<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Thu, Mar 17, 2016 at 2:41 PM, Rail Shafigulin <span dir="ltr"><<a href="mailto:rail@esenciatech.com" target="_blank">rail@esenciatech.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr"><div class="gmail_extra"><div class="gmail_quote"><span class="">On Thu, Mar 17, 2016 at 10:10 AM, Rail Shafigulin <span dir="ltr"><<a href="mailto:rail@esenciatech.com" target="_blank">rail@esenciatech.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr"><div class="gmail_extra"><div><div><div class="gmail_quote">On Wed, Mar 16, 2016 at 6:38 PM, Mehdi Amini <span dir="ltr"><<a href="mailto:mehdi.amini@apple.com" target="_blank">mehdi.amini@apple.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div style="word-wrap:break-word"><span><br><div><blockquote type="cite"><div>On Mar 16, 2016, at 5:38 PM, Rail Shafigulin <<a href="mailto:rail@esenciatech.com" target="_blank">rail@esenciatech.com</a>> wrote:</div><br><div><div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Wed, Mar 16, 2016 at 11:48 AM, Mehdi Amini <span dir="ltr"><<a href="mailto:mehdi.amini@apple.com" target="_blank">mehdi.amini@apple.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div style="word-wrap:break-word">Hi Rail,<div><br></div><div>Two hints to begin with:<div><br></div><div>1) Makes sure you example is vectorized on X86 for example<div>2) Is your target correctly overriding the TTI (declaring the vector register size for example) so that the vectorizer can kicks-in (see X86TTIImpl::getRegisterBitWidth for instance). Alternatively you can test the SLP vectorizer by passing to clang: -mllvm -slp-max-reg-size -mllvm 512 (I don't see an equivalent option for the loop vectorizer though).</div><div><br></div></div></div></div></blockquote><div>Well, it sort of worked. I added a getRegisterBitWidth(...) but then I got this error:</div><div><br></div><div><div>fatal error: error in backend: Cannot select: 0x5e949a8: v4i32 = BUILD_VECTOR 0x5e91ae8, 0x5e91ae8, 0x5e91ae8, 0x5e91ae8 [ORD=16] [ID=16]</div><div> 0x5e91ae8: i32 = Constant<0> [ID=5]</div><div> 0x5e91ae8: i32 = Constant<0> [ID=5]</div><div> 0x5e91ae8: i32 = Constant<0> [ID=5]</div><div> 0x5e91ae8: i32 = Constant<0> [ID=5]</div></div><div><br></div><div>What am I missing?</div></div></div></div></div></blockquote><div><br></div></div></span><div>I means that you have a vectorized IR that reached your backend, but your backend is not ready to deal with all the vector constructs in SelectionDAG. </div><div>You need to express how to legalize/select the BUILD_VECTOR in SelectionDAG to instructions that your target supports. You can look at what other targets are doing.</div><span><font color="#888888"><div><br></div><div>-- </div><div>Mehdi</div><div><br></div></font></span></div></blockquote></div></div></div></div></div></blockquote></span></div></div></div></blockquote><div><br></div><div>So I've added setOperationAction(ISD::BUILD_VECTOR, MVT::v4i32, Expand); to my code but that generated a following error:</div><div><br></div><div>fatal error: error in backend: Cannot select: 0x6a84dc8: i32 = extract_vector_elt 0x6a85388, 0x6a813b0 [ORD=9] [ID=16]</div><div> 0x6a85388: v4i32 = add 0x6a81098, 0x6a81e00 [ORD=8] [ID=15]</div><div> 0x6a81098: v4i32 = add 0x6a81bf0, 0x6a84168 [ORD=6] [ID=12]</div><div> 0x6a81bf0: v4i32,ch = CopyFromReg 0x6a2b7f0, 0x6a819e0 [ORD=5] [ID=8]</div><div> 0x6a819e0: v4i32 = Register %vreg4 [ID=1]</div><div> 0x6a84168: v4i32 = vector_shuffle 0x6a81bf0, 0x6a857a8<2,3,u,u> [ORD=5] [ID=10]</div><div> 0x6a81bf0: v4i32,ch = CopyFromReg 0x6a2b7f0, 0x6a819e0 [ORD=5] [ID=8]</div><div> 0x6a819e0: v4i32 = Register %vreg4 [ID=1]</div><div> 0x6a857a8: v4i32 = undef [ID=2]</div><div> 0x6a81e00: v4i32 = vector_shuffle 0x6a81098, 0x6a857a8<1,u,u,u> [ORD=7] [ID=14]</div><div> 0x6a81098: v4i32 = add 0x6a81bf0, 0x6a84168 [ORD=6] [ID=12]</div><div> 0x6a81bf0: v4i32,ch = CopyFromReg 0x6a2b7f0, 0x6a819e0 [ORD=5] [ID=8]</div><div> 0x6a819e0: v4i32 = Register %vreg4 [ID=1]</div><div> 0x6a84168: v4i32 = vector_shuffle 0x6a81bf0, 0x6a857a8<2,3,u,u> [ORD=5] [ID=10]</div><div> 0x6a81bf0: v4i32,ch = CopyFromReg 0x6a2b7f0, 0x6a819e0 [ORD=5] [ID=8]</div><div> 0x6a819e0: v4i32 = Register %vreg4 [ID=1]</div><div> 0x6a857a8: v4i32 = undef [ID=2]</div><div> 0x6a857a8: v4i32 = undef [ID=2]</div><div> 0x6a813b0: i32 = Constant<0> [ID=3]</div><div>In function: main</div><div><br></div><div>Then I've added</div><div>setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::i32, Expand);<br></div><div><br></div><div>but I still got the same error. So removed setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::i32, Expand); and added</div><div><br></div><div>setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::v4i32, Expand);<br></div><div><br></div><div>which produced a following error:</div><div><br></div><div><div>fatal error: error in backend: Cannot select: 0x7389250: v4i32 = vector_shuffle 0x73884e8, 0x738cbf8<1,u,u,u> [ORD=7] [ID=15]</div><div> 0x73884e8: v4i32 = add 0x7389040, 0x738b5b8 [ORD=6] [ID=13]</div><div> 0x7389040: v4i32,ch = CopyFromReg 0x73327f0, 0x7388e30 [ORD=5] [ID=9]</div><div> 0x7388e30: v4i32 = Register %vreg4 [ID=1]</div><div> 0x738b5b8: v4i32 = vector_shuffle 0x7389040, 0x738cbf8<2,3,u,u> [ORD=5] [ID=11]</div><div> 0x7389040: v4i32,ch = CopyFromReg 0x73327f0, 0x7388e30 [ORD=5] [ID=9]</div><div> 0x7388e30: v4i32 = Register %vreg4 [ID=1]</div><div> 0x738cbf8: v4i32 = undef [ID=2]</div><div> 0x738cbf8: v4i32 = undef [ID=2]</div><div>In function: main</div></div><div><br></div><div>Then I'v added setOperationAction(ISD::VECTOR_SHUFFLE, MVT::v4i32, Expand);</div><div><br></div><div>and then my clang just hang. There is no error, no warning clang just sits there and nothing happens. </div><div><br></div><div>I'm doing a lot of guess work in trying to understand what is going on. I would really appreciate any help on this.</div></div><br clear="all"><div><br></div>-- <br><div class="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div>Rail Shafigulin<br></div>Software Engineer <br>Esencia Technologies<br></div></div></div></div>
</div></div>