Re: [PATCH v7 0/2] MR1618: gdiplus: improve performance of GdipScaleMatrix

[Bartosz Kosiorek (＠gang65)]

On Fri Dec  2 20:46:15 2022 +0000, Alex Henrie wrote:
> > I could do whatever you want to get binary code in efficient form (as
>
> it is the output from this Merge Request). Could you please describe how
> constant propagation should looks like?
> If I understand correctly, Esme just wants you to test the performance
> of changing matrix_multiply without changing GdipScaleMatrix.

With only `gdiplus: improve performance of matrix multiplication by unrolling loop.` commit applied, I had:

Wine gdiplus.dll **with only `matrix_multiply` optimizations** and `matrix_multiply` inlining:
   * 500 x `GdipScaleMatrix` time (seconds): 0.21s 
   * 700 x `GdipMultiplyMatrix` time (seconds): 0.14s

Generally I cannot get value below 0.21s.