<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <div class="moz-cite-prefix">On 5/11/20 10:53 PM, Chris Lattner
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:F597CC32-CFBF-4530-A1F5-54E573E87A95@nondot.org">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <br class="">
      <div><br class="">
        <blockquote type="cite" class="">
          <div class="">On May 8, 2020, at 4:40 PM, Josh Stone via
            llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org"
              class="" moz-do-not-send="true">llvm-dev@lists.llvm.org</a>>
            wrote:</div>
          <br class="Apple-interchange-newline">
          <div class="">
            <meta http-equiv="Content-Type" content="text/html;
              charset=UTF-8" class="">
            <div class="">
              <div class="moz-cite-prefix">On 4/14/20 1:26 AM, Neil
                Henning via llvm-dev wrote:<br class="">
              </div>
              <blockquote type="cite"
cite="mid:CALRoqd1AGvyKpNAnwJUMhdW4pEGTM-pFh0KmmYw1rQc7sKL7zA@mail.gmail.com"
                class="">
                <meta http-equiv="content-type" content="text/html;
                  charset=UTF-8" class="">
                <div dir="ltr" class="">
                  <div class="">Hey list,</div>
                  <div class=""><br class="">
                  </div>
                  <div class="">TL;DR - LLVM 10 is around 7-8% slower
                    than LLVM 9 when compiling the same inputs.</div>
                  <div class=""><br class="">
                  </div>
                  [...] concerns from the Rust folks about LLVM 10's
                  slowdowns [...]<br class="">
                </div>
              </blockquote>
              <p class="">I'm one of those Rust folks, and in the last
                few days I've been trying to dig into this. I focused on
                one particular benchmark input -- I've attached that IR
                generated from Rust 1.43.0 for LLVM 9.0.1. I started
                running a git bisection to see if anything huge jumped
                out, but it seems to be gradual. Maybe others can take
                my input and find some pattern to the decline.</p>
            </div>
          </div>
        </blockquote>
        <div><br class="">
        </div>
        <div>Hi Josh,</div>
        <div><br class="">
        </div>
        <div>On behalf of the “LLVM people”, I’d like to publicly thank
          you for the this work to make sure that LLVM compile time
          performance stays acceptable over time.  I think that the Rust
          community’s diligence in terms of tracking, improving, and
          publicizing compile time performance has been incredibly
          valuable.  Thank you,</div>
      </div>
    </blockquote>
    <p>Thank you, Chris, I'll certainly share that feedback with the
      Rust community.</p>
    <p>If there's anything specific I can do to make such data more
      actionable, please let me know. I already learned from my Red Hat
      peers that I didn't have a full understanding of llc vs opt etc.,
      so I need to explore more to frame the test in a way that's closer
      to what rustc actually does.<br>
    </p>
    <p><br>
    </p>
    <p>Josh<br>
    </p>
  </body>
</html>