tree 77c58727b2bee511b483dd1d49645146a0a75015
parent eede9f26b2e5a593eebe4362cd0de389dc01d8af
author Kunwar Grover <groverkss@gmail.com> 1718645940 +0100
committer GitHub <noreply@github.com> 1718645940 +0000
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsFcBAABCAAQBQJmcHS0CRC1aQ7uu5UhlAAA/8QQAGm+Lo3JH2dUXzSCcSO1OvI7
 MwLmLdvHoId6Y+JCwe0O7/zlUR+/riu9ADxM+9smzRIV/A07/NTlNI6rPYblclEV
 oGjqjM0B3VH/OfOr//J02HxbUVaGwWnY9kpeCrYOnlQvBfEflMoQuBjYM998+K3/
 2nuOJzaLASxIc4IZkin7Cy2Jk1ZjJ91+iHi0qAlhRszd0WhAUQx1ej5SiPK89IS5
 P2Sy/R250t1Wxnep6M006/xxzpGe/56CWKEL2BXDW+mHVrQ6YS/NqtXuuGOgoMmJ
 HtB4G4oO7p0AuOGYMNqOxElodN4I2weZ2dD+l+EokBmn++y8ctwxOYM2UqMynHNi
 lPB8FuE1XfahHdcTYfcwmisF1wPo3ta8mXQOSAFYsF7T35DTZuGGr+QGg3Zl6Jw2
 FIj/R/jmtXZQoCSMG//0ztIx9O0zPCzOd0XVTRd6SbylIsRUr+juyE0g85kCXxrY
 XtlZ+In1cEjnmLbPccZlTOSJodlIxE2h9M0KJ6sVnEQw9qwCNXFj7k/eD/RyKKoD
 tyk5A+vwiH0Aw5CGj2hPZz0+tpT/Hs+QzwzfPjFqkdVxKmATDnF0jsJ0iJMRoWpw
 KzBAK/QhYaqyR5QtFgMF5TZIOOka68F7icJwbe1t1b/5Blw+kDTWY4psqn7fIdK/
 yb9EWLeXgx6Pkk4MalNq
 =lxMH
 -----END PGP SIGNATURE-----
 

[LinalgExt] Reland: Add online_attention op (#17681)

This patch adds a new online_attention op. This op represents a
partially reduced attention op which can be tiled along it's k2
reduction dimension. This op also has indexing maps, supports tiling on
all dimensions other than k1 dimension, and can decompose based on any
given indexing maps.

This patch also makes the CPU backend use online attention to decompose
and tile reduction dimension, allowing it to be tiled along N and batch
dimensions, and tiling using LLVMCPUTile.

This is a reland of https://github.com/iree-org/iree/pull/17658 , with
more conservative tile size selection to not unroll too much.