<div dir="ltr">Hi,<br><div><div class="gmail_extra"><br><div class="gmail_quote">2014-09-29 13:50 GMT+02:00 Matthias Kretz <span dir="ltr"><<a href="mailto:kretz@compeng.uni-frankfurt.de" target="_blank">kretz@compeng.uni-frankfurt.de</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi,<br>
<br>
just a quick answer to let you know that I have some high priority work to do<br>
the next two weeks...<br>
<br>
IIUC your code vectorizes inside each object (vertical vectorization) not over<br>
multiple objects (horizontal vectorization), right. In my experience vertical<br>
vectorization requires shuffles, horizontal vectorization doesn't.<br>
At this point Vc is much better suited for horizontal vectorization...<br>
<br></blockquote><div><br></div><div>Yes, I fully agree :) Also, gcc / icc will tell you hints towards horizontal vectorization<br>always. Vertical is usually more tough (and more fun).<br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Once you do vertical vectorization you rely on the number of entries in the<br>
SIMD register anyway and would be better of with something like<br>
Vc::simdarray<float, 4> instead of Vc::float_v. With Vc::simdarray a shuffle<br>
API will be much easier to define, because the number of entries is portable<br>
for a given type.<br>
<br></blockquote><div><br></div><div>That's very nice input. I'll have a look into developing something myself if I find some time.<br></div><div>Was this something you guys intended to support in the short future?<br></div><div><br>Thanks,<br><br></div><div>Daniel<br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
-Matthias<br>
<br>
--<br>
─────────────────────────────────────────────────────────────<br>
Dipl.-Phys. Matthias Kretz<br>
<span class=""><br>
Web: <a href="http://compeng.uni-frankfurt.de/?mkretz" target="_blank">http://compeng.uni-frankfurt.de/?mkretz</a><br>
<br>
SIMD easy and portable: <a href="http://compeng.uni-frankfurt.de/?vc" target="_blank">http://compeng.uni-frankfurt.de/?vc</a><br>
─────────────────────────────────────────────────────────────<br>
</span>_______________________________________________<br>
Vc mailing list<br>
<a href="mailto:Vc@compeng.uni-frankfurt.de">Vc@compeng.uni-frankfurt.de</a><br>
<a href="https://compeng.uni-frankfurt.de/mailman/listinfo/vc" target="_blank">https://compeng.uni-frankfurt.de/mailman/listinfo/vc</a><br>
</blockquote></div><br><br clear="all"><br>-- <br>Daniel Hugo Cámpora Pérez<br>European Organization for Nuclear Research (CERN)<br>PH LBC, LHCb Online Fellow<br>e-mail. <a href="mailto:dcampora@cern.ch" target="_blank">dcampora@cern.ch</a><br>tel. intern 166473<br><div>tel. +41 (0) 764876473<div><br></div></div>
</div></div></div>