<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Hi Nataraj<br>
    </p>
    <p>Which processor is that run on? (although too close to call, it
      implies LEA has a latency of 2 in that case)<br>
    </p>
    <p>Kit<br>
    </p>
    <div class="moz-cite-prefix">On 08/10/2023 14:06, Nataraj S Narayan
      via fpc-devel wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAK8HZO+JFu=UoV7f9qW+rURFz7CxmYy=A9rKx1rCjhSrTch=XA@mail.gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      <div dir="ltr">
        <div>Hi</div>
        <div><br>
        </div>
        <div>[nataraj@dflyHP ~]$ fpc ttt.pas</div>
        Free Pascal Compiler version 3.2.2 [2023/07/04] for x86_64<br>
        Copyright (c) 1993-2021 by Florian Klaempfl and others<br>
        Target OS: DragonFly for x86-64<br>
        Compiling ttt.pas<br>
        Linking ttt<br>
        /usr/local/bin/ld.bfd: warning:
        /usr/local/lib/fpc/3.2.2/units/x86_64-dragonfly/rtl/prt0.o:
        missing .note.GNU-stack section implies executable stack<br>
        /usr/local/bin/ld.bfd: NOTE: This behaviour is deprecated and
        will be removed in a future version of the linker<br>
        121 lines compiled, 14.9 sec<br>
        [nataraj@dflyHP ~]$ ./ttt <br>
           Pascal control case: 6.7 ns/call<br>
         Using LEA instruction: 4.2 ns/call<br>
        Using ADD instructions: 4.0 ns/call<br>
        <div>
          <div dir="ltr" class="gmail_signature"
            data-smartmail="gmail_signature">
            <div dir="ltr">
              <div>
                <div dir="ltr">
                  <div>
                    <div dir="ltr"><br>
                    </div>
                    <div dir="ltr"><br>
                    </div>
                    <div dir="ltr">Nataraj S Narayan
                      <div>Synergy Info Systems</div>
                      <div>Software & Technology Consultants</div>
                      <div>Ettumanoor, INDIA</div>
                      <div>Ph:+91 9443211326</div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </div>
        </div>
        <br>
      </div>
      <br>
      <div class="gmail_quote">
        <div dir="ltr" class="gmail_attr">On Sat, Oct 7, 2023 at 9:39 PM
          J. Gareth Moreton via fpc-devel <<a
            href="mailto:fpc-devel@lists.freepascal.org"
            moz-do-not-send="true" class="moz-txt-link-freetext">fpc-devel@lists.freepascal.org</a>>
          wrote:<br>
        </div>
        <blockquote class="gmail_quote" style="margin:0px 0px 0px
          0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">That's
          interesting; I am interested to see the assembly output for
          the <br>
          Pascal control cases.  As for the 64-bit version, that was my
          fault <br>
          since the assembly language is for Microsoft's ABI rather than
          the <br>
          System V ABI, so it was checking a register with an undefined
          value.  <br>
          Find attached the fixed test.<br>
          <br>
          Kit<br>
          <br>
          P.S. Results on my Intel(R) Core(TM) i7-10750H<br>
          <br>
              Pascal control case: 2.0 ns/call<br>
            Using LEA instruction: 1.7 ns/call<br>
          Using ADD instructions: 1.3 ns/call<br>
          <br>
          On 07/10/2023 16:51, Tomas Hajny via fpc-devel wrote:<br>
          > On 2023-10-07 03:57, J. Gareth Moreton via fpc-devel
          wrote:<br>
          ><br>
          ><br>
          > Hi Kit,<br>
          ><br>
          >> Do you think this should suffice? Originally it ran
          for 1,000,000<br>
          >> repetitions but I fear that will take way too long on
          a 486, so I<br>
          >> reduced it to 10,000.<br>
          ><br>
          > OK, I tried it now. First of all, after turning on the
          old machine, I <br>
          > realized that it wasn't Intel but AMD 486 DX4 - sorry for
          my bad <br>
          > memory. :-( I compiled and ran the test under OS/2 there
          (I was too <br>
          > lazy to boot it to DOS ;-) ), but I assume that it
          shouldn't make any <br>
          > substantial difference. The ADD and LEA results were
          basically the <br>
          > same there, both around 100 ns / call. The Pascal result
          was around <br>
          > twice as long. Interestingly, the Pascal result for FPC
          3.2.2 was <br>
          > around 10% longer than the same source compiled with FPC
          2.0.3 (the <br>
          > assembler versions were obviously the same for both FPC
          versions; I <br>
          > tried compiling it also with FPC 1.0.10 and the assembler
          versions <br>
          > were more than three times slower due to missing support
          for the <br>
          > nostackframe directive).<br>
          ><br>
          > I tested it under the AMD Athlon 1 GHz machine as well
          and again, the <br>
          > results for LEA and ADD are basically equal (both 3.1
          ns/call) and the <br>
          > result for Pascal slightly more than twice (7.3 ns/call).
