[JitDiff X64] [tannergooding] Update Matrix4x4 paths to use the new xplat in ...

[MihuBot]

Job completed in 21 minutes 40 seconds (remote runner delay: 1 minute 8 seconds).
dotnet/runtime#130192

Main commit: dotnet/runtime@58dfec7
PR commit: tannergooding/runtime@085db4a

Diffs

Diffs

Found 89 files with textual diffs.

Summary of Code Size diffs:
(Lower is better)

Total bytes of base: 158040784
Total bytes of diff: 158044704
Total bytes of delta: 3920 (0.00 % of base)
Total relative delta: 72.25
    diff is a regression.
    relative diff is a regression.


Total byte diff includes -1796 bytes from reconciling methods
	Base had    3 unique methods,     1937 unique bytes
	Diff had    1 unique methods,      141 unique bytes

Top file regressions (bytes):
        4739 : SixLabors.ImageSharp.dasm (0.21 % of base)

Top file improvements (bytes):
        -819 : System.Private.CoreLib.dasm (-0.01 % of base)

2 total files with Code Size differences (1 improved, 1 regressed), 352 unchanged.

Top method regressions (bytes):
         690 (5,307.69 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:Invert(byref,byref):bool (FullOpts)
         214 (9.15 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.TransformExtensions:Transform(SixLabors.ImageSharp.Processing.IImageProcessingContext,SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler):SixLabors.ImageSharp.Processing.IImageProcessingContext (FullOpts)
         211 (9.09 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.TransformExtensions:Transform(SixLabors.ImageSharp.Processing.IImageProcessingContext,SixLabors.ImageSharp.Rectangle,SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler):SixLabors.ImageSharp.Processing.IImageProcessingContext (FullOpts)
         207 (57.02 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:CheckDegenerate(System.Numerics.Matrix4x4) (FullOpts)
         177 (31.55 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (28.50 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendTranslation(SixLabors.ImageSharp.PointF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (29.26 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendTranslation(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (24.28 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (22.18 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependTranslation(SixLabors.ImageSharp.PointF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (22.63 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependTranslation(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (30.37 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.Processors.Transforms.ProjectiveTransformProcessor:.ctor(System.Numerics.Matrix4x4,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler,SixLabors.ImageSharp.Size):this (FullOpts)
         174 (22.86 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendRotationDegrees(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (23.11 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendRotationRadians(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (27.36 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendScale(float):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (25.97 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendScale(SixLabors.ImageSharp.SizeF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (26.61 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendScale(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (18.55 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependRotationDegrees(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (18.71 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependRotationRadians(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (21.22 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependScale(float):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (20.37 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependScale(SixLabors.ImageSharp.SizeF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (20.76 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependScale(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (17.77 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependSkewDegrees(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (18.26 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependSkewRadians(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (27.40 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.TransformExtensions:Transform(SixLabors.ImageSharp.Processing.IImageProcessingContext,SixLabors.ImageSharp.Rectangle,System.Numerics.Matrix4x4,SixLabors.ImageSharp.Size,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler):SixLabors.ImageSharp.Processing.IImageProcessingContext (FullOpts)
         172 (21.58 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendSkewDegrees(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         172 (22.31 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendSkewRadians(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         161 (1,238.46 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4:GetDeterminant():float:this (FullOpts)
         157 (64.08 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.Processors.Transforms.TransformUtilities:IsDegenerate(System.Numerics.Matrix4x4):bool (FullOpts)
         141 (Infinity of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:Shuffle2(System.Runtime.Intrinsics.Vector128`1[float],System.Runtime.Intrinsics.Vector128`1[float],byte,byte,byte,byte):System.Runtime.Intrinsics.Vector128`1[float] (FullOpts) (0 base, 1 diff methods)
         126 (8.97 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:Decompose(byref,byref,byref,byref):bool (FullOpts)
         108 (9.10 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:BuildMatrix(SixLabors.ImageSharp.Rectangle):System.Numerics.Matrix4x4:this (FullOpts)

Top method improvements (bytes):
       -1187 (-100.00 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:<Invert>g__SoftwareFallback|53_1(byref,byref):bool (FullOpts) (1 base, 0 diff methods)
        -707 (-100.00 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:<Invert>g__SseImpl|53_0(byref,byref):bool (FullOpts) (1 base, 0 diff methods)
         -43 (-100.00 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:ThrowPlatformNotSupportedException() (FullOpts) (1 base, 0 diff methods)

Top method regressions (percentages):
         141 (Infinity of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:Shuffle2(System.Runtime.Intrinsics.Vector128`1[float],System.Runtime.Intrinsics.Vector128`1[float],byte,byte,byte,byte):System.Runtime.Intrinsics.Vector128`1[float] (FullOpts) (0 base, 1 diff methods)
         690 (5,307.69 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:Invert(byref,byref):bool (FullOpts)
         161 (1,238.46 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4:GetDeterminant():float:this (FullOpts)
         157 (64.08 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.Processors.Transforms.TransformUtilities:IsDegenerate(System.Numerics.Matrix4x4):bool (FullOpts)
         207 (57.02 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:CheckDegenerate(System.Numerics.Matrix4x4) (FullOpts)
         177 (31.55 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (30.37 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.Processors.Transforms.ProjectiveTransformProcessor:.ctor(System.Numerics.Matrix4x4,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler,SixLabors.ImageSharp.Size):this (FullOpts)
         177 (29.26 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendTranslation(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (28.50 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendTranslation(SixLabors.ImageSharp.PointF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (27.40 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.TransformExtensions:Transform(SixLabors.ImageSharp.Processing.IImageProcessingContext,SixLabors.ImageSharp.Rectangle,System.Numerics.Matrix4x4,SixLabors.ImageSharp.Size,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler):SixLabors.ImageSharp.Processing.IImageProcessingContext (FullOpts)
         174 (27.36 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendScale(float):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (26.61 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendScale(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (25.97 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendScale(SixLabors.ImageSharp.SizeF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (24.28 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (23.11 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendRotationRadians(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (22.86 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendRotationDegrees(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (22.63 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependTranslation(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         172 (22.31 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendSkewRadians(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         177 (22.18 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependTranslation(SixLabors.ImageSharp.PointF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         172 (21.58 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendSkewDegrees(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (21.22 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependScale(float):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (20.76 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependScale(System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (20.37 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependScale(SixLabors.ImageSharp.SizeF):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (18.71 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependRotationRadians(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (18.55 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependRotationDegrees(float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (18.26 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependSkewRadians(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         174 (17.77 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:PrependSkewDegrees(float,float,System.Numerics.Vector2):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
         214 (9.15 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.TransformExtensions:Transform(SixLabors.ImageSharp.Processing.IImageProcessingContext,SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler):SixLabors.ImageSharp.Processing.IImageProcessingContext (FullOpts)
         108 (9.10 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:BuildMatrix(SixLabors.ImageSharp.Rectangle):System.Numerics.Matrix4x4:this (FullOpts)
         211 (9.09 % of base) : SixLabors.ImageSharp.dasm - SixLabors.ImageSharp.Processing.TransformExtensions:Transform(SixLabors.ImageSharp.Processing.IImageProcessingContext,SixLabors.ImageSharp.Rectangle,SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder,SixLabors.ImageSharp.Processing.Processors.Transforms.IResampler):SixLabors.ImageSharp.Processing.IImageProcessingContext (FullOpts)
         126 (8.97 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:Decompose(byref,byref,byref,byref):bool (FullOpts)

Top method improvements (percentages):
       -1187 (-100.00 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:<Invert>g__SoftwareFallback|53_1(byref,byref):bool (FullOpts) (1 base, 0 diff methods)
        -707 (-100.00 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:<Invert>g__SseImpl|53_0(byref,byref):bool (FullOpts) (1 base, 0 diff methods)
         -43 (-100.00 % of base) : System.Private.CoreLib.dasm - System.Numerics.Matrix4x4+Impl:ThrowPlatformNotSupportedException() (FullOpts) (1 base, 0 diff methods)

34 total methods with Code Size differences (3 improved, 31 regressed), 815792 unchanged.

