tree 00954d39471a0b00a657247101365282fc4ffa97
parent 03c744ead1482abde3ee9e70293215c5b557c629
author MaheshRavishankar <1663364+MaheshRavishankar@users.noreply.github.com> 1730124057 -0700
committer GitHub <noreply@github.com> 1730124057 -0700
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsFcBAABCAAQBQJnH5kZCRC1aQ7uu5UhlAAAEM4QAJ1Ia1CBy7FVsrHmbR/ytbXK
 SxT3JmSyCPgCiiL56R/lmnG+l57QIfcZdfbjxLih8r1fy1MTsBEMMU7V1MRU6IYU
 xbm/2PRz+aNdjsVFKmglHTJR3x61O8Cm3jQ+1/REGO6QqBimOnZO1ZBqVmzdwPkA
 A8uYbgz9hxBEjAntoK9MjDq0eCj6DZREP131L1WT3NOb/1gBsZiEU609SbKZMWf5
 FXVvWTu+pachm67sU7hb7OFOkItCwcL1liTBLi6xUpd3Ex3u9hbPGZTEoj2mhxJQ
 +SwPoAmhtaVG6wy5Qn3YQw3OeE1hGM1ANynJ31oOlbAwXDxrTCXyIlW9eLoxSBc3
 YZVPLs422z2GzC4vk73T3L5sUTP6XBGqwWTkuEomZ8Poyd8NR4zNE2aSSVqpQegv
 /tbYo1YyNX+H7PryYVL0V5WrwAZJtmcdFZPBGCBzAgSxp0C0pgKbvw7q0okSG8hH
 NyDEcBm4QfVRD/7reRtqTD56pHoi91xFWEQ76vWYNVv1+PI9FPkm3VBXRId+P8hX
 KOq0WMDg8/TUqfGnpoN3WEK2l0zIDFZiLhPH4T3w+EWtXiExAPKLEje0hI8UwsF/
 0z7ZapcHZSi7OiPeuGXxy6hcD/nG5u7hxm9gyNopJSo7jnE1/D5tiUN/jpd9A335
 bLTuFiuA4XcuvXwfIxqZ
 =dCHX
 -----END PGP SIGNATURE-----
 

Pass to block dynamic dimensions of operands of `iree_linalg_ext.attention`. (#18874)

The use of `IntegerRangeAnalysis` and `IntegerDivisibilityAnalysis`
gives range and divisibility information for constants passed to the
dispatch. This can be used to infer the range and divisibility
information for all tensor values in the dispatch. This PR adds an
analysis to do this.

This analysis is then used to expand the dimensions of operands of the
attention operation that are dynamic, but are known to be divisible by
a compile-time static value. This gets the operations into a form that
can be compiled by the AMDGPU backend and target the mfma intrinsics.

Signed-off-by: MaheshRavishankar <mahesh.ravishankar@gmail.com>

---------

Signed-off-by: MaheshRavishankar <mravisha@amd.com>