tree 140211e8c660b789482901537142a2d773baed87
parent 5e7bc5e080fb393c3ddb954fa9fd3714284f72c2
author Pierre van Houtryve <pierre.vanhoutryve@amd.com> 1747384263 +0200
committer GitHub <noreply@github.com> 1747384263 +0200
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsFcBAABCAAQBQJoJvfHCRC1aQ7uu5UhlAAALSoQAKkrkuG1iylE76hkG9ra7vhe
 WKGsR5VzW2fjbtxjJgEc95oJdyOcCBhv0/DBcBUgORPvRJ19GZsTGShi65PJQs4/
 RYxkwJngV7Kc/KHOaCiYzwBzUdsnj43dW85fIEpJPVLoBA3VS9aP882cV7QNITM/
 MNSSmRweLoJF1GaLnbhdFqFIV7BZfKYZ78cfv4PYkZkikg0a+mIE6gPg++V8R951
 HsotWJZeLqmH/FUmOLICXV09C2BTORD4Bm1PUUDbIfhg+eX/Zf9UDrUYa6l2F8SB
 DNog4fnU7lcUnnitorqdI8pxDVdTi5XfyK0FiZhwmEAMo77D0/FxeMKU2U4Q2bK3
 kYKgOTjt20/Ovk/oMzYjYhfWGRszVZqU9NsRZgABpaXxkgQyABq4vhoaFhC1hTC1
 aHTv0VGxvB4dyOY6uGnYL566w1vO95V/M4uHFTb6pXkL2YbP5lCcpUHI311t3oUH
 D+FudPm0+6y3Wm8rOtVR6ywWwJRrVXUbHx37Uol1UZT8KgtM4Zx4gPJdbODkQ/Vt
 4lTpcmOAXo1S0hoxfkbdgR6SeMQp+gYSyFlCUNY0IWSu0fZrVBsyyC548VYWMBdy
 T5Jibhlj/9/7K9C4JZEjtmez37PpbP5vQnst7vbFck13/2Dxs0k/QEFQv0mXOAp4
 Cfp3tu+lBr6f68iK2AkK
 =rMlr
 -----END PGP SIGNATURE-----
 

[AMDGPU] Do not promote uniform i16 operations to i32 in CGP (#140208)

For the majority of cases, this is a neutral or positive change.
There are even testcases that greatly benefit from it, but some regressions are possible.
There is #140040 for GlobalISel that'd need to be fixed but it's only a one instruction regression and I think it can be fixed later.

Solves #64591