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

Artifacts:

• PackageList.json (33 KB)
• diff-frameworks.txt (16 KB)
• ShortDiffsRegressions.md (53 KB)
• LongDiffsRegressions.md (695 KB)
• jit-diffs-pr.7z (141 MB)
• jit-diffs-main.7z (141 MB)

[MihuBot]

Top method regressions

690 (5,307.69 % of base) - System.Numerics.Matrix4x4+Impl:Invert(byref,byref):bool

 ; Assembly listing for method System.Numerics.Matrix4x4+Impl:Invert(byref,byref):bool (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
-; rsp based frame
-; fully interruptible
+; rbp based frame
+; partially interruptible
 ; No PGO data
+; 0 inlinees with PGO data; 3 single block inlinees; 20 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 OutArgs      [V02    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V00 arg0         [V00,T01] (  6,  6   )   byref  ->  rdi         single-def
+;  V01 arg1         [V01,T00] ( 10,  6   )   byref  ->  rsi         single-def
+;  V02 loc0         [V02,T06] ( 12, 12   )  simd16  ->  mm0         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V03 loc1         [V03,T09] ( 11, 11   )  simd16  ->  mm1         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V04 loc2         [V04,T10] ( 11, 11   )  simd16  ->  mm2         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V05 loc3         [V05,T11] ( 11, 11   )  simd16  ->  mm3         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V06 loc4         [V06,T25] (  3,  3   )  simd16  ->  mm4         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V07 loc5         [V07,T26] (  3,  3   )  simd16  ->  mm5         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V08 loc6         [V08,T27] (  3,  3   )  simd16  ->  mm3         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V09 loc7         [V09,T03] ( 13, 13   )  simd16  ->  mm4         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V10 loc8         [V10,T18] ( 10, 10   )  simd16  ->  registers   <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V11 loc9         [V11,T04] ( 13, 13   )  simd16  ->  registers   <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V12 loc10        [V12,T02] ( 16, 16   )  simd16  ->  registers   <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V13 loc11        [V13,T05] ( 13, 13   )  simd16  ->  registers   <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V14 loc12        [V14,T19] ( 10, 10   )  simd16  ->  registers   <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V15 loc13        [V15,T12] ( 11, 11   )  simd16  ->  mm5         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V16 loc14        [V16,T13] ( 11, 11   )  simd16  ->  mm7         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V17 loc15        [V17,T14] ( 11, 11   )  simd16  ->  mm9         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V18 loc16        [V18,T07] ( 12, 12   )  simd16  ->  registers   <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V19 loc17        [V19,T20] (  9,  9   )  simd16  ->  registers   <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V20 loc18        [V20,T08] ( 12, 11.50)  simd16  ->  mm16         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V21 loc19        [V21,T15] ( 11, 10.50)  simd16  ->  mm17         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V22 loc20        [V22,T16] ( 11, 10.50)  simd16  ->  mm18         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V23 loc21        [V23,T17] ( 11, 10.50)  simd16  ->  mm14         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V24 loc22        [V24,T31] (  2,  2   )  simd16  ->  mm1         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V25 loc23        [V25,T32] (  2,  2   )  simd16  ->  mm2         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V26 loc24        [V26,T33] (  2,  2   )  simd16  ->  mm3         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V27 loc25        [V27,T34] (  2,  2   )  simd16  ->  mm4         <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V28 loc26        [V28,T30] (  3,  2.50)   float  ->  mm0        
+;  V29 loc27        [V29,T28] (  5,  2.50)  simd16  ->  mm0         <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V30 loc28        [V30    ] (  0,  0   )  simd16  ->  zero-ref    <System.Numerics.Vector4>
+;# V31 OutArgs      [V31    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
+;  V32 tmp1         [V32,T22] (  3,  6   )  simd16  ->  mm1         "dup spill"
+;* V33 tmp2         [V33    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V34 tmp3         [V34    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V35 tmp4         [V35    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V36 tmp5         [V36    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V37 tmp6         [V37    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V38 tmp7         [V38    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V39 tmp8         [V39    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V40 tmp9         [V40    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V41 tmp10        [V41    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V42 tmp11        [V42    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V43 tmp12        [V43    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V44 tmp13        [V44    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V45 tmp14        [V45    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V46 tmp15        [V46    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V47 tmp16        [V47    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V48 tmp17        [V48    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V49 tmp18        [V49    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V50 tmp19        [V50    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V51 tmp20        [V51    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V52 tmp21        [V52    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V53 cse0         [V53,T21] (  8,  8   )  simd16  ->  mm13         "CSE #01: moderate"
+;  V54 cse1         [V54,T29] (  5,  2.50)  simd16  ->  mm0         "CSE #02: moderate"
+;  V55 rat0         [V55,T23] (  3,  6   )  simd16  ->  mm0         "ReplaceWithLclVar is creating a new local variable"
+;  V56 rat1         [V56,T24] (  3,  6   )  simd16  ->  mm0         "ReplaceWithLclVar is creating a new local variable"
 ;
 ; Lcl frame size = 0
 
