Hi James, TImode is an integer mode so we strongly prefer using integer registers to avoid inefficient allocations using SIMD registers. We might be able to use TFmode since that prefers Q registers. However we don't support TFmode LDP/STP unless emitted explicitly like in prolog/epilog. LDP of TImode isn't supported at all, so both these options are out.
We also can't use 256-bit modes like OImode since that's a special type for LD2/ST2, so it's not possible to emit LDP/STP that way. So yes, using V4SImode and allowing LDP/STP being generated as usual is far better than what we do today. It also allows targets to disable 256-bit LDPs if required. Cheers, Wilco