<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html;
      charset=windows-1252">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <p>Hi,</p>
    <p>the best way to move forward, and I guess the answer to your
      question, is to get your work to build and test on try.</p>
    <p>And now I'm going to flood you with jargon, sorry. Instead of
      trying to document that all here, I suggest you ask people for
      help on irc. Notably, #taskcluster, and #build. Folks there are
      your target audience for compiler changes, too.<br>
    </p>
    <p>First read is
      <a class="moz-txt-link-freetext" href="https://wiki.mozilla.org/ReleaseEngineering/TryServer">https://wiki.mozilla.org/ReleaseEngineering/TryServer</a>, just to get
      the first piece of jargon out.</p>
    <p>I suspect that the next steps are:</p>
    <p>I speculate that you can use your version of llvm if you can
      create a docker image that is able to compile Firefox. Please
      confirm that with #taskcluster, and get some help there on how to
      do that.</p>
    <p>Then you'd want to make sure that you can actually compile/run
      mozilla-central. In particular wrt to taskcluster, there's been
      various file moves and changes, so working off of release is going
      to make your path rockier. I understand that mozilla-central is
      more of a moving target, but that shouldn't be that much of a
      problem for you in practice, I hope.</p>
    <p>I suspect that just focusing on linux x64 is good for you for
      now?</p>
    <p>So jargon-thought-train:</p>
    <p>Validate my assumptions about all of this ;-) .<br>
    </p>
    <p>Make your setup work with mozilla-central.</p>
    <p>Work in docker.</p>
    <p>Make your mozilla-central's taskcluster tasks for linux x64 pick
      your docker image to build on try.</p>
    <p>Push to try for linux x64, with all tests and perf tests.</p>
    <p>Use treeherder and perfherder (web uis) to see tests and
      performance.</p>
    <p><br>
    </p>
    <p>If you want to focus on your toolchain impact over time instead
      of mozilla-central, I think you should be able to keep the same
      base version of your patch and just update the docker image, and
      then use perfherder to compare the results.</p>
    <p>HTH</p>
    <p>Axel<br>
    </p>
    <br>
    <div class="moz-cite-prefix">On 19/06/16 20:41, vivek pandya wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAHYgpoLzP4UFD=P0KbfcA94y6mkY6Cn9Jt1R9mUycaf3UvVK1w@mail.gmail.com">
      <div dir="ltr">
        <div>
          <div>Hello,</div>
          <div><br>
          </div>
          <div>I build FireFox-46.0.1 source with llvm to test
            interprocedural register allocation.</div>
          <div>The build was successful with out any runtime faliures,
            here are few stats:</div>
          <div><font size="1"><br>
            </font></div>
          <div><font size="1">Measure <span class="" style="white-space:pre">                                                                           </span>W/O
              IPRA<span class="" style="white-space:pre">     </span>WITH
              IPRA</font></div>
          <div><font size="1">=======<span class="" style="white-space:pre">                                                                            </span>========<span class="" style="white-space:pre">  </span>=========</font></div>
          <div><font size="1">Total Build Time<span class="" style="white-space:pre">                                                           </span>76
              mins<span class="" style="white-space:pre">             </span>82.3
              mins<span class="" style="white-space:pre">     </span>8%
              increment</font></div>
          <div><font size="1">Octane v2.0 JS Benchmark Score (higher is
              better)<span class="" style="white-space:pre">  </span>18675.69 <span class="" style="white-space:pre"> </span>19665.16<span class="" style="white-space:pre">  </span>5%
              improvement</font></div>
          <div><font size="1">Kraken JS Benchmark time (lower is better)<span class="" style="white-space:pre">                 </span>1416.2
              ms <span class="" style="white-space:pre">      </span>1421.3
              ms <span class="" style="white-space:pre">      </span>0.35%
              regression</font></div>
          <div><font size="1">JetStream JS Benchmark Score (higer is
              better)<span class="" style="white-space:pre">          </span>110.10<span class="" style="white-space:pre">            </span>112.88<span class="" style="white-space:pre">            </span>2.52%
              improvement</font></div>
          <div><br>
          </div>
          <div>Any suggestions are welcome on how to effectively measure
            performance improvement!</div>
          <div><br>
          </div>
          <div>Sincerely,</div>
          <div>Vivek</div>
        </div>
        <div><br>
        </div>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
firefox-dev mailing list
<a class="moz-txt-link-abbreviated" href="mailto:firefox-dev@mozilla.org">firefox-dev@mozilla.org</a>
<a class="moz-txt-link-freetext" href="https://mail.mozilla.org/listinfo/firefox-dev">https://mail.mozilla.org/listinfo/firefox-dev</a>
</pre>
    </blockquote>
    <br>
  </body>
</html>