          However, <br>
          > rather surprisingly for me, the overall test run was
          _much_ longer <br>
          > there?! Finally, I tried compiling the test on a 64-bit
          machine (AMD <br>
          > A9-9425) with Linux (compiled for 64-bits with FPC 3.2.3
          compiled from <br>
          > a fresh 3.2 branch). The Pascal version shows about 4
          ns/call, but the <br>
          > assembler version runs forever - well, certainly much
          longer than my <br>
          > patience lasts. I haven't tried to analyze the reasons,
          but that's <br>
          > what I get.<br>
          ><br>
          > Tomas<br>
          ><br>
          ><br>
          ><br>
          >><br>
          >> On 03/10/2023 06:30, Tomas Hajny via fpc-devel wrote:<br>
          >>> On October 3, 2023 03:32:34 +0200, "J. Gareth
          Moreton via fpc-devel" <br>
          >>> <<a
            href="mailto:fpc-devel@lists.freepascal.org" target="_blank"
            moz-do-not-send="true" class="moz-txt-link-freetext">fpc-devel@lists.freepascal.org</a>>
          wrote:<br>
          >>><br>
          >>><br>
          >>> Hii Kit,<br>
          >>><br>
          >>>> This is mainly to Florian, but also to anyone
          else who can answer <br>
          >>>> the question - at which point did a complex
          LEA instruction (using <br>
          >>>> all three input operands and some other
          specific circumstances) get <br>
          >>>> slow? Preliminary research suggests the 486
          was when it gained <br>
          >>>> extra latency, and then Sandy Bridge when it
          got particularly bad.  <br>
          >>>> Icy Lake seems to be the architecture where
          faster LEA instructions <br>
          >>>> are reintroduced, but I'm not sure about AMD
          processors.<br>
          >>> I cannot answer your question, but if you prepare
          a test program, I <br>
          >>> can run it on an Intel 486 DX2 100 Mhz and AMD
          Athlon 1 GHz machines <br>
          >>> if it helps you in any way (at least I hope the
          486 DX2 machine <br>
          >>> should be still able to start ;-) ).<br>
          >>><br>
          >>> Tomas<br>
          >>><br>
          >>> _______________________________________________<br>
          >>> fpc-devel maillist  -  <a
            href="mailto:fpc-devel@lists.freepascal.org" target="_blank"
            moz-do-not-send="true" class="moz-txt-link-freetext">fpc-devel@lists.freepascal.org</a><br>
          >>> <a
            href="https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel"
            rel="noreferrer" target="_blank" moz-do-not-send="true"
            class="moz-txt-link-freetext">https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel</a><br>
          >>><br>
          >> _______________________________________________<br>
          >> fpc-devel maillist  -  <a
            href="mailto:fpc-devel@lists.freepascal.org" target="_blank"
            moz-do-not-send="true" class="moz-txt-link-freetext">fpc-devel@lists.freepascal.org</a><br>
          >> <a
            href="https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel"
            rel="noreferrer" target="_blank" moz-do-not-send="true"
            class="moz-txt-link-freetext">https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel</a><br>
          > _______________________________________________<br>
          > fpc-devel maillist  -  <a
            href="mailto:fpc-devel@lists.freepascal.org" target="_blank"
            moz-do-not-send="true" class="moz-txt-link-freetext">fpc-devel@lists.freepascal.org</a><br>
          > <a
            href="https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel"
            rel="noreferrer" target="_blank" moz-do-not-send="true"
            class="moz-txt-link-freetext">https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel</a><br>
          >_______________________________________________<br>
          fpc-devel maillist  -  <a
            href="mailto:fpc-devel@lists.freepascal.org" target="_blank"
            moz-do-not-send="true" class="moz-txt-link-freetext">fpc-devel@lists.freepascal.org</a><br>
          <a
            href="https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel"
            rel="noreferrer" target="_blank" moz-do-not-send="true"
            class="moz-txt-link-freetext">https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel</a><br>
        </blockquote>
      </div>
      <br>
      <fieldset class="moz-mime-attachment-header"></fieldset>
      <pre class="moz-quote-pre" wrap="">_______________________________________________
fpc-devel maillist  -  <a class="moz-txt-link-abbreviated" href="mailto:fpc-devel@lists.freepascal.org">fpc-devel@lists.freepascal.org</a>
<a class="moz-txt-link-freetext" href="https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel">https://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel</a>
</pre>
    </blockquote>
  </body>
</html>