Re: 1.21 vs 1.3 encoding speed
Scott Brown <[email protected]> Mon, 28 Jul 2014 08:22:04 -0400
| Newsgroups | gmane.comp.audio.compression.flac.user |
|---|---|
| Message-ID | <CANJq19J2egKE+8zxhCrvRATU5_9_jDPhKeHnAizympmxVeynhQ@mail.gmail.com> |
--===============1033168066== Content-Type: multipart/alternative; boundary=089e0122f5fab6e13f04ff3ff91a --089e0122f5fab6e13f04ff3ff91a Content-Type: text/plain; charset=UTF-8 I forgot to post to this list yesterday, so here's my latest email to the dev list: --------- Forwarded message ---------- From: Scott Brown <[email protected]> Date: Sun, Jul 27, 2014 at 8:16 AM Subject: Re: [flac-dev] 1.21 vs 1.3 encoding speed To: Martijn van Beurden <[email protected]> Cc: [email protected] Adding -O3 now has these results: 51.81 seconds for the file that took 222 seconds without -O3. So that was it. Thank you! Scott On Sat, Jul 26, 2014 at 3:47 AM, Martijn van Beurden <[email protected]> wrote: op 25-07-14 19:32, Scott Brown schreef: > ./configure -enable-static -disable-shared CFLAGS=" -isysroot > /Developer/SDKs/MacOSX10.6.sdk -mmacosx-version-min=10.6" > make Well, the use of CFLAGS 'disables' the -O3 and unroll-loops optimisation. I'm quite sure that's the culprit. Add -O3 to your CFLAGS and it should be fixed. See the following lines in the configure script > if test "x$user_cflags" = x; then > CFLAGS="-O3 -funroll-loops -Wall -W -Winline" > fi Maybe we should change this 'overriding' behaviour of CFLAGS and add -O3 just like most other options are added to the CFLAGS? ______________________________ _________________ flac-dev mailing list [email protected] http://lists.xiph.org/mailman/listinfo/flac-dev On Sat, Jul 26, 2014 at 11:38 AM, <[email protected]> wrote: > Please cc: the results from dev list back here though > > I would run some tests on a diff platform but don't have access to my PC > for a few weeks > > Would also be interesting to see what decoding stats you get > > On Jul 25, 2014, at 7:38 AM, Scott Brown <[email protected]> wrote: > > I will post to the dev list, sorry about that. > > To give an idea, though, at flac level 8: > > 24/96 wav file > 1.21: 61.05 seconds. ratio=0.690 > 1.3: 222.48 seconds. ratio=0.690 > > smaller 16/44.1 wav file > 1.21: 14.28 seconds. ratio=0.487 > 1.3: 51.21 seconds. ratio=0.487 > > Scott > > > On Fri, Jul 25, 2014 at 2:39 AM, <[email protected]> wrote: > >> Are you sure they didn't change the default encoding level ? >> >> I would include some example timings to give a better idea of what >> "significantly slower" is >> >> Wonder if you see same for 44/16 files? >> >> On Jul 25, 2014, at 2:29 AM, Martijn van Beurden <[email protected]> >> wrote: >> >> Hi, >> >> You might want to report this to the flac-dev mailinglist instead of the >> flac (users) list. >> >> op 24-07-14 22:02, Scott Brown schreef: >> >> Hello, >> >> I'm on a Mac and I'm noticing that encoding via the flac command line is >> significantly slower with version 1.3.0 than 1.2.1. I'm encoding a 24/96 >> file to flac, both from wav and aiff and both formats are showing the same >> speed decrease when using 1.3.0. >> >> Also, 1.2.1 will not encode an aiff-c file, but 1.3.0 will. Is this >> change documented anywhere? >> >> Thanks, >> Scott >> >> >> _______________________________________________ >> Flac mailing [email protected]://lists.xiph.org/mailman/listinfo/flac >> >> >> _______________________________________________ >> Flac mailing list >> [email protected] >> http://lists.xiph.org/mailman/listinfo/flac >> >> >> _______________________________________________ >> Flac mailing list >> [email protected] >> http://lists.xiph.org/mailman/listinfo/flac >> >> > --089e0122f5fab6e13f04ff3ff91a Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable <div dir=3D"ltr">I forgot to post to this list yesterday, so here's my = latest email to the dev list:<br><br>--------- Forwarded message ----------= <br>From: <b class=3D"gmail_sendername">Scott Brown</b> <span dir=3D"ltr">&= lt;<a href=3D"mailto:[email protected]">[email protected]</a>></= span><br> Date: Sun, Jul 27, 2014 at 8:16 AM<br>Subject: Re: [flac-dev] 1.21 vs 1.3 e= ncoding speed<br>To: Martijn van Beurden <<a href=3D"mailto:mvanb1@gmail= .com">[email protected]</a>><br>Cc: <a href=3D"mailto:[email protected]">= [email protected]</a><br> <br><br><div dir=3D"ltr">Adding -O3 now has these results:<div><br></div><d= iv>51.81 seconds for the file that took 222 seconds without -O3. So that wa= s it.</div><div><br></div><div>Thank you!</div><span class=3D""><font color= =3D"#888888"><div> <br></div><div>Scott</div></font></span></div> <br><br>On Sat, Jul 26, 2014 at 3:47 AM, Martijn van Beurden <span dir=3D"l= tr"><<a href=3D"mailto:[email protected]" target=3D"_blank">mvanb1@gmail.= com</a>></span> wrote:<br> op 25-07-14 19:32, Scott Brown schreef:<br> <div>> ./configure -enable-static -disable-shared CFLAGS=3D" -isysr= oot<br> > /Developer/SDKs/MacOSX10.6.sdk -mmacosx-version-min=3D10.6"<br> > make<br> <br> </div>Well, the use of CFLAGS 'disables' the -O3 and unroll-loops<b= r> optimisation. I'm quite sure that's the culprit. Add -O3 to your<br= > CFLAGS and it should be fixed.