we're seeing several dozens of ICEs in apply_scale since this patch (PR121523). I didn't pay too much attention due to vacation etc. but now coming back to this. Any specific spot I should start looking?
I had a quick look and part? of the issue is that vect_gen_prolog_loop_niters returns -1 = unknown for BOUND when align_in_elems is not constant.
Then we use bound -1 = -2 as scale factor for the loop profile.
It seems like we should set bound = 0 in that case so the loop profile scaling doesn't do anything.
-- Regards Robin
