On 1/7/2026 12:11 PM, Richard Biener wrote:
3. https://gcc.gnu.org/pipermail/gcc-patches/2025-November/700657.html
[PATCH 2/2] ia64: Expand MAX_VECT_LEN to 16
It's not clear to me this is the right thing to do. It seems to me that
value should be driven by performance behavior, not as a workaround for
limitations of the middle-end diagnostics.
It's purely internal as size to local arrays in ia64 permute handling
(copy&pasted from other targets I guess). So it has to match the largest
mode and likely the diagnostic correctly complains it is short. So the
patch looks obvious to me (for V16QI).
THanks. Clearly I misunderstood how this stuff worked. Based on your
comment above, I've pushed this to the trunk.
jeff
