long]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[long](System.Runtime.Intrinsics.Vector512`1[long]):int (FullOpts) (1 base, 0 diff methods)
         -21 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[short](System.Runtime.Intrinsics.Vector128`1[short]):int (FullOpts) (1 base, 0 diff methods)
         -24 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[short](System.Runtime.Intrinsics.Vector256`1[short]):int (FullOpts) (1 base, 0 diff methods)
         -30 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[short](System.Runtime.Intrinsics.Vector512`1[short]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.__Canon](System.Runtime.Intrinsics.Vector128`1[System.__Canon]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.__Canon](System.Runtime.Intrinsics.Vector256`1[System.__Canon]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.__Canon](System.Runtime.Intrinsics.Vector512`1[System.__Canon]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Nullable`1[int]](System.Runtime.Intrinsics.Vector128`1[System.Nullable`1[int]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Nullable`1[int]](System.Runtime.Intrinsics.Vector256`1[System.Nullable`1[int]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Nullable`1[int]](System.Runtime.Intrinsics.Vector512`1[System.Nullable`1[int]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Numerics.Vector`1[float]](System.Runtime.Intrinsics.Vector128`1[System.Numerics.Vector`1[float]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Numerics.Vector`1[float]](System.Runtime.Intrinsics.Vector256`1[System.Numerics.Vector`1[float]]):int (FullOpts) (1 base, 0 diff methods)
         -19 (-100.00 % of base) : System.Numerics.Tensors.dasm - System.Numerics.Tensors.TensorPrimitives:IndexOfFirstMatch[System.Numerics.Vector`1[float]](System.Runtime.Intrinsics.Vector512`1[System.Numerics.Vector`1[float]]):int (FullOpts) (1 base, 0 diff methods)

41 total methods with Code Size differences (24 improved, 17 regressed), 285138 unchanged.

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

Artifacts:

• diff-frameworks.txt (17 KB)
• LongDiffsRegressions.md (74 KB)
• ShortDiffsRegressions.md (58 KB)
• jit-diffs-main.zip (155 MB)
• jit-diffs-pr.zip (155 MB)

[MihuBot]

Top method regressions

103 (13.41 % of base) - System.SpanHelpers:NonPackedIndexOfValueType[short,System.SpanHelpers+DontNegate`1[short]](byref,short,int):int

 ; Assembly listing for method System.SpanHelpers:NonPackedIndexOfValueType[short,System.SpanHelpers+DontNegate`1[short]](byref,short,int):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rbp based frame
 ; fully interruptible
 ; No PGO data
-; 0 inlinees with PGO data; 40 single block inlinees; 0 inlinees without PGO data
+; 0 inlinees with PGO data; 40 single block inlinees; 6 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T01] ( 24, 44.50)   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T08] (  8,  8.75)   short  ->  rsi         single-def
 ;  V02 arg2         [V02,T03] ( 22, 33.75)     int  ->  rdx        
 ;  V03 loc0         [V03,T00] ( 28, 59.50)    long  ->  rcx        
-;  V04 loc1         [V04,T18] (  7, 10.50)  simd64  ->  mm1         <System.Runtime.Intrinsics.Vector512`1[short]>
-;  V05 loc2         [V05,T19] (  5,  6   )  simd64  ->  mm0         <System.Runtime.Intrinsics.Vector512`1[short]>
+;  V04 loc1         [V04,T30] (  7, 10.50)  simd64  ->  mm1         <System.Runtime.Intrinsics.Vector512`1[short]>
+;  V05 loc2         [V05,T31] (  5,  6   )  simd64  ->  mm0         <System.Runtime.Intrinsics.Vector512`1[short]>
 ;  V06 loc3         [V06,T04] (  7, 17.50)   byref  ->  rsi        
 ;  V07 loc4         [V07,T09] (  3,  5   )   byref  ->  rcx         single-def
-;  V08 loc5         [V08,T16] ( 11, 16   )  simd32  ->  mm1         <System.Runtime.Intrinsics.Vector256`1[short]>
-;  V09 loc6         [V09,T20] (  4,  5.50)  simd32  ->  mm0         <System.Runtime.Intrinsics.Vector256`1[short]>
+;  V08 loc5         [V08,T28] ( 11, 16   )  simd32  ->  mm1         <System.Runtime.Intrinsics.Vector256`1[short]>
+;  V09 loc6         [V09,T32] (  4,  5.50)  simd32  ->  mm0         <System.Runtime.Intrinsics.Vector256`1[short]>
 ;  V10 loc7         [V10,T05] (  7, 17.50)   byref  ->  rsi        
 ;  V11 loc8         [V11,T10] (  3,  5   )   byref  ->  rcx         single-def
-;  V12 loc9         [V12,T17] ( 11, 16   )  simd16  ->  mm1         <System.Runtime.Intrinsics.Vector128`1[short]>
-;  V13 loc10        [V13,T21] (  4,  5.50)  simd16  ->  mm0         <System.Runtime.Intrinsics.Vector128`1[short]>
+;  V12 loc9         [V12,T29] ( 11, 16   )  simd16  ->  mm1         <System.Runtime.Intrinsics.Vector128`1[short]>
+;  V13 loc10        [V13,T33] (  4,  5.50)  simd16  ->  mm0         <System.Runtime.Intrinsics.Vector128`1[short]>
 ;  V14 loc11        [V14,T06] (  7, 17.50)   byref  ->  rsi        
 ;  V15 loc12        [V15,T11] (  3,  5   )   byref  ->  rcx         single-def
 ;# V16 OutArgs      [V16    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
 ;* V17 tmp1         [V17    ] (  0,  0   )    mask  ->  zero-ref    "spilled call-like call argument"
 ;* V18 tmp2         [V18    ] (  0,  0   )    mask  ->  zero-ref    "spilled call-like call argument"
 ;* V19 tmp3         [V19    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V20 tmp4         [V20    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V21 tmp5         [V21    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V22 tmp6         [V22    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V23 tmp7         [V23    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V24 tmp8         [V24    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V25 tmp9         [V25    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V26 tmp10        [V26    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V27 tmp11        [V27    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V28 tmp12        [V28    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V29 tmp13        [V29    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V30 tmp14        [V30    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V31 tmp15        [V31    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V32 tmp16        [V32    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V33 tmp17        [V33    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V34 tmp18        [V34    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V35 tmp19        [V35    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V36 tmp20        [V36    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V37 tmp21        [V37    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V38 tmp22        [V38    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V39 tmp23        [V39    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V40 tmp24        [V40    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V41 tmp25        [V41    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V42 tmp26        [V42    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
 ;* V43 tmp27        [V43    ] (  0,  0   )   short  ->  zero-ref    "Inlining Arg"
 ;* V44 tmp28        [V44    ] (  0,  0   )   ubyte  ->  zero-ref    "Inlining Arg"
-;* V45 tmp29        [V45    ] (  0,  0   )  simd32  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector256`1[short]>
-;* V46 tmp30        [V46    ] (  0,  0   )  simd32  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector256`1[short]>
-;* V47 tmp31        [V47    ] (  0,  0   )  simd16  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector128`1[short]>
-;* V48 tmp32        [V48    ] (  0,  0   )  simd16  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector128`1[short]>
-;  V49 tmp33        [V49,T07] ( 15, 15   )     int  ->  rax         "Single return block return value"
-;  V50 cse0         [V50,T12] (  3,  1.50)    long  ->  rax         "CSE #06: moderate"
-;  V51 cse1         [V51,T13] (  3,  1.50)    long  ->  rax         "CSE #10: moderate"
-;  V52 cse2         [V52,T14] (  3,  1.50)    long  ->  rax         "CSE #11: moderate"
-;  V53 cse3         [V53,T02] ( 22, 46.25)     int  ->  registers   hoist multi-def "CSE #01: aggressive"
-;  V54 cse4         [V54,T15] (  3,  1.50)    mask  ->   k1         "CSE #07: moderate"
+;  V45 tmp29        [V45,T22] (  2,  1   )     int  ->  rax         "Inline return value spill temp"
+;  V46 tmp30        [V46,T12] (  3,  1.50)     int  ->  rax         "Inline stloc first use temp"
+;  V47 tmp31        [V47,T23] (  2,  1   )     int  ->  rcx         "Inline return value spill temp"
+;  V48 tmp32        [V48,T13] (  3,  1.50)     int  ->  rcx         "Inline stloc first use temp"
+;* V49 tmp33        [V49    ] (  0,  0   )  simd32  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector256`1[short]>
+;  V50 tmp34        [V50,T24] (  2,  1   )     int  ->  rax         "Inline return value spill temp"
+;  V51 tmp35        [V51,T14] (  3,  1.50)     int  ->  rax         "Inline stloc first use temp"
+;* V52 tmp36        [V52    ] (  0,  0   )  simd32  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector256`1[short]>
+;  V53 tmp37        [V53,T25] (  2,  1   )     int  ->  rdx         "Inline return value spill temp"
+;  V54 tmp38        [V54,T15] (  3,  1.50)     int  ->  rdx         "Inline stloc first use temp"
+;* V55 tmp39        [V55    ] (  0,  0   )  simd16  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector128`1[short]>
+;  V56 tmp40        [V56,T26] (  2,  1   )     int  ->  rax         "Inline return value spill temp"
+;  V57 tmp41        [V57,T16] (  3,  1.50)     int  ->  rax         "Inline stloc first use temp"
+;* V58 tmp42        [V58    ] (  0,  0   )  simd16  ->  zero-ref    "Inlining Arg" <System.Runtime.Intrinsics.Vector128`1[short]>
+;  V59 tmp43        [V59,T27] (  2,  1   )     int  ->  rdx         "Inline return value spill temp"
+;  V60 tmp44        [V60,T17] (  3,  1.50)     int  ->  rdx         "Inline stloc first use temp"
+;  V61 tmp45        [V61,T07] ( 15, 15   )     int  ->  rax         "Single return block return value"
+;  V62 cse0         [V62,T18] (  3,  1.50)    long  ->  rax         "CSE #07: moderate"
+;  V63 cse1         [V63,T19] (  3,  1.50)    long  ->  rax         "CSE #10: moderate"
+;  V64 cse2         [V64,T20] (  3,  1.50)    long  ->  rax         "CSE #11: moderate"
+;  V65 cse3         [V65,T02] ( 22, 46.25)     int  ->  registers   hoist multi-def "CSE #01: aggressive"
+;  V66 cse4         [V66,T21] (  3,  1.50)    mask  ->   k1         "CSE #06: moderate"
 ;
 ; Lcl frame size = 0
 