 G_M1478_IG01:
-						;; size=0 bbWeight=1 PerfScore 0.00
+       push     rbp
+       mov      rbp, rsp
+						;; size=4 bbWeight=1 PerfScore 1.25
 G_M1478_IG02:
-       mov      rax, 0xD1FFAB1E      ; code for System.Numerics.Matrix4x4+Impl:<Invert>g__SseImpl|53_0(byref,byref):bool
-						;; size=10 bbWeight=1 PerfScore 0.25
+       vmovups  xmm0, xmmword ptr [rdi]
+       vmovups  xmm1, xmmword ptr [rdi+0x10]
+       vmovups  xmm2, xmmword ptr [rdi+0x20]
+       vmovups  xmm3, xmmword ptr [rdi+0x30]
+       vmovhlps xmm4, xmm1, xmm0
+       vmovlhps xmm5, xmm2, xmm3
+       vmovhlps xmm3, xmm3, xmm2
+       vmovlhps xmm1, xmm0, xmm1
+       vshufps  xmm0, xmm1, xmm5, -120
+       vshufps  xmm1, xmm1, xmm5, -35
+       vshufps  xmm2, xmm4, xmm3, -120
+       vshufps  xmm3, xmm4, xmm3, -35
+       vpermilps xmm4, xmm2, 80
+       vpermilps xmm5, xmm3, -18
+       vpermilps xmm6, xmm0, 80
+       vpermilps xmm7, xmm1, -18
+       vshufps  xmm8, xmm2, xmm0, -120
+       vshufps  xmm9, xmm3, xmm1, -35
+       vmulps   xmm5, xmm4, xmm5
+       vmulps   xmm7, xmm7, xmm6
+       vmulps   xmm9, xmm8, xmm9
+       vpermilps xmm4, xmm2, -18
+       vpermilps xmm6, xmm3, 80
+       vpermilps xmm8, xmm0, -18
+       vpermilps xmm10, xmm1, 80
+       vshufps  xmm11, xmm2, xmm0, -35
+       vshufps  xmm12, xmm3, xmm1, -120
+       vbroadcastss xmm13, dword ptr [reloc @RWD00]
+       vxorps   xmm4, xmm13, xmm4
+       vfmadd231ps xmm5, xmm6, xmm4
+       vxorps   xmm6, xmm13, xmm8
+       vfmadd231ps xmm7, xmm10, xmm6
+       vxorps   xmm10, xmm13, xmm11
+       vfmadd231ps xmm9, xmm12, xmm10
+       vshufps  xmm10, xmm5, xmm9, 93
+       vpermilps xmm4, xmm1, 73
+       vshufps  xmm6, xmm10, xmm5, 50
+       vpermilps xmm8, xmm0, 18
+       vshufps  xmm10, xmm10, xmm5, -103
+       vshufps  xmm14, xmm7, xmm9, -3
+       vpermilps xmm11, xmm3, 73
+       vshufps  xmm12, xmm14, xmm7, 50
+       vpermilps xmm15, xmm2, 18
+       vshufps  xmm14, xmm14, xmm7, -103
+       vmulps   xmm16, xmm4, xmm6
+       vmulps   xmm17, xmm10, xmm8
+       vmulps   xmm18, xmm11, xmm12
+       vmulps   xmm14, xmm14, xmm15
+       vshufps  xmm10, xmm5, xmm9, 4
+       vpermilps xmm4, xmm1, -98
+						;; size=265 bbWeight=1 PerfScore 73.00
 G_M1478_IG03:
-       tail.jmp [rax]System.Numerics.Matrix4x4+Impl:<Invert>g__SseImpl|53_0(byref,byref):bool
-						;; size=3 bbWeight=1 PerfScore 2.00
+       vshufps  xmm6, xmm5, xmm10, -109
+       vpermilps xmm8, xmm0, 123
+       vshufps  xmm10, xmm5, xmm10, 38
+       vshufps  xmm11, xmm7, xmm9, -92
+       vpermilps xmm12, xmm3, -98
+       vshufps  xmm15, xmm7, xmm11, -109
+       vpermilps xmm19, xmm2, 123
+       vshufps  xmm11, xmm7, xmm11, 38
+       vxorps   xmm4, xmm13, xmm4
+       vfmadd231ps xmm16, xmm6, xmm4
+       vxorps   xmm6, xmm13, xmm8
+       vfmadd231ps xmm17, xmm10, xmm6
+       vxorps   xmm8, xmm13, xmm12
+       vfmadd231ps xmm18, xmm15, xmm8
+       vxorps   xmm10, xmm13, xmm19
+       vfmadd231ps xmm14, xmm11, xmm10
+       vpermilps xmm4, xmm1, 51
+       vshufps  xmm12, xmm5, xmm9, 74
+       vpermilps xmm6, xmm12, 44
+       vpermilps xmm8, xmm0, -115
+       vshufps  xmm15, xmm5, xmm9, 76
+       vpermilps xmm10, xmm15, -109
+       vpermilps xmm12, xmm3, 51
+       vshufps  xmm19, xmm7, xmm9, -22
+       vpermilps xmm15, xmm19, 44
+       vpermilps xmm19, xmm2, -115
+       vshufps  xmm11, xmm7, xmm9, -20
+       vpermilps xmm11, xmm11, -109
+       vmulps   xmm4, xmm4, xmm6
+       vmulps   xmm8, xmm10, xmm8
+       vmulps   xmm12, xmm12, xmm15
+       vmulps   xmm19, xmm11, xmm19
+       vsubps   xmm1, xmm16, xmm4
+       vaddps   xmm16, xmm16, xmm4
+       vaddps   xmm2, xmm8, xmm17
+       vsubps   xmm17, xmm17, xmm8
+       vsubps   xmm3, xmm18, xmm12
+       vaddps   xmm18, xmm12, xmm18
+       vaddps   xmm4, xmm14, xmm19
+       vsubps   xmm14, xmm14, xmm19
+       vshufps  xmm16, xmm16, xmm1, -40
+       vshufps  xmm17, xmm17, xmm2, -40
+       vshufps  xmm18, xmm18, xmm3, -40
+       vshufps  xmm14, xmm14, xmm4, -40
+       vpermilps xmm16, xmm16, -40
+       vpermilps xmm17, xmm17, -40
+       vpermilps xmm18, xmm18, -40
+       vpermilps xmm14, xmm14, -40
+       vmulps   xmm0, xmm16, xmm0
+						;; size=294 bbWeight=1 PerfScore 80.33
+G_M1478_IG04:
+       vpermilps xmm1, xmm0, -79
+       vaddps   xmm0, xmm1, xmm0
+       vpermilps xmm1, xmm0, 78
+       vaddps   xmm0, xmm1, xmm0
+       vandps   xmm1, xmm0, xmmword ptr [reloc @RWD04]
+       vmovss   xmm2, dword ptr [reloc @RWD20]
+       vucomiss xmm2, xmm1
+       ja       SHORT G_M1478_IG07
+						;; size=42 bbWeight=1 PerfScore 15.00
+G_M1478_IG05:
+       vbroadcastss xmm0, xmm0
+       vbroadcastss xmm1, dword ptr [reloc @RWD24]
+       vdivps   xmm0, xmm1, xmm0
+       vmulps   xmm1, xmm16, xmm0
+       vmovups  xmmword ptr [rsi], xmm1
+       vmulps   xmm1, xmm17, xmm0
+       vmovups  xmmword ptr [rsi+0x10], xmm1
+       vmulps   xmm1, xmm18, xmm0
+       vmovups  xmmword ptr [rsi+0x20], xmm1
+       vmulps   xmm0, xmm14, xmm0
+       vmovups  xmmword ptr [rsi+0x30], xmm0
+       mov      eax, 1
+						;; size=64 bbWeight=0.50 PerfScore 14.62
+G_M1478_IG06:
+       pop      rbp
+       ret      
+						;; size=2 bbWeight=0.50 PerfScore 0.75
+G_M1478_IG07:
+       vbroadcastss xmm0, dword ptr [reloc @RWD28]
+       vmovups  xmmword ptr [rsi], xmm0
+       vmovups  xmmword ptr [rsi+0x10], xmm0
+       vmovups  xmmword ptr [rsi+0x20], xmm0
+       vmovups  xmmword ptr [rsi+0x30], xmm0
+       xor      eax, eax
+						;; size=30 bbWeight=0.50 PerfScore 3.12
+G_M1478_IG08:
+       pop      rbp
+       ret      
+						;; size=2 bbWeight=0.50 PerfScore 0.75
+RWD00  	dd	80000000h		;        -0
+RWD04  	dq	7FFFFFFF7FFFFFFFh, 7FFFFFFF7FFFFFFFh
+RWD20  	dd	00000001h		; 1.4013e-45
+RWD24  	dd	3F800000h		;         1
+RWD28  	dd	FFC00000h		;      -nan
 
-; Total bytes of code 13, prolog size 0, PerfScore 2.25, instruction count 2, allocated bytes for code 13 (MethodHash=46b7fa39) for method System.Numerics.Matrix4x4+Impl:Invert(byref,byref):bool (FullOpts)
+
+; Total bytes of code 703, prolog size 4, PerfScore 188.83, instruction count 131, allocated bytes for code 703 (MethodHash=46b7fa39) for method System.Numerics.Matrix4x4+Impl:Invert(byref,byref):bool (FullOpts)
 ; ============================================================

207 (57.02 % of base) - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:CheckDegenerate(System.Numerics.Matrix4x4)

