tree 79880a0f8692da2e6f428df7572c2031982d882f
parent b7aa3b76033695bedd94ca6f102503661997e83f
author Quinn Dawkins <quinn.dawkins@gmail.com> 1715134673 -0700
committer GitHub <noreply@github.com> 1715134673 +0000
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsFcBAABCAAQBQJmOuDRCRC1aQ7uu5UhlAAAUfsQADY1jDZKs9NLixmpVfl460Xo
 WZjZDhkhF5Ygs7uvj8qBdb6ePx7OOyOahl908Q8FuuVSidB6WY247mWMZRbX0KcM
 5PJOIpkMFDmDF+gFGp/tzMnBjjdVDJNvGfolZCJUqLedGXGyl3uZi5VAHcf8gBC2
 4MHfohzwefDOxExeYEF0YISb3q+SkNQuTFeHZTMfoUpVXI7XY4ENFb8IzdNyv3mq
 PnODUdtdywl/XBaDiT4dcBQGXqLzTv6AwA1eqbwVDc8eS+suGMM2li1154clVIrH
 4MMMg6fRHx5gwng0AYaqM6rJAOcqOunI/ooS2JAMjSo2Z+zj6sX0dFRzUUWlwvHb
 CWdFNbE1IrBbylChyW9eXNvImQ5U8+9zPHpXk3I5x25Jcuunj4boEBiYXTJn0GwA
 P1uQEGvq9DWd6dyh94P8FvWfkCbA0tAXw+AQy/bYFQHEYJVtbB3uCtatilcq18JV
 vC6NLEJkbsA9LzGH4DKS8XK90qNZwQIlYhrVz2mBcCJT4WZm0rm9nQ9Nss+xXvaj
 8HbS3gnB0eYMTl9eAFViOptxDIPjMcUPgX1cm1dwCdtTCzCagW/waxlXQGLSE/PS
 5bamh2CLsK8hKNr9ip3Zyh03pKrSXkCXzzfm1HeGsG6JRqB2xFoIUcJpqKXc0KzM
 MzagZx1wfCxYASxVAZ/u
 =epDt
 -----END PGP SIGNATURE-----
 

[Codegen] Add pattern for lowering iree_gpu.shuffle_tensor (#17269)

This is a direct lowering that relies on bufferization to produce the
correct allocation and inserts a barrier between the slices. This is
an experimental lowering that plays fast and loose with barrier placement
(as does all barrier usage within the compiler).