 G_M46731_IG01:
        push     rbp
        mov      rbp, rsp
 						;; size=4 bbWeight=1 PerfScore 1.25
 G_M46731_IG02:
        cmp      edx, 8
        jl       G_M46731_IG15
 						;; size=9 bbWeight=1 PerfScore 1.25
 G_M46731_IG03:
        cmp      edx, 32
        jge      G_M46731_IG11
        cmp      edx, 16
        jge      G_M46731_IG07
        movsx    rax, si
        vpbroadcastw xmm0, eax
        mov      rsi, rdi
        lea      eax, [rdx-0x08]
        add      rax, rax
        lea      rcx, bword ptr [rsi+rax]
        vpcmpeqw xmm1, xmm0, xmmword ptr [rdi]
        vptest   xmm1, xmm1
        jne      SHORT G_M46731_IG05
        align    [0 bytes for IG04]
 						;; size=52 bbWeight=0.50 PerfScore 6.12
 G_M46731_IG04:
        add      rsi, 16
        cmp      rsi, rcx
        ja       SHORT G_M46731_IG06
        vpcmpeqw xmm1, xmm0, xmmword ptr [rsi]
        vptest   xmm1, xmm1
        je       SHORT G_M46731_IG04
 						;; size=20 bbWeight=4 PerfScore 34.00
 G_M46731_IG05:
-       mov      rax, rsi
-       sub      rax, rdi
-       shr      rax, 1
        vpmovw2m k1, xmm1
-       kmovb    edi, k1
+       kmovb    eax, k1
 		  ;; NOP compensation instructions of 2 bytes.
-       xor      ecx, ecx
-       tzcnt    ecx, edi
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
+       mov      rcx, rsi
+       sub      rcx, rdi
+       shr      rcx, 1
        add      eax, ecx
        jmp      G_M46731_IG30
        align    [0 bytes for IG08]
-						;; size=34 bbWeight=0.50 PerfScore 4.75
+						;; size=43 bbWeight=0.50 PerfScore 5.00
 G_M46731_IG06:
        mov      edi, edx
        test     dil, 7
        je       G_M46731_IG20
        vpcmpeqw xmm1, xmm0, xmmword ptr [rcx]
        vptest   xmm1, xmm1
        je       G_M46731_IG20
-       shr      rax, 1
        vpmovw2m k1, xmm1
        kmovb    esi, k1
 		  ;; NOP compensation instructions of 2 bytes.
        xor      edx, edx
        tzcnt    edx, esi
+       mov      ecx, -1
+       cmp      edx, 32
+       cmove    edx, ecx
+       shr      rax, 1
        add      eax, edx
        jmp      G_M46731_IG30
-						;; size=55 bbWeight=0.50 PerfScore 8.75
+						;; size=66 bbWeight=0.50 PerfScore 9.12
 G_M46731_IG07:
        movsx    rax, si
        vpbroadcastw ymm0, eax
        mov      rsi, rdi
        lea      eax, [rdx-0x10]
        add      rax, rax
        lea      rcx, bword ptr [rsi+rax]
        vpcmpeqw ymm1, ymm0, ymmword ptr [rdi]
        vptest   ymm1, ymm1
        jne      SHORT G_M46731_IG09
 						;; size=34 bbWeight=0.50 PerfScore 6.38
 G_M46731_IG08:
        add      rsi, 32
        cmp      rsi, rcx
        ja       SHORT G_M46731_IG10
        vpcmpeqw ymm1, ymm0, ymmword ptr [rsi]
        vptest   ymm1, ymm1
        je       SHORT G_M46731_IG08
 						;; size=20 bbWeight=4 PerfScore 42.00
 G_M46731_IG09:
-       mov      rax, rsi
-       sub      rax, rdi
-       shr      rax, 1
        vpmovw2m k1, ymm1
-       kmovw    edi, k1
+       kmovw    eax, k1
 		  ;; NOP compensation instructions of 2 bytes.
-       xor      ecx, ecx
-       tzcnt    ecx, edi
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
+       mov      rcx, rsi
+       sub      rcx, rdi
+       shr      rcx, 1
        add      eax, ecx
        jmp      G_M46731_IG30
-       align    [6 bytes for IG12]
-						;; size=40 bbWeight=0.50 PerfScore 4.75
+       align    [14 bytes for IG12]
+						;; size=57 bbWeight=0.50 PerfScore 5.00
 G_M46731_IG10:
        mov      edi, edx
        test     dil, 15
        je       G_M46731_IG20
        vpcmpeqw ymm1, ymm0, ymmword ptr [rcx]
        vptest   ymm1, ymm1
        je       G_M46731_IG20
-       shr      rax, 1
        vpmovw2m k1, ymm1
        kmovw    esi, k1
 		  ;; NOP compensation instructions of 2 bytes.
        xor      edx, edx
        tzcnt    edx, esi
+       mov      ecx, -1
+       cmp      edx, 32
+       cmove    edx, ecx
+       shr      rax, 1
        add      eax, edx
        jmp      G_M46731_IG30
-						;; size=55 bbWeight=0.50 PerfScore 9.75
+						;; size=66 bbWeight=0.50 PerfScore 10.12
 G_M46731_IG11:
        movsx    rax, si
        vpbroadcastw zmm0, eax
        mov      rsi, rdi
        lea      eax, [rdx-0x20]
        add      rax, rax
        lea      rcx, bword ptr [rsi+rax]
        vmovups  zmm1, zmmword ptr [rdi]
        vpcmpeqw k1, zmm1, zmm0
        kortestd k1, k1
 		  ;; NOP compensation instructions of 3 bytes.
        jne      SHORT G_M46731_IG13
 						;; size=45 bbWeight=0.50 PerfScore 5.12
 G_M46731_IG12:
        add      rsi, 64
        cmp      rsi, rcx
        ja       SHORT G_M46731_IG14
        vmovups  zmm1, zmmword ptr [rsi]
        vpcmpeqw k1, zmm1, zmm0
        kortestd k1, k1
+		  ;; NOP compensation instructions of 3 bytes.
        je       SHORT G_M46731_IG12
-						;; size=28 bbWeight=4 PerfScore 32.00
+						;; size=31 bbWeight=4 PerfScore 32.00
 G_M46731_IG13:
-       mov      rax, rsi
-       sub      rax, rdi
-       shr      rax, 1
        vpcmpeqw k1, zmm1, zmm0
        kmovd    edx, k1