 ; Assembly listing for method SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:CheckDegenerate(System.Numerics.Matrix4x4) (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rbp based frame
 ; partially interruptible
 ; No PGO data
-; 0 inlinees with PGO data; 20 single block inlinees; 3 inlinees without PGO data
+; 0 inlinees with PGO data; 23 single block inlinees; 4 inlinees without PGO data
 ; Final local variable assignments
 ;
-;  V00 arg0         [V00,T02] (  1,  1   )  struct (64) [rbp+0x10]  do-not-enreg[S] single-def <System.Numerics.Matrix4x4>
+;  V00 arg0         [V00,T01] (  4,  4   )  struct (64) [rbp+0x10]  do-not-enreg[SF] single-def <System.Numerics.Matrix4x4>
 ;# V01 OutArgs      [V01    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
-;  V02 tmp1         [V02,T20] (  4,  0   )     ref  ->  rbx         class-hnd exact single-def "NewObj constructor temp" <SixLabors.ImageSharp.Processing.Processors.Transforms.DegenerateTransformException>
-;  V03 tmp2         [V03,T01] (  2,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
-;  V04 tmp3         [V04    ] ( 18, 36   )  struct (64) [rbp-0x48]  do-not-enreg[XS] addr-exposed ld-addr-op "Inlining Arg" <System.Numerics.Matrix4x4>
-;* V05 tmp4         [V05,T21] (  0,  0   )   ubyte  ->  zero-ref    "Inline return value spill temp"
+;  V02 tmp1         [V02,T40] (  4,  0   )     ref  ->  rbx         class-hnd exact single-def "NewObj constructor temp" <SixLabors.ImageSharp.Processing.Processors.Transforms.DegenerateTransformException>
+;  V03 tmp2         [V03,T03] (  2,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
+;  V04 tmp3         [V04,T00] ( 20, 40   )  struct (64) [rbp-0x48]  do-not-enreg[SF] ld-addr-op "Inlining Arg" <System.Numerics.Matrix4x4>
+;* V05 tmp4         [V05,T41] (  0,  0   )   ubyte  ->  zero-ref    "Inline return value spill temp"
 ;* V06 tmp5         [V06    ] (  0,  0   )  struct (64) zero-ref    do-not-enreg[SF] "Inlining Arg" <System.Numerics.Matrix4x4>
 ;* V07 tmp6         [V07    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V08 tmp7         [V08    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V09 tmp8         [V09    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V10 tmp9         [V10    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V11 tmp10        [V11    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V12 tmp11        [V12    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V13 tmp12        [V13    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V14 tmp13        [V14    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V15 tmp14        [V15    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V16 tmp15        [V16    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V17 tmp16        [V17    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V18 tmp17        [V18    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V19 tmp18        [V19    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V20 tmp19        [V20    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V21 tmp20        [V21    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V22 tmp21        [V22    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
-;  V23 tmp22        [V23,T00] (  3,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
-;  V24 tmp23        [V24,T03] (  3,  5   )   float  ->  mm0         "Inlining Arg"
-;  V25 tmp24        [V25,T04] (  3,  3   )   float  ->  mm0         "V06.[000..004)"
-;  V26 tmp25        [V26,T05] (  3,  2   )   float  ->  mm1         "V06.[004..008)"
-;  V27 tmp26        [V27,T06] (  3,  2   )   float  ->  mm2         "V06.[008..012)"
-;  V28 tmp27        [V28,T07] (  3,  2   )   float  ->  mm3         "V06.[012..016)"
-;  V29 tmp28        [V29,T08] (  3,  2   )   float  ->  mm4         "V06.[016..020)"
-;  V30 tmp29        [V30,T09] (  3,  2   )   float  ->  mm5         "V06.[020..024)"
-;  V31 tmp30        [V31,T10] (  3,  2   )   float  ->  mm6         "V06.[024..028)"
-;  V32 tmp31        [V32,T11] (  3,  2   )   float  ->  mm7         "V06.[028..032)"
-;  V33 tmp32        [V33,T12] (  3,  2   )   float  ->  mm8         "V06.[032..036)"
-;  V34 tmp33        [V34,T13] (  3,  2   )   float  ->  mm9         "V06.[036..040)"
-;  V35 tmp34        [V35,T14] (  3,  2   )   float  ->  mm10         "V06.[040..044)"
-;  V36 tmp35        [V36,T15] (  3,  2   )   float  ->  mm11         "V06.[044..048)"
-;  V37 tmp36        [V37,T16] (  3,  2   )   float  ->  mm12         "V06.[048..052)"
-;  V38 tmp37        [V38,T17] (  3,  2   )   float  ->  mm13         "V06.[052..056)"
-;  V39 tmp38        [V39,T18] (  3,  2   )   float  ->  mm14         "V06.[056..060)"
-;  V40 tmp39        [V40,T19] (  3,  2   )   float  ->  mm15         "V06.[060..064)"
+;* V23 tmp22        [V23    ] (  0,  0   )  simd16  ->  zero-ref    "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V24 tmp23        [V24    ] (  0,  0   )  simd16  ->  zero-ref    "dup spill"
+;  V25 tmp24        [V25,T29] (  2,  2   )  simd16  ->  mm4         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V26 tmp25        [V26,T30] (  2,  2   )  simd16  ->  mm5         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V27 tmp26        [V27,T31] (  2,  2   )  simd16  ->  mm6         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V28 tmp27        [V28,T32] (  2,  2   )  simd16  ->  mm7         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V29 tmp28        [V29,T33] (  2,  2   )  simd16  ->  mm8         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V30 tmp29        [V30,T34] (  2,  2   )  simd16  ->  mm2         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V31 tmp30        [V31,T35] (  2,  2   )  simd16  ->  mm9         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V32 tmp31        [V32,T36] (  2,  2   )  simd16  ->  mm3         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V33 tmp32        [V33    ] (  0,  0   )  simd16  ->  zero-ref    "dup spill"
+;  V34 tmp33        [V34,T37] (  2,  2   )  simd16  ->  mm10         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V35 tmp34        [V35,T38] (  2,  2   )  simd16  ->  mm11         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V36 tmp35        [V36    ] (  0,  0   )  simd16  ->  zero-ref    "impSpillStackEnsure"
+;  V37 tmp36        [V37,T39] (  2,  2   )  simd16  ->  mm1         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V38 tmp37        [V38,T10] (  3,  3   )  simd16  ->  mm2         "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V39 tmp38        [V39,T06] (  3,  6   )  simd16  ->  mm3         "dup spill"
+;* V40 tmp39        [V40    ] (  0,  0   )  simd16  ->  zero-ref    "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V41 tmp40        [V41    ] (  0,  0   )  simd16  ->  zero-ref    "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V42 tmp41        [V42    ] (  0,  0   )  simd16  ->  zero-ref    "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V43 tmp42        [V43,T07] (  3,  6   )  simd16  ->  mm0         "dup spill"
+;  V44 tmp43        [V44,T11] (  3,  3   )  simd16  ->  mm0         ld-addr-op "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V45 tmp44        [V45    ] (  0,  0   )   float  ->  zero-ref    "impAppendStmt"
+;* V46 tmp45        [V46    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V47 tmp46        [V47,T02] (  3,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
+;  V48 tmp47        [V48,T09] (  3,  5   )   float  ->  mm0         "Inlining Arg"
+;  V49 tmp48        [V49,T12] (  3,  3   )  simd16  ->  mm0         "V04.[000..016)"
+;  V50 tmp49        [V50,T08] (  5,  5   )  simd16  ->  mm1         "V04.[016..032)"
+;  V51 tmp50        [V51,T04] (  6,  6   )  simd16  ->  mm2         "V04.[032..048)"
+;  V52 tmp51        [V52,T05] (  6,  6   )  simd16  ->  mm3         "V04.[048..064)"
+;  V53 tmp52        [V53,T13] (  3,  3   )   float  ->  mm4         "V06.[000..004)"
+;  V54 tmp53        [V54,T14] (  3,  2   )   float  ->  mm5         "V06.[004..008)"
+;  V55 tmp54        [V55,T15] (  3,  2   )   float  ->  mm6         "V06.[008..012)"
+;  V56 tmp55        [V56,T16] (  3,  2   )   float  ->  mm7         "V06.[012..016)"
+;  V57 tmp56        [V57,T17] (  3,  2   )   float  ->  mm8         "V06.[016..020)"
+;  V58 tmp57        [V58,T18] (  3,  2   )   float  ->  mm9         "V06.[020..024)"
+;  V59 tmp58        [V59,T19] (  3,  2   )   float  ->  mm10         "V06.[024..028)"
+;  V60 tmp59        [V60,T20] (  3,  2   )   float  ->  mm11         "V06.[028..032)"
+;  V61 tmp60        [V61,T21] (  3,  2   )   float  ->  mm12         "V06.[032..036)"
+;  V62 tmp61        [V62,T22] (  3,  2   )   float  ->  mm13         "V06.[036..040)"
+;  V63 tmp62        [V63,T23] (  3,  2   )   float  ->  mm14         "V06.[040..044)"
+;  V64 tmp63        [V64,T24] (  3,  2   )   float  ->  mm15         "V06.[044..048)"
+;  V65 tmp64        [V65,T25] (  3,  2   )   float  ->  mm16         "V06.[048..052)"
+;  V66 tmp65        [V66,T26] (  3,  2   )   float  ->  mm17         "V06.[052..056)"
+;  V67 tmp66        [V67,T27] (  3,  2   )   float  ->  mm18         "V06.[056..060)"
+;  V68 tmp67        [V68,T28] (  3,  2   )   float  ->  mm19         "V06.[060..064)"
 ;
 ; Lcl frame size = 72
 
