tree c746a5a40b15c5b01fb5bfeb55745f770ea563fc
parent 2dddf02a12a2219fd80d457b5bf32dd155eb3b66
author Lei Zhang <antiagainst@google.com> 1686593760 -0400
committer GitHub <noreply@github.com> 1686593760 -0700
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsBcBAABCAAQBQJkh2DgCRBK7hj4Ov3rIwAAlFYIACioSAGfQQN3NMdElVvmIba6
 sNdhGxDD+vwY0msIolQLnbKE6cpKd2kLciLeq4pmrY3gH6qU9kJL3cF2igZCYQZz
 u244jLmhWgTc+PLqv+SVD/FnTxhjwmmbVtzV+F6xURocJwS2yeU8pXfe1FZdieRF
 Dr7SuMq8jZDYTN8OdUDEd2/u2kJz39UCgsK3FQ2oZh5sul5S7Jio7WPXjv3UVZMT
 ictyGKA6g0MSETNc+ryfaF9CiqWQBOYNKL3CDj4p4EdaBgcpE4UXqp/jIylLnGhW
 82uvjLJ0uwVlvnl2KN4rrDP6OhOiNODQ/YHlgVNqxsHKfe4cZICcOAg1ZhBaKEA=
 =FuHl
 -----END PGP SIGNATURE-----
 

[cuda] Port over descriptor set and pipeline layout (#14038)

This commit reuses the existing CUDA HAL driver's descriptor set
and pipeline layout implementation. Along the way, dropped the
context wrapper and adds more documentation.

Progress towards https://github.com/openxla/iree/issues/13245