-       xor      ecx, ecx
-       tzcnt    rcx, rdx
+		  ;; NOP compensation instructions of 2 bytes.
+       xor      eax, eax
+       tzcnt    rax, rdx
+       mov      ecx, -1
+       cmp      eax, 64
+       cmove    eax, ecx
+       mov      rcx, rsi
+       sub      rcx, rdi
+       shr      rcx, 1
        add      eax, ecx
        jmp      G_M46731_IG30
        align    [0 bytes for IG16]
-						;; size=33 bbWeight=0.50 PerfScore 4.00
+						;; size=46 bbWeight=0.50 PerfScore 4.38
 G_M46731_IG14:
        mov      edx, edx
        test     dl, 31
        je       G_M46731_IG20
        vmovups  zmm1, zmmword ptr [rcx]
        vpcmpeqw k1, zmm1, zmm0
        kortestd k1, k1
+		  ;; NOP compensation instructions of 3 bytes.
        je       G_M46731_IG20
-       shr      rax, 1
        kmovd    edi, k1
+		  ;; NOP compensation instructions of 2 bytes.
        xor      ecx, ecx
        tzcnt    rcx, rdi
+       mov      edx, -1
+       cmp      ecx, 64
+       cmove    ecx, edx
+       shr      rax, 1
        add      eax, ecx
        jmp      G_M46731_IG30
-						;; size=55 bbWeight=0.50 PerfScore 7.50
+						;; size=71 bbWeight=0.50 PerfScore 7.88
 G_M46731_IG15:
        xor      ecx, ecx
        cmp      edx, 8
        jl       G_M46731_IG17
 						;; size=11 bbWeight=0.50 PerfScore 0.75
 G_M46731_IG16:
        add      edx, -8
        movsx    rax, word  ptr [rdi+2*rcx]
        movsx    r8, si
        cmp      eax, r8d
        je       G_M46731_IG29
        movsx    rax, word  ptr [rdi+2*rcx+0x02]
        cmp      eax, r8d
        je       G_M46731_IG26
        movsx    rax, word  ptr [rdi+2*rcx+0x04]
        cmp      eax, r8d
        je       G_M46731_IG27
        movsx    rax, word  ptr [rdi+2*rcx+0x06]
        cmp      eax, r8d
        je       G_M46731_IG28
        movsx    rax, word  ptr [rdi+2*rcx+0x08]
        cmp      eax, r8d
        je       G_M46731_IG25
        movsx    rax, word  ptr [rdi+2*rcx+0x0A]
        cmp      eax, r8d
        je       G_M46731_IG24
        movsx    rax, word  ptr [rdi+2*rcx+0x0C]
        cmp      eax, r8d
        je       G_M46731_IG23
        movsx    rax, word  ptr [rdi+2*rcx+0x0E]
        cmp      eax, r8d
-       je       SHORT G_M46731_IG22
+       je       G_M46731_IG22
        add      rcx, 8
        cmp      edx, 8
        jge      G_M46731_IG16
-						;; size=135 bbWeight=4 PerfScore 176.00
+						;; size=139 bbWeight=4 PerfScore 176.00
 G_M46731_IG17:
        cmp      edx, 4
        jl       SHORT G_M46731_IG18
        add      edx, -4
        movsx    rax, word  ptr [rdi+2*rcx]
        movsx    r8, si
        cmp      eax, r8d
-       je       SHORT G_M46731_IG29
+       je       G_M46731_IG29
        movsx    rax, word  ptr [rdi+2*rcx+0x02]
        cmp      eax, r8d
        je       SHORT G_M46731_IG26
        movsx    rax, word  ptr [rdi+2*rcx+0x04]
        cmp      eax, r8d
        je       SHORT G_M46731_IG27
        movsx    rax, word  ptr [rdi+2*rcx+0x06]
        cmp      eax, r8d
        je       SHORT G_M46731_IG28
        add      rcx, 4
-						;; size=59 bbWeight=0.50 PerfScore 11.50
+						;; size=63 bbWeight=0.50 PerfScore 11.50
 G_M46731_IG18:
        test     edx, edx
        jle      SHORT G_M46731_IG20
        movsx    r8, si
-       align    [0 bytes for IG19]
-						;; size=8 bbWeight=0.75 PerfScore 1.12
+       align    [15 bytes for IG19]
+						;; size=23 bbWeight=0.75 PerfScore 1.31
 G_M46731_IG19:
        dec      edx
        movsx    rax, word  ptr [rdi+2*rcx]
        cmp      eax, r8d
        je       SHORT G_M46731_IG29
        inc      rcx
        test     edx, edx
        jg       SHORT G_M46731_IG19
 						;; size=19 bbWeight=4 PerfScore 28.00
 G_M46731_IG20:
        mov      eax, -1
 						;; size=5 bbWeight=0.50 PerfScore 0.12
 G_M46731_IG21:
        vzeroupper 
        pop      rbp
        ret      
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG22:
        lea      eax, [rcx+0x07]
        jmp      SHORT G_M46731_IG30
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG23:
        lea      eax, [rcx+0x06]
        jmp      SHORT G_M46731_IG30
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG24:
        lea      eax, [rcx+0x05]
        jmp      SHORT G_M46731_IG30
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG25:
        lea      eax, [rcx+0x04]
        jmp      SHORT G_M46731_IG30
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG26:
        lea      eax, [rcx+0x01]
        jmp      SHORT G_M46731_IG30
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG27:
        lea      eax, [rcx+0x02]
        jmp      SHORT G_M46731_IG30
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG28:
        lea      eax, [rcx+0x03]
        jmp      SHORT G_M46731_IG30
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 G_M46731_IG29:
        mov      eax, ecx
 						;; size=2 bbWeight=0.50 PerfScore 0.12
 G_M46731_IG30:
        vzeroupper 
        pop      rbp
        ret      
 						;; size=5 bbWeight=0.50 PerfScore 1.25
 