 G_M31528_IG01:
        push     rbp
        push     rbx
        sub      rsp, 72
        lea      rbp, [rsp+0x50]
 						;; size=11 bbWeight=1 PerfScore 2.75
 G_M31528_IG02:
-       vmovdqu32 zmm0, zmmword ptr [rbp+0x10]
-       vmovdqu32 zmmword ptr [rbp-0x48], zmm0
-       vmovss   xmm0, dword ptr [rbp-0x48]
-       vmovss   xmm1, dword ptr [rbp-0x44]
-       vmovss   xmm2, dword ptr [rbp-0x40]
-       vmovss   xmm3, dword ptr [rbp-0x3C]
-       vmovss   xmm4, dword ptr [rbp-0x38]
-       vmovss   xmm5, dword ptr [rbp-0x34]
-       vmovss   xmm6, dword ptr [rbp-0x30]
-       vmovss   xmm7, dword ptr [rbp-0x2C]
-       vmovss   xmm8, dword ptr [rbp-0x28]
-       vmovss   xmm9, dword ptr [rbp-0x24]
-       vmovss   xmm10, dword ptr [rbp-0x20]
-       vmovss   xmm11, dword ptr [rbp-0x1C]
-       vmovss   xmm12, dword ptr [rbp-0x18]
-       vmovss   xmm13, dword ptr [rbp-0x14]
-       vmovss   xmm14, dword ptr [rbp-0x10]
-       vmovss   xmm15, dword ptr [rbp-0x0C]
-       vucomiss xmm0, xmm0
-       jp       G_M31528_IG09
-						;; size=110 bbWeight=1 PerfScore 39.00
-G_M31528_IG03:
-       vucomiss xmm1, xmm1
-       jp       G_M31528_IG09
-       vucomiss xmm2, xmm2
-       jp       G_M31528_IG09
-       vucomiss xmm3, xmm3
-       jp       G_M31528_IG09
+       vmovups  xmm0, xmmword ptr [rbp+0x10]
+       vmovups  xmm1, xmmword ptr [rbp+0x20]
+       vmovups  xmm2, xmmword ptr [rbp+0x30]
+       vmovups  xmm3, xmmword ptr [rbp+0x40]
+       vmovups  xmmword ptr [rbp-0x48], xmm0
+       vmovups  xmmword ptr [rbp-0x38], xmm1
+       vmovups  xmmword ptr [rbp-0x28], xmm2
+       vmovups  xmmword ptr [rbp-0x18], xmm3
+       vmovss   xmm4, dword ptr [rbp-0x48]
+       vmovss   xmm5, dword ptr [rbp-0x44]
+       vmovss   xmm6, dword ptr [rbp-0x40]
+       vmovss   xmm7, dword ptr [rbp-0x3C]
+       vmovss   xmm8, dword ptr [rbp-0x38]
+       vmovss   xmm9, dword ptr [rbp-0x34]
+       vmovss   xmm10, dword ptr [rbp-0x30]
+       vmovss   xmm11, dword ptr [rbp-0x2C]
+       vmovss   xmm12, dword ptr [rbp-0x28]
+       vmovss   xmm13, dword ptr [rbp-0x24]
+       vmovss   xmm14, dword ptr [rbp-0x20]
+       vmovss   xmm15, dword ptr [rbp-0x1C]
+       vmovss   xmm16, dword ptr [rbp-0x18]
+       vmovss   xmm17, dword ptr [rbp-0x14]
+       vmovss   xmm18, dword ptr [rbp-0x10]
+       vmovss   xmm19, dword ptr [rbp-0x0C]
        vucomiss xmm4, xmm4
        jp       G_M31528_IG09
+						;; size=138 bbWeight=1 PerfScore 43.00
+G_M31528_IG03:
        vucomiss xmm5, xmm5
        jp       G_M31528_IG09
        vucomiss xmm6, xmm6
        jp       G_M31528_IG09
        vucomiss xmm7, xmm7
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm8, xmm8
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm9, xmm9
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm10, xmm10
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm11, xmm11
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm12, xmm12
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm13, xmm13
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm14, xmm14
-       jp       SHORT G_M31528_IG09
+       jp       G_M31528_IG09
        vucomiss xmm15, xmm15
-       jp       SHORT G_M31528_IG09
-						;; size=122 bbWeight=0.50 PerfScore 22.50
+       jp       G_M31528_IG09
+       vucomiss xmm16, xmm16
+       jp       G_M31528_IG09
+       vucomiss xmm17, xmm17
+       jp       G_M31528_IG09
+       vucomiss xmm18, xmm18
+       jp       G_M31528_IG09
+       vucomiss xmm19, xmm19
+       jp       G_M31528_IG09
+						;; size=166 bbWeight=0.50 PerfScore 22.50
 G_M31528_IG04:
-       lea      rdi, [rbp-0x48]
-       mov      rax, 0xD1FFAB1E      ; code for System.Numerics.Matrix4x4+Impl:GetDeterminant():float:this
-       call     [rax]System.Numerics.Matrix4x4+Impl:GetDeterminant():float:this
+       vpermilps xmm4, xmm2, 22
+       vpermilps xmm5, xmm2, 64
+       vpermilps xmm6, xmm3, -17
+       vpermilps xmm7, xmm3, -74
+       vpermilps xmm8, xmm2, -17
+       vpermilps xmm2, xmm2, -74
+       vpermilps xmm9, xmm3, 22
+       vpermilps xmm3, xmm3, 64
+       vpermilps xmm10, xmm1, 1
+       vpermilps xmm11, xmm1, 90
+       vpermilps xmm1, xmm1, -65
+       vmulps   xmm5, xmm5, xmm7
+       vmulps   xmm2, xmm2, xmm3
+       vsubps   xmm2, xmm5, xmm2
+       vmulps   xmm3, xmm4, xmm6
+       vmulps   xmm4, xmm8, xmm9
+       vsubps   xmm3, xmm3, xmm4
+       vshufps  xmm4, xmm3, xmm2, 45
+       vmulps   xmm4, xmm11, xmm4
+       vpermilps xmm3, xmm3, -112
+       vmulps   xmm3, xmm10, xmm3
+       vsubps   xmm3, xmm3, xmm4
+       vpermilps xmm2, xmm2, 83
+       vmulps   xmm1, xmm2, xmm1
+       vaddps   xmm1, xmm1, xmm3
+       vmulps   xmm0, xmm1, xmm0
+       vpermilps xmm1, xmm0, 49
+       vsubps   xmm0, xmm0, xmm1
+       vmovaps  xmm1, xmm0
+       vunpckhps xmm0, xmm0, xmm0
+       vaddss   xmm0, xmm1, xmm0
        vucomiss xmm0, dword ptr [reloc @RWD00]
        ja       SHORT G_M31528_IG06
-						;; size=26 bbWeight=1 PerfScore 8.75
+						;; size=164 bbWeight=1 PerfScore 63.25
 G_M31528_IG05:
        xor      edi, edi
        jmp      SHORT G_M31528_IG07
 						;; size=4 bbWeight=0.50 PerfScore 1.12
 G_M31528_IG06:
        vmovss   xmm1, dword ptr [reloc @RWD04]
        vucomiss xmm1, xmm0
        seta     dil
        movzx    rdi, dil
 						;; size=20 bbWeight=0.50 PerfScore 2.38
 G_M31528_IG07:
        test     edi, edi
        jne      SHORT G_M31528_IG09
 						;; size=4 bbWeight=1 PerfScore 1.25
 G_M31528_IG08:
-       vzeroupper 
        add      rsp, 72
        pop      rbx
        pop      rbp
        ret      
-						;; size=10 bbWeight=1 PerfScore 3.25
+						;; size=7 bbWeight=1 PerfScore 2.25
 G_M31528_IG09:
        mov      rdi, 0xD1FFAB1E      ; SixLabors.ImageSharp.Processing.Processors.Transforms.DegenerateTransformException
        call     CORINFO_HELP_NEWSFAST
        mov      rbx, rax
        mov      rdi, rbx
        mov      rax, 0xD1FFAB1E      ; code for System.Exception:.ctor():this
        call     [rax]System.Exception:.ctor():this
        mov      rdi, 0xD1FFAB1E      ; 'Matrix is degenerate. Check input values.'
        mov      gword ptr [rbx+0x10], rdi
        mov      rdi, rbx
        call     CORINFO_HELP_THROW
        int3     
 						;; size=56 bbWeight=0 PerfScore 0.00
 RWD00  	dd	B58637BEh		;    -1e-06
 RWD04  	dd	358637BEh		;     1e-06
 
 
-; Total bytes of code 363, prolog size 11, PerfScore 81.00, instruction count 83, allocated bytes for code 363 (MethodHash=06c984d7) for method SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:CheckDegenerate(System.Numerics.Matrix4x4) (FullOpts)
+; Total bytes of code 570, prolog size 11, PerfScore 138.50, instruction count 116, allocated bytes for code 570 (MethodHash=06c984d7) for method SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:CheckDegenerate(System.Numerics.Matrix4x4) (FullOpts)
 ; ============================================================

177 (31.55 % of base) - SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this