<br> <br> See the following lines in the configure script<br> <br> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 if test "x$user_cflags" =3D x; t= hen<br> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 CFLAGS=3D"-O3 -funroll-= loops -Wall -W -Winline"<br> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 fi<br> <br> Maybe we should change this 'overriding' behaviour of CFLAGS and<br= > add -O3 just like most other options are added to the CFLAGS?<br> <br> ______________________________<div>_________________<br> flac-dev mailing list<br> <a href=3D"mailto:[email protected]" target=3D"_blank">[email protected]</a= ><br> <a href=3D"http://lists.xiph.org/mailman/listinfo/flac-dev" target=3D"_blan= k">http://lists.xiph.org/mailman/listinfo/flac-dev</a></div><br><div><div c= lass=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On Sat, Jul 26, 201= 4 at 11:38 AM, <span dir=3D"ltr"><<a href=3D"mailto:[email protected]"= target=3D"_blank">[email protected]</a>></span> wrote:<br> <blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-= left:1px solid rgb(204,204,204);padding-left:1ex"><div dir=3D"auto"><div>Pl= ease cc: the results from dev list back here though</div><div><br></div><di= v> I would run some tests on a diff platform but don't have access to my P= C for a few weeks</div> <div><br></div><div>Would also be interesting to see what decoding stats yo= u get</div><div><div><div><br>On Jul 25, 2014, at 7:38 AM, Scott Brown <= <a href=3D"mailto:[email protected]" target=3D"_blank">scottcbrown@gmai= l.com</a>> wrote:<br> <br></div><blockquote type=3D"cite"><div><div dir=3D"ltr"><div>I will post = to the dev list, sorry about that.<br><br>To give an idea, though, at flac = level 8:<br><br></div><div>24/96 wav file<br></div><div>1.21: 61.05 seconds= . ratio=3D0.690<br> </div><div>1.3: 222.48 seconds. ratio=3D0.690<br> </div><div><br></div><div>smaller 16/44.1 wav file<br></div><div>1.21: 14.2= 8 seconds. ratio=3D0.487<br></div><div>1.3: 51.21 seconds. ratio=3D0.487<br= ><br></div>Scott<br></div><div class=3D"gmail_extra"><br><br><div class=3D"= gmail_quote"> On Fri, Jul 25, 2014 at 2:39 AM, <span dir=3D"ltr"><<a href=3D"mailto:d= [email protected]" target=3D"_blank">[email protected]</a>></span> wrot= e:<br><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;b= order-left:1px solid rgb(204,204,204);padding-left:1ex"> <div dir=3D"auto"><div>Are you sure they didn't change the default enco= ding level ?</div><div><br></div><div>I would include some example timings = to give a better idea of what "significantly slower" is</div><div= > <br></div><div>Wonder if you see same for 44/16 files?</div><div><div><div>= <br>On Jul 25, 2014, at 2:29 AM, Martijn van Beurden <<a href=3D"mailto:= [email protected]" target=3D"_blank">[email protected]</a>> wrote:<br> <br></div><blockquote type=3D"cite"><div> =20 =20 =20 =20 Hi,<br> <br> You might want to report this to the flac-dev mailinglist instead of the flac (users) list.<br> <br> <div>op 24-07-14 22:02, Scott Brown schreef:<br> </div> <blockquote type=3D"cite"> <div dir=3D"ltr"> <div> <div> <div>Hello,<br> <br> </div> I'm on a Mac and I'm noticing that encoding via the fla= c command line is significantly slower with version 1.3.0 than 1.2.1. I'm encoding a 24/96 file to flac, both from wav and aiff and both formats are showing the same speed decrease when using 1.3.0.<br> <br> </div> Also, 1.2.1 will not encode an aiff-c file, but 1.3.0 will. Is this change documented anywhere?<br> <br> </div> Thanks,<br> Scott<br> </div> <br> <fieldset></fieldset> <br> <pre>_______________________________________________ Flac mailing list <a href=3D"mailto:[email protected]" target=3D"_blank">[email protected]</a> <a href=3D"http://lists.xiph.org/mailman/listinfo/flac" target=3D"_blank">h= ttp://lists.xiph.org/mailman/listinfo/flac</a> </pre> </blockquote> <br> =20 </div></blockquote><blockquote type=3D"cite"><div><span>___________________= ____________________________</span><br><span>Flac mailing list</span><br><s= pan><a href=3D"mailto:[email protected]" target=3D"_blank">[email protected]</a></s= pan><br> <span><a href=3D"http://lists.xiph.org/mailman/listinfo/flac" target=3D"_bl= ank">http://lists.xiph.org/mailman/listinfo/flac</a></span><br></div></bloc= kquote></div></div></div><br>______________________________________________= _<br> Flac mailing list<br> <a href=3D"mailto:[email protected]" target=3D"_blank">[email protected]</a><br> <a href=3D"http://lists.xiph.org/mailman/listinfo/flac" target=3D"_blank">h= ttp://lists.xiph.org/mailman/listinfo/flac</a><br> <br></blockquote></div><br></div> </div></blockquote></div></div></div></blockquote></div><br></div></div></d= iv> --089e0122f5fab6e13f04ff3ff91a-- --===============1033168066== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Flac mailing list [email protected] http://lists.xiph.org/mailman/listinfo/flac --===============1033168066==--