-; Total bytes of code 768, prolog size 4, PerfScore 396.50, instruction count 207, allocated bytes for code 786 (MethodHash=bc854974) for method System.SpanHelpers:NonPackedIndexOfValueType[short,System.SpanHelpers+DontNegate`1[short]](byref,short,int):int (FullOpts)
+; Total bytes of code 871, prolog size 4, PerfScore 398.69, instruction count 223, allocated bytes for code 871 (MethodHash=bc854974) for method System.SpanHelpers:NonPackedIndexOfValueType[short,System.SpanHelpers+DontNegate`1[short]](byref,short,int):int (FullOpts)
 ; ============================================================

84 (14.07 % of base) - System.SpanHelpers:IndexOfNullByte(ptr):int

 ; Assembly listing for method System.SpanHelpers:IndexOfNullByte(ptr):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rbp based frame
 ; fully interruptible
 ; No PGO data
-; 0 inlinees with PGO data; 9 single block inlinees; 0 inlinees without PGO data
+; 0 inlinees with PGO data; 9 single block inlinees; 5 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T02] ( 23, 88   )    long  ->  rdi         single-def
 ;  V01 loc0         [V01,T00] ( 61,240.50)    long  ->  rbx        
 ;  V02 loc1         [V02,T01] ( 16,108   )    long  ->  r15        
 ;* V03 loc2         [V03    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
-;  V04 loc3         [V04,T06] (  3,  4.50)     int  ->  rcx        
+;  V04 loc3         [V04,T17] (  4,  6.50)  simd16  ->  mm0         <System.Runtime.Intrinsics.Vector128`1[byte]>
 ;* V05 loc4         [V05    ] (  0,  0   )  simd32  ->  zero-ref    <System.Runtime.Intrinsics.Vector256`1[byte]>
-;  V06 loc5         [V06,T07] (  3,  4.50)     int  ->  rax        
+;  V06 loc5         [V06,T18] (  4,  6.50)  simd32  ->  mm0         <System.Runtime.Intrinsics.Vector256`1[byte]>
 ;* V07 loc6         [V07    ] (  0,  0   )  simd64  ->  zero-ref    <System.Runtime.Intrinsics.Vector512`1[byte]>
-;  V08 loc7         [V08,T03] (  5, 34.50)    long  ->  r14        
+;  V08 loc7         [V08,T16] (  7, 51.50)  simd64  ->  mm0         <System.Runtime.Intrinsics.Vector512`1[byte]>
 ;* V09 loc8         [V09    ] (  0,  0   )  simd32  ->  zero-ref    <System.Runtime.Intrinsics.Vector256`1[byte]>
-;  V10 loc9         [V10,T08] (  3,  4.50)     int  ->  rax        
+;  V10 loc9         [V10,T19] (  4,  6.50)  simd32  ->  mm0         <System.Runtime.Intrinsics.Vector256`1[byte]>
 ;* V11 loc10        [V11    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
-;  V12 loc11        [V12,T09] (  3,  4.50)     int  ->  rax        
+;  V12 loc11        [V12,T20] (  4,  6.50)  simd16  ->  mm0         <System.Runtime.Intrinsics.Vector128`1[byte]>
 ;* V13 loc12        [V13    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
-;* V14 loc13        [V14    ] (  0,  0   )     int  ->  zero-ref   
+;* V14 loc13        [V14    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
 ;* V15 loc14        [V15    ] (  0,  0   )  simd32  ->  zero-ref    <System.Runtime.Intrinsics.Vector256`1[byte]>
-;* V16 loc15        [V16    ] (  0,  0   )     int  ->  zero-ref   
+;* V16 loc15        [V16    ] (  0,  0   )  simd32  ->  zero-ref    <System.Runtime.Intrinsics.Vector256`1[byte]>
 ;* V17 loc16        [V17    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
-;* V18 loc17        [V18    ] (  0,  0   )     int  ->  zero-ref   
+;* V18 loc17        [V18    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
 ;* V19 loc18        [V19    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
 ;* V20 loc19        [V20    ] (  0,  0   )  simd16  ->  zero-ref    <System.Runtime.Intrinsics.Vector128`1[byte]>
-;* V21 loc20        [V21    ] (  0,  0   )     int  ->  zero-ref   
-;# V22 OutArgs      [V22    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
-;* V23 tmp1         [V23    ] (  0,  0   )    long  ->  zero-ref    "Inline stloc first use temp"
-;* V24 tmp2         [V24    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
-;* V25 tmp3         [V25    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
-;* V26 tmp4         [V26    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
-;* V27 tmp5         [V27    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
-;* V28 tmp6         [V28    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
-;  V29 tmp7         [V29,T04] ( 14, 15   )     int  ->  rax         "Single return block return value"
-;  V30 cse0         [V30,T05] (  3, 12   )    long  ->  rax         "CSE #05: aggressive"
-;  V31 rat0         [V31,T11] (  3,  6   )  simd64  ->  mm0         "ReplaceWithLclVar is creating a new local variable"
-;  V32 rat1         [V32,T10] (  3, 96   )  simd64  ->  mm0         "ReplaceWithLclVar is creating a new local variable"
+;# V21 OutArgs      [V21    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;* V22 tmp1         [V22    ] (  0,  0   )    long  ->  zero-ref    "Inline stloc first use temp"
+;* V23 tmp2         [V23    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
+;  V24 tmp3         [V24,T10] (  2,  1   )     int  ->  rdi         "Inline return value spill temp"
+;  V25 tmp4         [V25,T05] (  3,  1.50)     int  ->  rdi         "Inline stloc first use temp"
+;* V26 tmp5         [V26    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
+;  V27 tmp6         [V27,T11] (  2,  1   )     int  ->  rax         "Inline return value spill temp"
+;  V28 tmp7         [V28,T06] (  3,  1.50)     int  ->  rax         "Inline stloc first use temp"
+;* V29 tmp8         [V29    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
+;  V30 tmp9         [V30,T12] (  2,  1   )     int  ->  rax         "Inline return value spill temp"
+;  V31 tmp10        [V31,T07] (  3,  1.50)     int  ->  rax         "Inline stloc first use temp"
+;* V32 tmp11        [V32    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
+;  V33 tmp12        [V33,T13] (  2,  1   )     int  ->  rax         "Inline return value spill temp"
+;  V34 tmp13        [V34,T08] (  3,  1.50)     int  ->  rax         "Inline stloc first use temp"
+;* V35 tmp14        [V35    ] (  0,  0   )    long  ->  zero-ref    "Inlining Arg"
+;  V36 tmp15        [V36,T14] (  2,  1   )     int  ->  rax         "Inline return value spill temp"
+;  V37 tmp16        [V37,T09] (  3,  1.50)     int  ->  rax         "Inline stloc first use temp"
+;  V38 tmp17        [V38,T03] ( 14, 15   )     int  ->  rax         "Single return block return value"
+;  V39 cse0         [V39,T04] (  3, 12   )    long  ->  rax         "CSE #05: aggressive"
+;  V40 rat0         [V40,T21] (  3,  6   )  simd64  ->  mm0         "ReplaceWithLclVar is creating a new local variable"
+;  V41 rat1         [V41,T15] (  3, 96   )  simd64  ->  mm0         "ReplaceWithLclVar is creating a new local variable"
 ;
-; Lcl frame size = 8
+; Lcl frame size = 0
 