 ; Assembly listing for method SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
 ; Emitting BLENDED_CODE for x64 + VEX + EVEX on Unix
 ; FullOpts code
 ; optimized code
 ; rbp based frame
 ; partially interruptible
 ; No PGO data
-; 0 inlinees with PGO data; 23 single block inlinees; 6 inlinees without PGO data
+; 0 inlinees with PGO data; 26 single block inlinees; 7 inlinees without PGO data
 ; Final local variable assignments
 ;
-;  V00 this         [V00,T02] (  4,  4   )     ref  ->  rbx         this class-hnd single-def <SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder>
-;  V01 arg1         [V01,T09] (  1,  1   )  struct (64) [rbp+0x10]  do-not-enreg[S] single-def <System.Numerics.Matrix4x4>
-;  V02 loc0         [V02,T04] (  3,  3   )     ref  ->  r15         class-hnd exact single-def <SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder+<>c__DisplayClass33_0>
+;  V00 this         [V00,T04] (  4,  4   )     ref  ->  rbx         this class-hnd single-def <SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder>
+;  V01 arg1         [V01,T11] (  1,  1   )  struct (64) [rbp+0x10]  do-not-enreg[S] single-def <System.Numerics.Matrix4x4>
+;  V02 loc0         [V02,T06] (  3,  3   )     ref  ->  r15         class-hnd exact single-def <SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder+<>c__DisplayClass33_0>
 ;# V03 OutArgs      [V03    ] (  1,  1   )  struct ( 0) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace" <Empty>
 ;* V04 tmp1         [V04    ] (  0,  0   )     ref  ->  zero-ref    class-hnd exact "NewObj constructor temp" <SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder+<>c__DisplayClass33_0>
-;  V05 tmp2         [V05,T01] (  5,  8   )     ref  ->  r14         class-hnd exact single-def "NewObj constructor temp" <System.Func`2[SixLabors.ImageSharp.Size,System.Numerics.Matrix4x4]>
+;  V05 tmp2         [V05,T03] (  5,  8   )     ref  ->  r14         class-hnd exact single-def "NewObj constructor temp" <System.Func`2[SixLabors.ImageSharp.Size,System.Numerics.Matrix4x4]>
 ;* V06 tmp3         [V06    ] (  0,  0   )  struct (64) zero-ref    do-not-enreg[S] "Inlining Arg" <System.Numerics.Matrix4x4>
-;  V07 tmp4         [V07,T27] (  4,  0   )     ref  ->  rbx         class-hnd exact single-def "NewObj constructor temp" <SixLabors.ImageSharp.Processing.Processors.Transforms.DegenerateTransformException>
-;  V08 tmp5         [V08,T08] (  2,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
-;  V09 tmp6         [V09    ] ( 18, 36   )  struct (64) [rbp-0x58]  do-not-enreg[XS] addr-exposed ld-addr-op "Inlining Arg" <System.Numerics.Matrix4x4>
-;* V10 tmp7         [V10,T28] (  0,  0   )   ubyte  ->  zero-ref    "Inline return value spill temp"
+;  V07 tmp4         [V07,T48] (  4,  0   )     ref  ->  rbx         class-hnd exact single-def "NewObj constructor temp" <SixLabors.ImageSharp.Processing.Processors.Transforms.DegenerateTransformException>
+;  V08 tmp5         [V08,T10] (  2,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
+;  V09 tmp6         [V09,T00] ( 20, 40   )  struct (64) [rbp-0x58]  do-not-enreg[SF] ld-addr-op "Inlining Arg" <System.Numerics.Matrix4x4>
+;* V10 tmp7         [V10,T49] (  0,  0   )   ubyte  ->  zero-ref    "Inline return value spill temp"
 ;* V11 tmp8         [V11    ] (  0,  0   )  struct (64) zero-ref    do-not-enreg[SF] "Inlining Arg" <System.Numerics.Matrix4x4>
 ;* V12 tmp9         [V12    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V13 tmp10        [V13    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V14 tmp11        [V14    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V15 tmp12        [V15    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V16 tmp13        [V16    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V17 tmp14        [V17    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V18 tmp15        [V18    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V19 tmp16        [V19    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V20 tmp17        [V20    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V21 tmp18        [V21    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V22 tmp19        [V22    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V23 tmp20        [V23    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V24 tmp21        [V24    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V25 tmp22        [V25    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V26 tmp23        [V26    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
 ;* V27 tmp24        [V27    ] (  0,  0   )   float  ->  zero-ref    "Inlining Arg"
-;  V28 tmp25        [V28,T07] (  3,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
-;  V29 tmp26        [V29,T10] (  3,  5   )   float  ->  mm0         "Inlining Arg"
-;  V30 tmp27        [V30,T00] (  7, 12   )     ref  ->  rdi         class-hnd single-def "Inlining Arg" <System.Collections.Generic.List`1[System.Func`2[SixLabors.ImageSharp.Size,System.Numerics.Matrix4x4]]>
-;  V31 tmp28        [V31,T06] (  3,  2.50)     ref  ->  rax         class-hnd single-def "Inline stloc first use temp" <System.Func`2[SixLabors.ImageSharp.Size,System.Numerics.Matrix4x4][]>
-;  V32 tmp29        [V32,T03] (  4,  3   )     int  ->  r15         single-def "Inline stloc first use temp"
-;  V33 tmp30        [V33,T11] (  3,  3   )   float  ->  mm0         single-def "V11.[000..004)"
-;  V34 tmp31        [V34,T12] (  3,  2   )   float  ->  mm1         single-def "V11.[004..008)"
-;  V35 tmp32        [V35,T13] (  3,  2   )   float  ->  mm2         single-def "V11.[008..012)"
-;  V36 tmp33        [V36,T14] (  3,  2   )   float  ->  mm3         single-def "V11.[012..016)"
-;  V37 tmp34        [V37,T15] (  3,  2   )   float  ->  mm4         single-def "V11.[016..020)"
-;  V38 tmp35        [V38,T16] (  3,  2   )   float  ->  mm5         single-def "V11.[020..024)"
-;  V39 tmp36        [V39,T17] (  3,  2   )   float  ->  mm6         single-def "V11.[024..028)"
-;  V40 tmp37        [V40,T18] (  3,  2   )   float  ->  mm7         single-def "V11.[028..032)"
-;  V41 tmp38        [V41,T19] (  3,  2   )   float  ->  mm8         single-def "V11.[032..036)"
-;  V42 tmp39        [V42,T20] (  3,  2   )   float  ->  mm9         single-def "V11.[036..040)"
