<HTML><BODY style="word-wrap: break-word; -khtml-nbsp-mode: space; -khtml-line-break: after-white-space; "><BR><DIV><DIV>On Apr 23, 2007, at 5:08 PM, Chris Lattner wrote:</DIV><BR class="Apple-interchange-newline"><BLOCKQUOTE type="cite"><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">On Sun, 22 Apr 2007, Christopher Lamb wrote:</DIV> <BLOCKQUOTE type="cite"><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">1. Is there a good reason that v2f32 types are excluded from the</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">isFloatingPoint filter? Looks like a bug to me.</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; min-height: 14px; "><BR></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><SPAN class="Apple-converted-space"> </SPAN>v2f32<SPAN class="Apple-converted-space"> </SPAN>=<SPAN class="Apple-converted-space"> </SPAN>22, <SPAN class="Apple-converted-space"> </SPAN>//<SPAN class="Apple-converted-space"> </SPAN>2 x f32</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><SPAN class="Apple-converted-space"> </SPAN>v4f32<SPAN class="Apple-converted-space"> </SPAN>=<SPAN class="Apple-converted-space"> </SPAN>23, <SPAN class="Apple-converted-space"> </SPAN>//<SPAN class="Apple-converted-space"> </SPAN>4 x f32 <== start ??</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><SPAN class="Apple-converted-space"> </SPAN>v2f64<SPAN class="Apple-converted-space"> </SPAN>=<SPAN class="Apple-converted-space"> </SPAN>24, <SPAN class="Apple-converted-space"> </SPAN>//<SPAN class="Apple-converted-space"> </SPAN>2 x f64 <== end</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; min-height: 14px; "><BR></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><SPAN class="Apple-converted-space"> </SPAN>static inline bool isFloatingPoint(ValueType VT) {</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><SPAN class="Apple-converted-space"> </SPAN>return (VT >= f32 && VT <= f128) || (VT >= v4f32 && VT <= v2f64);</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><SPAN class="Apple-converted-space"> </SPAN>}</DIV> </BLOCKQUOTE><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; min-height: 14px; "><BR></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">Definitely a bug.<SPAN class="Apple-converted-space"> </SPAN>I don't think any existing targets support v2f32 which<SPAN class="Apple-converted-space"> </SPAN></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">is probably why it wasn't noticed.</DIV></BLOCKQUOTE><DIV><BR class="khtml-block-placeholder"></DIV><DIV>Fix is in.</DIV><BR><BLOCKQUOTE type="cite"><BLOCKQUOTE type="cite"><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">2. My problem seems to stem from what appears to be under-constrained</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">typing of patterns. With vectors there is a challenge because</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">currently there is no type constraint available to indicate that an</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">operand can/cannot be a vector type, or a constraint that operands</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">must have the same number of elements (either both scalers or vectors</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">with the same number of elements, but not necessarily the same</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">element type or element size).</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; min-height: 14px; "><BR></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">Even with these constraints, it would be difficult to apply them to</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">the standard nodes in a general way, though I could be wrong. Should</DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">standard nodes have vector aware type constraints?</DIV> </BLOCKQUOTE><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; min-height: 14px; "><BR></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">It depends on which nodes you mean.<SPAN class="Apple-converted-space"> </SPAN>"and" for example, can apply to any<SPAN class="Apple-converted-space"> </SPAN></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">integer type, including vectors.<SPAN class="Apple-converted-space"> </SPAN>The various compares can only apply to<SPAN class="Apple-converted-space"> </SPAN></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">scalars, as it would be otherwise ambiguous how the comparison is done<SPAN class="Apple-converted-space"> </SPAN></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">(i.e. is it true if all element satisfy<SPAN class="Apple-converted-space"> </SPAN>the predicate, or if any of<SPAN class="Apple-converted-space"> </SPAN></DIV><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">them?)</DIV></BLOCKQUOTE><BR><BLOCKQUOTE type="cite"><DIV style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; ">Do you have a specific example in mind?</DIV></BLOCKQUOTE></DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV>A strategy that we have had luck with in the past is the concept of a "vector boolean" type, which is the result of a comparison between two vectors. It's then necessary to perform a reduction (and, or) on the elements in order to produce a scalar boolean value that's suitable for a branch. This also allows stuff like letting 'select' accept a vector boolean predicate and vector values to perform a vector select.</DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV>In these cases you'd want to enforce the kinds of operand type constraints as mentioned above to ensure that the types are all acceptable.</DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV> <SPAN class="Apple-style-span" style="border-collapse: separate; border-spacing: 0px 0px; color: rgb(0, 0, 0); font-family: Helvetica; font-size: 12px; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; text-align: auto; -khtml-text-decorations-in-effect: none; text-indent: 0px; -apple-text-size-adjust: auto; text-transform: none; orphans: 2; white-space: normal; widows: 2; word-spacing: 0px; "><SPAN class="Apple-style-span" style="border-collapse: separate; border-spacing: 0px 0px; color: rgb(0, 0, 0); font-family: Helvetica; font-size: 12px; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; text-align: auto; -khtml-text-decorations-in-effect: none; text-indent: 0px; -apple-text-size-adjust: auto; text-transform: none; orphans: 2; white-space: normal; widows: 2; word-spacing: 0px; "><DIV>--</DIV><DIV>Christopher Lamb</DIV><DIV><BR class="khtml-block-placeholder"></DIV><BR class="Apple-interchange-newline"></SPAN></SPAN> </DIV><BR></BODY></HTML>