 G_M44554_IG01:
        push     rbp
        push     r15
-       push     r14
        push     rbx
-       push     rax
-       lea      rbp, [rsp+0x20]
-						;; size=12 bbWeight=1 PerfScore 5.50
+       lea      rbp, [rsp+0x10]
+						;; size=9 bbWeight=1 PerfScore 3.50
 G_M44554_IG02:
        xor      ebx, ebx
        mov      eax, edi
        and      eax, 15
        neg      eax
        add      eax, 16
        and      eax, 15
        mov      r15d, eax
        cmp      r15, 8
        jae      G_M44554_IG30
 						;; size=28 bbWeight=1 PerfScore 3.00
 G_M44554_IG03:
        cmp      r15, 4
        jb       SHORT G_M44554_IG05
 						;; size=6 bbWeight=4 PerfScore 5.00
 G_M44554_IG04:
        add      r15, -4
        cmp      byte  ptr [rdi+rbx], 0
        je       G_M44554_IG31
        cmp      byte  ptr [rdi+rbx+0x01], 0
        je       G_M44554_IG33
        cmp      byte  ptr [rdi+rbx+0x02], 0
        je       G_M44554_IG35
        cmp      byte  ptr [rdi+rbx+0x03], 0
        je       G_M44554_IG37
        add      rbx, 4
 						;; size=51 bbWeight=2 PerfScore 33.00
 G_M44554_IG05:
        test     r15, r15
        je       SHORT G_M44554_IG07
        align    [0 bytes for IG06]
 						;; size=5 bbWeight=6 PerfScore 7.50
 G_M44554_IG06:
        dec      r15
        cmp      byte  ptr [rdi+rbx], 0
        je       G_M44554_IG31
        inc      rbx
        test     r15, r15
        jne      SHORT G_M44554_IG06
 						;; size=21 bbWeight=16 PerfScore 92.00
 G_M44554_IG07:
        cmp      rbx, 0xD1FFAB1E
        jae      G_M44554_IG47
        mov      eax, edi
        lea      rcx, [rax+rbx]
        test     cl, 31
        je       SHORT G_M44554_IG11
 						;; size=24 bbWeight=4 PerfScore 13.00
 G_M44554_IG08:
        vxorps   xmm0, xmm0, xmm0
        vpcmpeqb xmm0, xmm0, xmmword ptr [rdi+rbx]
-       vpmovmskb ecx, xmm0
-       test     ecx, ecx
+       vptest   xmm0, xmm0
        je       SHORT G_M44554_IG10
-						;; size=17 bbWeight=2 PerfScore 13.17
+						;; size=16 bbWeight=2 PerfScore 14.67
 G_M44554_IG09:
-       xor      eax, eax
-       tzcnt    eax, ecx
+       vpmovmskb eax, xmm0
+       xor      edi, edi
+       tzcnt    edi, eax
+       mov      eax, -1
+       cmp      edi, 32
+       cmove    edi, eax
+       mov      eax, edi
        add      eax, ebx
        jmp      G_M44554_IG19
        align    [0 bytes for IG17]
-						;; size=13 bbWeight=0.50 PerfScore 2.25
+						;; size=30 bbWeight=0.50 PerfScore 3.75
 G_M44554_IG10:
        add      rbx, 16
 						;; size=4 bbWeight=2 PerfScore 0.50
 G_M44554_IG11:
        add      rax, rbx
        test     al, 63
        je       SHORT G_M44554_IG15
 						;; size=7 bbWeight=4 PerfScore 6.00
 G_M44554_IG12:
        vxorps   ymm0, ymm0, ymm0
        vpcmpeqb ymm0, ymm0, ymmword ptr [rdi+rbx]
-       vpmovmskb eax, ymm0
-       test     eax, eax
+       vptest   ymm0, ymm0
        je       SHORT G_M44554_IG14
-						;; size=17 bbWeight=2 PerfScore 15.17
+						;; size=16 bbWeight=2 PerfScore 18.67
 G_M44554_IG13:
-       xor      edi, edi
-       tzcnt    edi, eax
-       mov      eax, edi
+       vpmovmskb edi, ymm0
+       xor      eax, eax
+       tzcnt    eax, edi
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
        add      eax, ebx
-       jmp      SHORT G_M44554_IG19
-						;; size=12 bbWeight=0.50 PerfScore 2.38
+       jmp      G_M44554_IG19
+						;; size=28 bbWeight=0.50 PerfScore 4.12
 G_M44554_IG14:
        add      rbx, 32
 						;; size=4 bbWeight=2 PerfScore 0.50
 G_M44554_IG15:
        mov      eax, ebx
        neg      eax
        add      eax, 0xD1FFAB1E
        and      eax, -64
        mov      r15d, eax
        cmp      r15, rbx
        jbe      SHORT G_M44554_IG20
 						;; size=20 bbWeight=4 PerfScore 10.00
 G_M44554_IG16:
        vmovups  zmm0, zmmword ptr [rdi+rbx]
        vptestnmb k1, zmm0, zmm0
-       kmovq    r14, k1
-       test     r14, r14
+       vpmovm2b zmm0, k1
+       vptestmb k1, zmm0, zmm0
+       kortestq k1, k1
        jne      SHORT G_M44554_IG18
-						;; size=23 bbWeight=1 PerfScore 10.25
+						;; size=32 bbWeight=1 PerfScore 14.00
 G_M44554_IG17:
        add      rbx, 64
        cmp      r15, rbx
        jbe      SHORT G_M44554_IG20
        vmovups  zmm0, zmmword ptr [rdi+rbx]
        vptestnmb k1, zmm0, zmm0
-       kmovq    r14, k1
-       test     r14, r14
+       vpmovm2b zmm0, k1
+       vptestmb k1, zmm0, zmm0
+       kortestq k1, k1
        je       SHORT G_M44554_IG17
-						;; size=32 bbWeight=16 PerfScore 188.00
+						;; size=41 bbWeight=16 PerfScore 248.00
 G_M44554_IG18:
-       xor      eax, eax
-       tzcnt    rax, r14
-       mov      eax, eax
+       vpmovb2m k1, zmm0
+       kmovq    rax, k1
+       tzcnt    rax, rax
+       mov      ecx, -1
+       cmp      eax, 64
+       cmove    eax, ecx
        add      eax, ebx
-						;; size=11 bbWeight=0.50 PerfScore 1.38
+						;; size=29 bbWeight=0.50 PerfScore 3.50
 G_M44554_IG19:
        vzeroupper 
-       add      rsp, 8
        pop      rbx
-       pop      r14
        pop      r15
        pop      rbp
        ret      
-						;; size=14 bbWeight=1 PerfScore 4.25
+						;; size=8 bbWeight=1 PerfScore 3.50
 G_M44554_IG20:
        mov      eax, ebx
        neg      eax
        add      eax, 0xD1FFAB1E
        and      eax, -32
        cmp      rax, rbx
        jbe      SHORT G_M44554_IG24
 						;; size=17 bbWeight=4 PerfScore 9.00
 G_M44554_IG21:
        vxorps   ymm0, ymm0, ymm0
        vpcmpeqb ymm0, ymm0, ymmword ptr [rdi+rbx]
-       vpmovmskb eax, ymm0
-       test     eax, eax
+       vptest   ymm0, ymm0
        je       SHORT G_M44554_IG23
-						;; size=17 bbWeight=2 PerfScore 15.17
+						;; size=16 bbWeight=2 PerfScore 18.67
 G_M44554_IG22:
-       xor      edi, edi
-       tzcnt    edi, eax
-       mov      eax, edi
+       vpmovmskb edi, ymm0
+       xor      eax, eax
+       tzcnt    eax, edi
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
        add      eax, ebx
        jmp      SHORT G_M44554_IG19
-						;; size=12 bbWeight=0.50 PerfScore 2.38
+						;; size=25 bbWeight=0.50 PerfScore 4.12
 G_M44554_IG23:
        add      rbx, 32
 						;; size=4 bbWeight=2 PerfScore 0.50
 G_M44554_IG24:
        mov      eax, ebx
        neg      eax
        add      eax, 0xD1FFAB1E
        and      eax, -16
        cmp      rax, rbx
        jbe      SHORT G_M44554_IG28
 						;; size=17 bbWeight=4 PerfScore 9.00
 G_M44554_IG25:
        vxorps   xmm0, xmm0, xmm0
        vpcmpeqb xmm0, xmm0, xmmword ptr [rdi+rbx]
-       vpmovmskb eax, xmm0
-       test     eax, eax
+       vptest   xmm0, xmm0
        je       SHORT G_M44554_IG27
-						;; size=17 bbWeight=2 PerfScore 13.17
+						;; size=16 bbWeight=2 PerfScore 14.67
 G_M44554_IG26:
+       vpmovmskb eax, xmm0
        tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
        add      eax, ebx
        jmp      SHORT G_M44554_IG19
-						;; size=8 bbWeight=0.50 PerfScore 2.12
+						;; size=23 bbWeight=0.50 PerfScore 3.50
 G_M44554_IG27:
        add      rbx, 16
 						;; size=4 bbWeight=2 PerfScore 0.50
 G_M44554_IG28:
        cmp      rbx, 0xD1FFAB1E
        jae      G_M44554_IG47
        mov      r15, rbx
        neg      r15
        add      r15, 0xD1FFAB1E
 						;; size=26 bbWeight=4 PerfScore 8.00
 G_M44554_IG29:
        cmp      r15, 8
        jb       G_M44554_IG03
 						;; size=10 bbWeight=8 PerfScore 10.00
 G_M44554_IG30:
        add      r15, -8
        cmp      byte  ptr [rdi+rbx], 0
        jne      SHORT G_M44554_IG32
 						;; size=10 bbWeight=4 PerfScore 17.00
 G_M44554_IG31:
        mov      eax, ebx
        jmp      G_M44554_IG19
 						;; size=7 bbWeight=0.50 PerfScore 1.12
 G_M44554_IG32:
        cmp      byte  ptr [rdi+rbx+0x01], 0
        jne      SHORT G_M44554_IG34
 						;; size=7 bbWeight=4 PerfScore 16.00
 G_M44554_IG33:
        lea      eax, [rbx+0x01]
        jmp      G_M44554_IG19
 						;; size=8 bbWeight=0.50 PerfScore 1.25
 G_M44554_IG34:
        cmp      byte  ptr [rdi+rbx+0x02], 0
        jne      SHORT G_M44554_IG36
 						;; size=7 bbWeight=4 PerfScore 16.00
 G_M44554_IG35:
        lea      eax, [rbx+0x02]
        jmp      G_M44554_IG19
 						;; size=8 bbWeight=0.50 PerfScore 1.25
 G_M44554_IG36:
        cmp      byte  ptr [rdi+rbx+0x03], 0
        jne      SHORT G_M44554_IG38
 						;; size=7 bbWeight=4 PerfScore 16.00
 G_M44554_IG37:
        lea      eax, [rbx+0x03]
        jmp      G_M44554_IG19
 						;; size=8 bbWeight=0.50 PerfScore 1.25
 G_M44554_IG38:
        cmp      byte  ptr [rdi+rbx+0x04], 0
        jne      SHORT G_M44554_IG40
 						;; size=7 bbWeight=4 PerfScore 16.00
 G_M44554_IG39:
        lea      eax, [rbx+0x04]
        jmp      G_M44554_IG19
 						;; size=8 bbWeight=0.50 PerfScore 1.25
 G_M44554_IG40:
        cmp      byte  ptr [rdi+rbx+0x05], 0
        jne      SHORT G_M44554_IG42
 						;; size=7 bbWeight=4 PerfScore 16.00
 G_M44554_IG41:
        lea      eax, [rbx+0x05]
        jmp      G_M44554_IG19
 						;; size=8 bbWeight=0.50 PerfScore 1.25
 G_M44554_IG42:
        cmp      byte  ptr [rdi+rbx+0x06], 0
        jne      SHORT G_M44554_IG44
 						;; size=7 bbWeight=4 PerfScore 16.00
 G_M44554_IG43:
        lea      eax, [rbx+0x06]
        jmp      G_M44554_IG19
 						;; size=8 bbWeight=0.50 PerfScore 1.25
 G_M44554_IG44:
        cmp      byte  ptr [rdi+rbx+0x07], 0
        jne      SHORT G_M44554_IG46
 						;; size=7 bbWeight=4 PerfScore 16.00
 G_M44554_IG45:
        lea      eax, [rbx+0x07]
        jmp      G_M44554_IG19
 						;; size=8 bbWeight=0.50 PerfScore 1.25
 G_M44554_IG46:
        add      rbx, 8
        jmp      G_M44554_IG29
 						;; size=9 bbWeight=4 PerfScore 9.00
 G_M44554_IG47:
        mov      rax, 0xD1FFAB1E      ; code for System.SpanHelpers:ThrowMustBeNullTerminatedString()
        call     [rax]System.SpanHelpers:ThrowMustBeNullTerminatedString()
        int3     
 						;; size=13 bbWeight=0 PerfScore 0.00
 