-;  V43 tmp40        [V43,T21] (  3,  2   )   float  ->  mm10         single-def "V11.[040..044)"
-;  V44 tmp41        [V44,T22] (  3,  2   )   float  ->  mm11         single-def "V11.[044..048)"
-;  V45 tmp42        [V45,T23] (  3,  2   )   float  ->  mm12         single-def "V11.[048..052)"
-;  V46 tmp43        [V46,T24] (  3,  2   )   float  ->  mm13         single-def "V11.[052..056)"
-;  V47 tmp44        [V47,T25] (  3,  2   )   float  ->  mm14         single-def "V11.[056..060)"
-;  V48 tmp45        [V48,T26] (  3,  2   )   float  ->  mm15         single-def "V11.[060..064)"
-;  V49 cse0         [V49,T05] (  3,  3   )   byref  ->  rdi         "CSE #01: aggressive"
+;* V28 tmp25        [V28    ] (  0,  0   )  simd16  ->  zero-ref    single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V29 tmp26        [V29    ] (  0,  0   )  simd16  ->  zero-ref    "dup spill"
+;  V30 tmp27        [V30,T37] (  2,  2   )  simd16  ->  mm4         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V31 tmp28        [V31,T38] (  2,  2   )  simd16  ->  mm5         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V32 tmp29        [V32,T39] (  2,  2   )  simd16  ->  mm6         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V33 tmp30        [V33,T40] (  2,  2   )  simd16  ->  mm7         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V34 tmp31        [V34,T41] (  2,  2   )  simd16  ->  mm8         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V35 tmp32        [V35,T42] (  2,  2   )  simd16  ->  mm2         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V36 tmp33        [V36,T43] (  2,  2   )  simd16  ->  mm9         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V37 tmp34        [V37,T44] (  2,  2   )  simd16  ->  mm3         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V38 tmp35        [V38    ] (  0,  0   )  simd16  ->  zero-ref    "dup spill"
+;  V39 tmp36        [V39,T45] (  2,  2   )  simd16  ->  mm10         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V40 tmp37        [V40,T46] (  2,  2   )  simd16  ->  mm11         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V41 tmp38        [V41    ] (  0,  0   )  simd16  ->  zero-ref    "impSpillStackEnsure"
+;  V42 tmp39        [V42,T47] (  2,  2   )  simd16  ->  mm1         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V43 tmp40        [V43,T18] (  3,  3   )  simd16  ->  mm2         single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V44 tmp41        [V44,T14] (  3,  6   )  simd16  ->  mm3         "dup spill"
+;* V45 tmp42        [V45    ] (  0,  0   )  simd16  ->  zero-ref    "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V46 tmp43        [V46    ] (  0,  0   )  simd16  ->  zero-ref    "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V47 tmp44        [V47    ] (  0,  0   )  simd16  ->  zero-ref    "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V48 tmp45        [V48,T15] (  3,  6   )  simd16  ->  mm0         "dup spill"
+;  V49 tmp46        [V49,T19] (  3,  3   )  simd16  ->  mm0         ld-addr-op single-def "Inline stloc first use temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;* V50 tmp47        [V50    ] (  0,  0   )   float  ->  zero-ref    "impAppendStmt"
+;* V51 tmp48        [V51    ] (  0,  0   )  simd16  ->  zero-ref    "Inline return value spill temp" <System.Runtime.Intrinsics.Vector128`1[float]>
+;  V52 tmp49        [V52,T09] (  3,  2   )   ubyte  ->  rdi         "Inline return value spill temp"
+;  V53 tmp50        [V53,T17] (  3,  5   )   float  ->  mm0         "Inlining Arg"
+;  V54 tmp51        [V54,T01] (  7, 12   )     ref  ->  rdi         class-hnd single-def "Inlining Arg" <System.Collections.Generic.List`1[System.Func`2[SixLabors.ImageSharp.Size,System.Numerics.Matrix4x4]]>
+;  V55 tmp52        [V55,T08] (  3,  2.50)     ref  ->  rax         class-hnd single-def "Inline stloc first use temp" <System.Func`2[SixLabors.ImageSharp.Size,System.Numerics.Matrix4x4][]>
+;  V56 tmp53        [V56,T05] (  4,  3   )     int  ->  r15         single-def "Inline stloc first use temp"
+;  V57 tmp54        [V57,T20] (  3,  3   )  simd16  ->  mm0         single-def "V09.[000..016)"
+;  V58 tmp55        [V58,T16] (  5,  5   )  simd16  ->  mm1         single-def "V09.[016..032)"
+;  V59 tmp56        [V59,T12] (  6,  6   )  simd16  ->  mm2         single-def "V09.[032..048)"
+;  V60 tmp57        [V60,T13] (  6,  6   )  simd16  ->  mm3         single-def "V09.[048..064)"
+;  V61 tmp58        [V61,T21] (  3,  3   )   float  ->  mm4         single-def "V11.[000..004)"
+;  V62 tmp59        [V62,T22] (  3,  2   )   float  ->  mm5         single-def "V11.[004..008)"
+;  V63 tmp60        [V63,T23] (  3,  2   )   float  ->  mm6         single-def "V11.[008..012)"
+;  V64 tmp61        [V64,T24] (  3,  2   )   float  ->  mm7         single-def "V11.[012..016)"
+;  V65 tmp62        [V65,T25] (  3,  2   )   float  ->  mm8         single-def "V11.[016..020)"
+;  V66 tmp63        [V66,T26] (  3,  2   )   float  ->  mm9         single-def "V11.[020..024)"
+;  V67 tmp64        [V67,T27] (  3,  2   )   float  ->  mm10         single-def "V11.[024..028)"
+;  V68 tmp65        [V68,T28] (  3,  2   )   float  ->  mm11         single-def "V11.[028..032)"
+;  V69 tmp66        [V69,T29] (  3,  2   )   float  ->  mm12         single-def "V11.[032..036)"
+;  V70 tmp67        [V70,T30] (  3,  2   )   float  ->  mm13         single-def "V11.[036..040)"
+;  V71 tmp68        [V71,T31] (  3,  2   )   float  ->  mm14         single-def "V11.[040..044)"
+;  V72 tmp69        [V72,T32] (  3,  2   )   float  ->  mm15         single-def "V11.[044..048)"
+;  V73 tmp70        [V73,T33] (  3,  2   )   float  ->  mm16         single-def "V11.[048..052)"
+;  V74 tmp71        [V74,T34] (  3,  2   )   float  ->  mm17         single-def "V11.[052..056)"
+;  V75 tmp72        [V75,T35] (  3,  2   )   float  ->  mm18         single-def "V11.[056..060)"
+;  V76 tmp73        [V76,T36] (  3,  2   )   float  ->  mm19         single-def "V11.[060..064)"
+;  V77 tmp74        [V77,T02] (  5, 10   )   byref  ->  rdi         single-def "Spilling address for field-by-field copy"
+;  V78 cse0         [V78,T07] (  3,  3   )   byref  ->  rdi         "CSE #01: aggressive"
 ;
 ; Lcl frame size = 72
 
