<div dir="auto"><div><br><div class="gmail_extra"><br><div class="gmail_quote">On Dec 11, 2017 6:54 PM, "Sean Silva" <<a href="mailto:chisophugis@gmail.com">chisophugis@gmail.com</a>> wrote:<br type="attribution"><blockquote class="quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>It looks like in various places in SelectionDAG and related machinery we appear to implicitly or explicitly assume that TypeExpandInteger only applies to scalars.<br></div><div><br>Does anybody know why that is?</div><div><br></div><div>For our target (Pixel Visual Core: <a href="https://www.blog.google/products/pixel/pixel-visual-core-image-processing-and-machine-learning-pixel-2/" target="_blank">https://www.blog.google/<wbr>products/pixel/pixel-visual-<wbr>core-image-processing-and-<wbr>machine-learning-pixel-2/</a>) expanding a N-bit vector op into two (N/2)-bit vector ops</div></div></blockquote></div></div></div><div dir="auto"><br></div><div dir="auto">To be clear: the vector ops are all of the same number of elements. Just the bit width of the element type changes.</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_extra"><div class="gmail_quote"><blockquote class="quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div> is a natural way to legalize some of our vector operations, but we're finding that we need to modify a bunch of random places in the code to make it work. I'm curious if there is a cleaner way, or some big picture thing I'm missing.</div><div><br></div><div>Thanks,</div><div><br></div><div>-- Sean Silva</div></div>
</blockquote></div><br></div></div></div>