-; Total bytes of code 597, prolog size 12, PerfScore 630.54, instruction count 175, allocated bytes for code 603 (MethodHash=583b51f5) for method System.SpanHelpers:IndexOfNullByte(ptr):int (FullOpts)
+; Total bytes of code 681, prolog size 9, PerfScore 710.04, instruction count 187, allocated bytes for code 690 (MethodHash=583b51f5) for method System.SpanHelpers:IndexOfNullByte(ptr):int (FullOpts)
 ; ============================================================

9 (40.91 % of base) - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector128`1[byte]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector128`1[byte]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd16  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector128`1[byte]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd16  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector128`1[byte]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M36195_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M36195_IG02:
-       sub      rsi, rdi
-       mov      eax, esi
        vmovaps  xmm0, xmmword ptr [rsp+0x08]
-       vpmovmskb ecx, xmm0
-       tzcnt    ecx, ecx
-       add      eax, ecx
-						;; size=21 bbWeight=1 PerfScore 7.75
+       vpmovmskb eax, xmm0
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
+       sub      rsi, rdi
+       add      eax, esi
+						;; size=30 bbWeight=1 PerfScore 8.25
 G_M36195_IG03:
        ret      
 						;; size=1 bbWeight=1 PerfScore 1.00
 
-; Total bytes of code 22, prolog size 0, PerfScore 8.75, instruction count 7, allocated bytes for code 22 (MethodHash=fde1729c) for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector128`1[byte]):int (FullOpts)
+; Total bytes of code 31, prolog size 0, PerfScore 9.25, instruction count 9, allocated bytes for code 31 (MethodHash=fde1729c) for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector128`1[byte]):int (FullOpts)
 ; ============================================================

9 (36.00 % of base) - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector256`1[byte]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector256`1[byte]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd32  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector256`1[byte]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd32  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector256`1[byte]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M36681_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M36681_IG02:
-       sub      rsi, rdi
-       mov      eax, esi
        vmovups  ymm0, ymmword ptr [rsp+0x08]
-       vpmovmskb ecx, ymm0
-       tzcnt    ecx, ecx
-       add      eax, ecx
-						;; size=21 bbWeight=1 PerfScore 9.75
+       vpmovmskb eax, ymm0
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
+       sub      rsi, rdi
+       add      eax, esi
+						;; size=30 bbWeight=1 PerfScore 10.25
 G_M36681_IG03:
        vzeroupper 
        ret      
 						;; size=4 bbWeight=1 PerfScore 2.00
 
-; Total bytes of code 25, prolog size 0, PerfScore 11.75, instruction count 8, allocated bytes for code 25 (MethodHash=645670b6) for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector256`1[byte]):int (FullOpts)
+; Total bytes of code 34, prolog size 0, PerfScore 12.25, instruction count 10, allocated bytes for code 34 (MethodHash=645670b6) for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector256`1[byte]):int (FullOpts)
 ; ============================================================

9 (23.68 % of base) - System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector512`1[byte]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector512`1[byte]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd64  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector512`1[byte]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd64  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector512`1[byte]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M31470_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M31470_IG02:
-       sub      rsi, rdi
-       mov      eax, esi
        vmovups  zmm0, zmmword ptr [rsp+0x08]
        vpmovb2m k1, zmm0
-       kmovq    rcx, k1
-       tzcnt    rcx, rcx
-       add      eax, ecx
-						;; size=34 bbWeight=1 PerfScore 9.75
+       kmovq    rax, k1
+       tzcnt    rax, rax
+       mov      ecx, -1
+       cmp      eax, 64
+       cmove    eax, ecx
+       sub      rsi, rdi
+       add      eax, esi
+						;; size=43 bbWeight=1 PerfScore 10.25
 G_M31470_IG03:
        vzeroupper 
        ret      
 						;; size=4 bbWeight=1 PerfScore 2.00
 
-; Total bytes of code 38, prolog size 0, PerfScore 11.75, instruction count 9, allocated bytes for code 41 (MethodHash=4fcf8511) for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector512`1[byte]):int (FullOpts)
+; Total bytes of code 47, prolog size 0, PerfScore 12.25, instruction count 11, allocated bytes for code 50 (MethodHash=4fcf8511) for method System.SpanHelpers:ComputeFirstIndex[byte](byref,byref,System.Runtime.Intrinsics.Vector512`1[byte]):int (FullOpts)
 ; ============================================================

9 (34.62 % of base) - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector128`1[double]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector128`1[double]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd16  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector128`1[double]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd16  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector128`1[double]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M40611_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M40611_IG02:
+       vmovaps  xmm0, xmmword ptr [rsp+0x08]
+       vmovmskpd rax, xmm0
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
        sub      rsi, rdi
        shr      rsi, 3
-       mov      eax, esi
-       vmovaps  xmm0, xmmword ptr [rsp+0x08]
-       vmovmskpd rcx, xmm0
-       tzcnt    ecx, ecx
-       add      eax, ecx
-						;; size=25 bbWeight=1 PerfScore 8.25
+       add      eax, esi
+						;; size=34 bbWeight=1 PerfScore 8.75
 G_M40611_IG03:
        ret      
 						;; size=1 bbWeight=1 PerfScore 1.00
 
-; Total bytes of code 26, prolog size 0, PerfScore 9.25, instruction count 8, allocated bytes for code 26 (MethodHash=a59b615c) for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector128`1[double]):int (FullOpts)
+; Total bytes of code 35, prolog size 0, PerfScore 9.75, instruction count 10, allocated bytes for code 35 (MethodHash=a59b615c) for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector128`1[double]):int (FullOpts)
 ; ============================================================

9 (31.03 % of base) - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector256`1[double]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector256`1[double]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd32  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector256`1[double]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd32  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector256`1[double]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M53193_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M53193_IG02:
+       vmovups  ymm0, ymmword ptr [rsp+0x08]
+       vmovmskpd rax, ymm0
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
        sub      rsi, rdi
        shr      rsi, 3
-       mov      eax, esi
-       vmovups  ymm0, ymmword ptr [rsp+0x08]
-       vmovmskpd rcx, ymm0
-       tzcnt    ecx, ecx
-       add      eax, ecx
-						;; size=25 bbWeight=1 PerfScore 11.25
+       add      eax, esi
+						;; size=34 bbWeight=1 PerfScore 11.75
 G_M53193_IG03:
        vzeroupper 
        ret      
 						;; size=4 bbWeight=1 PerfScore 2.00
 