 G_M3948_IG01:
        push     rbp
        push     r15
        push     r14
        push     rbx
        sub      rsp, 72
        lea      rbp, [rsp+0x60]
        mov      rbx, rdi
 						;; size=18 bbWeight=1 PerfScore 5.00
 G_M3948_IG02:
        mov      rdi, 0xD1FFAB1E      ; SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder+<>c__DisplayClass33_0
        call     CORINFO_HELP_NEWSFAST
        mov      r15, rax
        lea      rdi, bword ptr [r15+0x08]
        vmovdqu32 zmm0, zmmword ptr [rbp+0x10]
        vmovdqu32 zmmword ptr [rdi], zmm0
-       vmovdqu32 zmm0, zmmword ptr [rdi]
-       vmovdqu32 zmmword ptr [rbp-0x58], zmm0
-       vmovss   xmm0, dword ptr [rbp-0x58]
-       vmovss   xmm1, dword ptr [rbp-0x54]
-       vmovss   xmm2, dword ptr [rbp-0x50]
-       vmovss   xmm3, dword ptr [rbp-0x4C]
-       vmovss   xmm4, dword ptr [rbp-0x48]
-       vmovss   xmm5, dword ptr [rbp-0x44]
-       vmovss   xmm6, dword ptr [rbp-0x40]
-       vmovss   xmm7, dword ptr [rbp-0x3C]
-       vmovss   xmm8, dword ptr [rbp-0x38]
-       vmovss   xmm9, dword ptr [rbp-0x34]
-       vmovss   xmm10, dword ptr [rbp-0x30]
-       vmovss   xmm11, dword ptr [rbp-0x2C]
-       vmovss   xmm12, dword ptr [rbp-0x28]
-       vmovss   xmm13, dword ptr [rbp-0x24]
-       vmovss   xmm14, dword ptr [rbp-0x20]
-       vmovss   xmm15, dword ptr [rbp-0x1C]
-       vucomiss xmm0, xmm0
-       jp       G_M3948_IG12
-						;; size=144 bbWeight=1 PerfScore 47.00
-G_M3948_IG03:
-       vucomiss xmm1, xmm1
-       jp       G_M3948_IG12
-       vucomiss xmm2, xmm2
-       jp       G_M3948_IG12
-       vucomiss xmm3, xmm3
-       jp       G_M3948_IG12
+       vmovups  xmm0, xmmword ptr [rdi]
+       vmovups  xmm1, xmmword ptr [rdi+0x10]
+       vmovups  xmm2, xmmword ptr [rdi+0x20]
+       vmovups  xmm3, xmmword ptr [rdi+0x30]
+       vmovups  xmmword ptr [rbp-0x58], xmm0
+       vmovups  xmmword ptr [rbp-0x48], xmm1
+       vmovups  xmmword ptr [rbp-0x38], xmm2
+       vmovups  xmmword ptr [rbp-0x28], xmm3
+       vmovss   xmm4, dword ptr [rbp-0x58]
+       vmovss   xmm5, dword ptr [rbp-0x54]
+       vmovss   xmm6, dword ptr [rbp-0x50]
+       vmovss   xmm7, dword ptr [rbp-0x4C]
+       vmovss   xmm8, dword ptr [rbp-0x48]
+       vmovss   xmm9, dword ptr [rbp-0x44]
+       vmovss   xmm10, dword ptr [rbp-0x40]
+       vmovss   xmm11, dword ptr [rbp-0x3C]
+       vmovss   xmm12, dword ptr [rbp-0x38]
+       vmovss   xmm13, dword ptr [rbp-0x34]
+       vmovss   xmm14, dword ptr [rbp-0x30]
+       vmovss   xmm15, dword ptr [rbp-0x2C]
+       vmovss   xmm16, dword ptr [rbp-0x28]
+       vmovss   xmm17, dword ptr [rbp-0x24]
+       vmovss   xmm18, dword ptr [rbp-0x20]
+       vmovss   xmm19, dword ptr [rbp-0x1C]
        vucomiss xmm4, xmm4
        jp       G_M3948_IG12
+						;; size=175 bbWeight=1 PerfScore 54.00
+G_M3948_IG03:
        vucomiss xmm5, xmm5
        jp       G_M3948_IG12
        vucomiss xmm6, xmm6
        jp       G_M3948_IG12
        vucomiss xmm7, xmm7
        jp       G_M3948_IG12
        vucomiss xmm8, xmm8
        jp       G_M3948_IG12
        vucomiss xmm9, xmm9
        jp       G_M3948_IG12
        vucomiss xmm10, xmm10
        jp       G_M3948_IG12
        vucomiss xmm11, xmm11
        jp       G_M3948_IG12
        vucomiss xmm12, xmm12
        jp       G_M3948_IG12
        vucomiss xmm13, xmm13
        jp       G_M3948_IG12
        vucomiss xmm14, xmm14
        jp       G_M3948_IG12
        vucomiss xmm15, xmm15
        jp       G_M3948_IG12
-						;; size=158 bbWeight=0.50 PerfScore 22.50
+       vucomiss xmm16, xmm16
+       jp       G_M3948_IG12
+       vucomiss xmm17, xmm17
+       jp       G_M3948_IG12
+       vucomiss xmm18, xmm18
+       jp       G_M3948_IG12
+       vucomiss xmm19, xmm19
+       jp       G_M3948_IG12
+						;; size=166 bbWeight=0.50 PerfScore 22.50
 G_M3948_IG04:
-       lea      rdi, [rbp-0x58]
-       mov      rax, 0xD1FFAB1E      ; code for System.Numerics.Matrix4x4+Impl:GetDeterminant():float:this
-       call     [rax]System.Numerics.Matrix4x4+Impl:GetDeterminant():float:this
+       vpermilps xmm4, xmm2, 22
+       vpermilps xmm5, xmm2, 64
+       vpermilps xmm6, xmm3, -17
+       vpermilps xmm7, xmm3, -74
+       vpermilps xmm8, xmm2, -17
+       vpermilps xmm2, xmm2, -74
+       vpermilps xmm9, xmm3, 22
+       vpermilps xmm3, xmm3, 64
+       vpermilps xmm10, xmm1, 1
+       vpermilps xmm11, xmm1, 90
+       vpermilps xmm1, xmm1, -65
+       vmulps   xmm5, xmm5, xmm7
+       vmulps   xmm2, xmm2, xmm3
+       vsubps   xmm2, xmm5, xmm2
+       vmulps   xmm3, xmm4, xmm6
+       vmulps   xmm4, xmm8, xmm9
+       vsubps   xmm3, xmm3, xmm4
+       vshufps  xmm4, xmm3, xmm2, 45
+       vmulps   xmm4, xmm11, xmm4
+       vpermilps xmm3, xmm3, -112
+       vmulps   xmm3, xmm10, xmm3
+       vsubps   xmm3, xmm3, xmm4
+       vpermilps xmm2, xmm2, 83
+       vmulps   xmm1, xmm2, xmm1
+       vaddps   xmm1, xmm1, xmm3
+       vmulps   xmm0, xmm1, xmm0
+       vpermilps xmm1, xmm0, 49
+       vsubps   xmm0, xmm0, xmm1
+       vmovaps  xmm1, xmm0
+       vunpckhps xmm0, xmm0, xmm0
+       vaddss   xmm0, xmm1, xmm0
        vucomiss xmm0, dword ptr [reloc @RWD00]
        ja       SHORT G_M3948_IG06
-						;; size=26 bbWeight=1 PerfScore 8.75
+						;; size=164 bbWeight=1 PerfScore 63.25
 G_M3948_IG05:
        xor      edi, edi
        jmp      SHORT G_M3948_IG07
 						;; size=4 bbWeight=0.50 PerfScore 1.12
 G_M3948_IG06:
        vmovss   xmm1, dword ptr [reloc @RWD04]
        vucomiss xmm1, xmm0
        seta     dil
        movzx    rdi, dil
 						;; size=20 bbWeight=0.50 PerfScore 2.38
 G_M3948_IG07:
        test     edi, edi
        jne      G_M3948_IG12
        mov      rdi, 0xD1FFAB1E      ; System.Func`2[SixLabors.ImageSharp.Size,System.Numerics.Matrix4x4]
        call     CORINFO_HELP_NEWSFAST
        mov      r14, rax
        lea      rdi, bword ptr [r14+0x10]
        mov      rsi, r15
        call     CORINFO_HELP_ASSIGN_REF
        mov      rdi, 0xD1FFAB1E      ; code for SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder+<>c__DisplayClass33_0:<AppendMatrix>b__0(SixLabors.ImageSharp.Size):System.Numerics.Matrix4x4:this
        mov      qword ptr [r14+0x18], rdi
        mov      rdi, gword ptr [rbx+0x08]
        inc      dword ptr [rdi+0x14]
        mov      rax, gword ptr [rdi+0x08]
        mov      r15d, dword ptr [rdi+0x10]
        cmp      dword ptr [rax+0x08], r15d
        ja       SHORT G_M3948_IG09
 						;; size=73 bbWeight=1 PerfScore 17.75
 G_M3948_IG08:
        mov      rsi, r14
        mov      rax, 0xD1FFAB1E      ; code for System.Collections.Generic.List`1[System.__Canon]:AddWithResize(System.__Canon):this
        call     [rax]System.Collections.Generic.List`1[System.__Canon]:AddWithResize(System.__Canon):this
        jmp      SHORT G_M3948_IG10
 						;; size=17 bbWeight=0.50 PerfScore 2.75
 G_M3948_IG09:
        lea      esi, [r15+0x01]
        mov      dword ptr [rdi+0x10], esi
        mov      esi, r15d
        mov      rdi, rax
        mov      rdx, r14
        mov      rax, 0xD1FFAB1E      ; code for CORINFO_HELP_ARRADDR_ST
        call     [rax]CORINFO_HELP_ARRADDR_ST
 						;; size=28 bbWeight=0.50 PerfScore 2.75
 G_M3948_IG10:
        mov      rax, rbx
 						;; size=3 bbWeight=1 PerfScore 0.25
 G_M3948_IG11:
        vzeroupper 
        add      rsp, 72
        pop      rbx
        pop      r14
        pop      r15
        pop      rbp
        ret      
 						;; size=14 bbWeight=1 PerfScore 4.25
 G_M3948_IG12:
        mov      rdi, 0xD1FFAB1E      ; SixLabors.ImageSharp.Processing.Processors.Transforms.DegenerateTransformException
        call     CORINFO_HELP_NEWSFAST
        mov      rbx, rax
        mov      rdi, rbx
        mov      rax, 0xD1FFAB1E      ; code for System.Exception:.ctor():this
        call     [rax]System.Exception:.ctor():this
        mov      rdi, 0xD1FFAB1E      ; 'Matrix is degenerate. Check input values.'
        mov      gword ptr [rbx+0x10], rdi
        mov      rdi, rbx
        call     CORINFO_HELP_THROW
        int3     
 						;; size=56 bbWeight=0 PerfScore 0.00
 RWD00  	dd	B58637BEh		;    -1e-06
 RWD04  	dd	358637BEh		;     1e-06
 
 
-; Total bytes of code 561, prolog size 15, PerfScore 114.50, instruction count 120, allocated bytes for code 561 (MethodHash=49fdf093) for method SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
+; Total bytes of code 738, prolog size 15, PerfScore 176.00, instruction count 154, allocated bytes for code 738 (MethodHash=49fdf093) for method SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:AppendMatrix(System.Numerics.Matrix4x4):SixLabors.ImageSharp.Processing.ProjectiveTransformBuilder:this (FullOpts)
 ; ============================================================

Note: some changes were skipped as they were too large to fit into a comment.

Larger list of diffs: https://gist.github.com/MihuBot/e5eb2ed1f9466e745774c8a71f1da2dc

[MihuBot]

@tannergooding