<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p>After speed test:<br>
GCC 7.0.1 and GCC 8.0.0 generates identical EXEs with -std=c++11
and -std=gnu++11.<br>
The rest relative encoding time (to GCC 6.3 gnu++11)<br>
</p>
<table cellspacing="0" cellpadding="0" width="128" border="0">
<colgroup><col style="width:48pt" span="2" width="64"> </colgroup><tbody>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt;width:48pt" height="19" width="64">x265c48</td>
<td class="xl65" style="width:48pt" align="right" width="64">101,16%</td>
</tr>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt" height="19">x265g48</td>
<td class="xl65" align="right">101,15%</td>
</tr>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt" height="19">x265c49</td>
<td class="xl65" align="right">100,67%</td>
</tr>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt" height="19">x265g49</td>
<td class="xl65" align="right">100,92%</td>
</tr>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt" height="19">x265c54</td>
<td class="xl65" align="right">101,50%</td>
</tr>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt" height="19">x265g54</td>
<td class="xl65" align="right">101,49%</td>
</tr>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt" height="19">x265c63</td>
<td class="xl65" align="right">99,95%</td>
</tr>
<tr style="height:14.5pt" height="19">
<td style="height:14.5pt" height="19">x265g63</td>
<td class="xl65" align="right">100,00%</td>
</tr>
</tbody>
</table>
<br>
For GCC 4.8.5 and GCC 5.4 -- gnu++11 wins, for GCC 4.9.4 and GCC 6.3
-- c++11 wins.<br>
<br>
I think it is OK to set -std=c++11 for GCC.<br>
<div class="moz-cite-prefix"><br>
<br>
W dniu 2017-04-26 o 12:27, Mateusz Brzostek pisze:<br>
</div>
<blockquote type="cite"
cite="mid:c7eb65a2-4fdd-57c7-3d1b-49bb294f95f0@msystem.waw.pl">
<pre wrap="">I will make speed test -std=c++11 vs. -std=gnu++11 (in the night) and if x265 build with GCC -std=c++11 is not slower than with -std=gnu++11 I will be convinced to -std=c++11.
W dniu 2017-04-26 o 08:07, Michael Lackner pisze:
</pre>
<blockquote type="cite">
<pre wrap="">Actually, I compiled x265 2.4+2 just yesterday using GCC 6.2.0 on Linux, and I had it set
to '-std=c++11'. It compiled and ran just fine doing a quick test run with some 1080p content.
On a side note: Also worked just fine with clang/llvm 3.9.1 on FreeBSD 10.3 UNIX,
currently encoding 4K/UHD stuff.
On 04/25/2017 04:43 PM, Stephen Hutchinson wrote:
</pre>
<blockquote type="cite">
<pre wrap="">On 4/25/2017 4:25 AM, Mateusz Brzostek wrote:
</pre>
<blockquote type="cite">
<pre wrap="">After tests with GCC from 4.8 to 8.0 I think it is OK to change to c++11 from c++98 (for
-DENABLE_DYNAMIC_HDR10=ON and OFF).
Selur (Hybrid author) wrote that for Clang it should be c++11 instead of gnu++11, so we
can set for Clang c++11, for GCC gnu++11.
Please review.
</pre>
</blockquote>
<pre wrap="">Something that's been on my mind since the discussion around
the -std= being set started is: is x265 even using the GNU
extensions that require setting gnu++* variants as the -std?
Or is it just being set like that because it's GCC?
If GCC-specific stuff isn't being used, or if Clang is tolerant
enough of any GNU extensions that are used, you could just set
it as -std=c++11 for both Clang and GCC and not have to branch.
</pre>
</blockquote>
</blockquote>
<pre wrap="">
_______________________________________________
x265-devel mailing list
<a class="moz-txt-link-abbreviated" href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a>
<a class="moz-txt-link-freetext" href="https://mailman.videolan.org/listinfo/x265-devel">https://mailman.videolan.org/listinfo/x265-devel</a>
</pre>
</blockquote>
<br>
</body>
</html>