-; Total bytes of code 29, prolog size 0, PerfScore 13.25, instruction count 9, allocated bytes for code 29 (MethodHash=60193036) for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector256`1[double]):int (FullOpts)
+; Total bytes of code 38, prolog size 0, PerfScore 13.75, instruction count 11, allocated bytes for code 38 (MethodHash=60193036) for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector256`1[double]):int (FullOpts)
 ; ============================================================

9 (21.95 % of base) - System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector512`1[double]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector512`1[double]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd64  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector512`1[double]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd64  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector512`1[double]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M23918_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M23918_IG02:
-       sub      rsi, rdi
-       shr      rsi, 3
-       mov      eax, esi
        vmovups  zmm0, zmmword ptr [rsp+0x08]
        vpmovq2m k1, zmm0
-       kmovb    ecx, k1
-       tzcnt    rcx, rcx
-       add      eax, ecx
-						;; size=37 bbWeight=1 PerfScore 10.25
+       kmovb    eax, k1
+       tzcnt    rax, rax
+       mov      ecx, -1
+       cmp      eax, 64
+       cmove    eax, ecx
+       sub      rsi, rdi
+       shr      rsi, 3
+       add      eax, esi
+						;; size=46 bbWeight=1 PerfScore 10.75
 G_M23918_IG03:
        vzeroupper 
        ret      
 						;; size=4 bbWeight=1 PerfScore 2.00
 
-; Total bytes of code 41, prolog size 0, PerfScore 12.25, instruction count 10, allocated bytes for code 43 (MethodHash=c636a291) for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector512`1[double]):int (FullOpts)
+; Total bytes of code 50, prolog size 0, PerfScore 12.75, instruction count 12, allocated bytes for code 52 (MethodHash=c636a291) for method System.SpanHelpers:ComputeFirstIndex[double](byref,byref,System.Runtime.Intrinsics.Vector512`1[double]):int (FullOpts)
 ; ============================================================

9 (34.62 % of base) - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector128`1[int]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd16  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector128`1[int]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd16  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector128`1[int]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M22083_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M22083_IG02:
+       vmovaps  xmm0, xmmword ptr [rsp+0x08]
+       vmovmskps rax, xmm0
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
        sub      rsi, rdi
        shr      rsi, 2
-       mov      eax, esi
-       vmovaps  xmm0, xmmword ptr [rsp+0x08]
-       vmovmskps rcx, xmm0
-       tzcnt    ecx, ecx
-       add      eax, ecx
-						;; size=25 bbWeight=1 PerfScore 8.25
+       add      eax, esi
+						;; size=34 bbWeight=1 PerfScore 8.75
 G_M22083_IG03:
        ret      
 						;; size=1 bbWeight=1 PerfScore 1.00
 
-; Total bytes of code 26, prolog size 0, PerfScore 9.25, instruction count 8, allocated bytes for code 26 (MethodHash=ed37a9bc) for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts)
+; Total bytes of code 35, prolog size 0, PerfScore 9.75, instruction count 10, allocated bytes for code 35 (MethodHash=ed37a9bc) for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts)
 ; ============================================================

9 (31.03 % of base) - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector256`1[int]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector256`1[int]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd32  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector256`1[int]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd32  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector256`1[int]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M13673_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M13673_IG02:
+       vmovups  ymm0, ymmword ptr [rsp+0x08]
+       vmovmskps rax, ymm0
+       tzcnt    eax, eax
+       mov      ecx, -1
+       cmp      eax, 32
+       cmove    eax, ecx
        sub      rsi, rdi
        shr      rsi, 2
-       mov      eax, esi
-       vmovups  ymm0, ymmword ptr [rsp+0x08]
-       vmovmskps rcx, ymm0
-       tzcnt    ecx, ecx
-       add      eax, ecx
-						;; size=25 bbWeight=1 PerfScore 11.25
+       add      eax, esi
+						;; size=34 bbWeight=1 PerfScore 11.75
 G_M13673_IG03:
        vzeroupper 
        ret      
 						;; size=4 bbWeight=1 PerfScore 2.00
 
-; Total bytes of code 29, prolog size 0, PerfScore 13.25, instruction count 9, allocated bytes for code 29 (MethodHash=ec90ca96) for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector256`1[int]):int (FullOpts)
+; Total bytes of code 38, prolog size 0, PerfScore 13.75, instruction count 11, allocated bytes for code 38 (MethodHash=ec90ca96) for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector256`1[int]):int (FullOpts)
 ; ============================================================

9 (21.95 % of base) - System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector512`1[int]):int

 ; Assembly listing for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector512`1[int]):int (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rsp based frame
 ; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 0 single block inlinees; 1 inlinees without PGO data
 ; Final local variable assignments
 ;
 ;  V00 arg0         [V00,T00] (  3,  3   )   byref  ->  rdi         single-def
 ;  V01 arg1         [V01,T01] (  3,  3   )   byref  ->  rsi         single-def
-;  V02 arg2         [V02,T02] (  1,  1   )  simd64  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector512`1[int]>
+;  V02 arg2         [V02,T04] (  1,  1   )  simd64  ->  [rsp+0x08]  single-def <System.Runtime.Intrinsics.Vector512`1[int]>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V04 tmp1         [V04,T03] (  2,  2   )     int  ->  rax         "Inline return value spill temp"
+;  V05 tmp2         [V05,T02] (  3,  3   )     int  ->  rax         "Inline stloc first use temp"
 ;
 ; Lcl frame size = 0
 
 G_M8366_IG01:
 						;; size=0 bbWeight=1 PerfScore 0.00
 G_M8366_IG02:
-       sub      rsi, rdi
-       shr      rsi, 2
-       mov      eax, esi
        vmovups  zmm0, zmmword ptr [rsp+0x08]
        vpmovd2m k1, zmm0
-       kmovw    ecx, k1
-       tzcnt    rcx, rcx
-       add      eax, ecx
-						;; size=37 bbWeight=1 PerfScore 10.25
+       kmovw    eax, k1
+       tzcnt    rax, rax
+       mov      ecx, -1
+       cmp      eax, 64
+       cmove    eax, ecx
+       sub      rsi, rdi
+       shr      rsi, 2
+       add      eax, esi
+						;; size=46 bbWeight=1 PerfScore 10.75
 G_M8366_IG03:
        vzeroupper 
        ret      
 						;; size=4 bbWeight=1 PerfScore 2.00
 
-; Total bytes of code 41, prolog size 0, PerfScore 12.25, instruction count 10, allocated bytes for code 43 (MethodHash=5433df51) for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector512`1[int]):int (FullOpts)
+; Total bytes of code 50, prolog size 0, PerfScore 12.75, instruction count 12, allocated bytes for code 52 (MethodHash=5433df51) for method System.SpanHelpers:ComputeFirstIndex[int](byref,byref,System.Runtime.Intrinsics.Vector512`1[int]):int (FullOpts)
 ; ============================================================

Larger list of diffs: https://gist.github.com/MihuBot/bb7830cdee363a573fa85b2bffc79b63

[MihuBot]

@EgorBo