"-function polynomial" problem

[snibgo]

Prerequisites

• [y] I have written a descriptive issue title
• [y] I have verified that I am using the latest version of ImageMagick
• [y] I have searched open and closed issues to ensure it has not already been reported

Description

"-function polynomial" gives slightly wrong answers on Windows 8.1, 7.0.10-27.

See also my previous report, https://www.imagemagick.org/discourse-server/viewtopic.php?f=3&t=37520

Steps to Reproduce

f:\prose\PICTURES>%IMG7%magick xc:gray(50%) +write txt: -function polynomial 1,0 txt:

# ImageMagick pixel enumeration: 1,1,65535,gray
0,0: (32767.5)  #800080008000  gray(50%)
# ImageMagick pixel enumeration: 1,1,65535,gray
0,0: (32211.3)  #7DD37DD37DD3  gray(49.1512%) <=== wrong answer

System Configuration

Windows 8.1.

IM version information:

f:\prose\PICTURES>%IMG7%magick -version
Version: ImageMagick 7.0.10-27 Q16 x64 2020-08-10 http://www.imagemagick.org
Copyright: Copyright (C) 1999-2018 ImageMagick Studio LLC
License: http://www.imagemagick.org/script/license.php
Visual C++: 192628806
Features: Cipher DPC HDRI Modules OpenCL OpenMP(2.0)
Delegates (built-in): bzlib cairo flif freetype gslib heic jng jp2 jpeg lcms lqr
 lzma openexr pangocairo png ps raw rsvg tiff webp xml zlib

[fmw42]

It seems to work fine for me on Mac OSX with IM 7.0.10.27 Q16.

magick xc:"gray(50%)" +write txt: -function polynomial 1,0 txt:
# ImageMagick pixel enumeration: 1,1,65535,gray
0,0: (32767.5)  #800080008000  gray(50%)
# ImageMagick pixel enumeration: 1,1,65535,gray
0,0: (32767.5)  #800080008000  gray(50%)

[snibgo]

Thanks, Fred. Yes, this seems to be a Windows-only problem.

[dlemstra]

This appears to be a rounding issue in the OpenCL code. The C codes uses double but the OpenCL code uses float instead.