<html><body>
<p><font size="2" face="sans-serif">Thanks, I have also forwarded your N document and associated discussion to the OpenMP MM group including Mark Batty</font><br>
<font size="2" face="sans-serif"><br>
</font><br>
<font size="3" color="#003F69" face="serif">_________________________________________________________</font><font size="1" color="#808080" face="Arial"><b><br>
Regards, Michael<br>
ISOCPP.org Director, VP <a href="http://isocpp.org/wiki/faq/wg21#michael-wong">http://isocpp.org/wiki/faq/wg21#michael-wong</a><br>
OpenMP CEO: <a href="http://openmp.org/wp/about-openmp/">http://openmp.org/wp/about-openmp/</a> <br>
My Blogs: <a href="http://ibm.co/pCvPHR">http://ibm.co/pCvPHR</a> <br>
C++11 status: <a href="http://tinyurl.com/43y8xgf">http://tinyurl.com/43y8xgf</a> <br>
Boost test results <a href="http://www.ibm.com/support/docview.wss?rs=2239&context=SSJT9L&uid=swg27006911">http://www.ibm.com/support/docview.wss?rs=2239&context=SSJT9L&uid=swg27006911</a> <br>
C/C++ Compilers Feature Request Page <a href="http://www.ibm.com/developerworks/rfe/?PROD_ID=700">http://www.ibm.com/developerworks/rfe/?PROD_ID=700</a> <br>
Chair of WG21 SG5 Transactional Memory: <a href="https://groups.google.com/a/isocpp.org/forum/?hl=en&fromgroups#!forum/tm">https://groups.google.com/a/isocpp.org/forum/?hl=en&fromgroups#!forum/tm</a> </b></font><font size="3" face="serif"><br>
</font><font size="1" color="#808080" face="Arial"><br>
IBM Corporation<br>
XL C++ Compiler kernel Development <br>
IBM z Systems Software,IBM Systems Unit <br>
IBM Canada Ltd., C2/KD2/8200/MKM <br>
8200 Warden Avenue <br>
Markham, Ontario L6G 1C7 <br>
W:905-413-3283 F:905-413-4839</font><font size="1" color="#00B0DA" face="Arial"><u><br>
</u></font><a href="http://www.openmpcon.org/" target="_new"><font size="1" color="#00B0DA" face="Arial"><u>OpenMPCon 2015</u></font></a><font size="1" color="#808080" face="Arial"><br>
<br>
</font><img src="cid:1__=0ABBF4B0DFF7188D8f9e8a93df938@ca.ibm.com" width="250" height="100"><br>
<br>
<img width="16" height="16" src="cid:2__=0ABBF4B0DFF7188D8f9e8a93df938@ca.ibm.com" border="0" alt="Inactive hide details for JF Bastien ---04/10/2015 01:22:40 PM---Architecture-specific NaCl basically uses whatever the underly"><font size="2" color="#424282" face="sans-serif">JF Bastien ---04/10/2015 01:22:40 PM---Architecture-specific NaCl basically uses whatever the underlying ISA expects (so answers will be "s</font><br>
<br>
<font size="1" color="#5F5F5F" face="sans-serif">From: </font><font size="1" face="sans-serif">JF Bastien <jfb@chromium.org></font><br>
<font size="1" color="#5F5F5F" face="sans-serif">To: </font><font size="1" face="sans-serif">Hal Finkel <hfinkel@anl.gov></font><br>
<font size="1" color="#5F5F5F" face="sans-serif">Cc: </font><font size="1" face="sans-serif">LLVM Dev <llvmdev@cs.uiuc.edu>, Chandler Carruth <chandlerc@gmail.com>, Daniel Berlin <dberlin@dberlin.org>, Robin Morisset <robin.morisset@normalesup.org>, t p northover <t.p.northover@gmail.com>, James Molloy <james@jamesmolloy.co.uk>, Tom Stellard <tom@stellard.net>, renato golin <renato.golin@linaro.org>, Michael Wong/Toronto/IBM@IBMCA, Alexey Bataev <a.bataev@gmx.com>, Bill Schmidt <wschmidt@linux.vnet.ibm.com></font><br>
<font size="1" color="#5F5F5F" face="sans-serif">Date: </font><font size="1" face="sans-serif">04/10/2015 01:22 PM</font><br>
<font size="1" color="#5F5F5F" face="sans-serif">Subject: </font><font size="1" face="sans-serif">Re: Optimization on Atomics (and the OpenMP memory model)</font><br>
<font size="1" color="#5F5F5F" face="sans-serif">Sent by: </font><font size="1" face="sans-serif">jfb@google.com</font><br>
<hr width="100%" size="2" align="left" noshade style="color:#8091A5; "><br>
<br>
<br>
<font size="3" face="serif">Architecture-specific NaCl basically uses whatever the underlying ISA expects (so answers will be "same as the above"), save for any OS interactions which go through NaCl's own syscalls (which then sometimes punt to the host OS). The most interesting here is futex, which is a restricted version of Linux' futex and gets emulated on non-Linux guests.</font><br>
<br>
<font size="3" face="serif">For PNaCl none of the C++ constructs get lowered past their C++ semantics (they look like functions calls). futex is still a syscall for PNaCl. C++ constructs only get lowered once the actual ISA is known (translation time) at which point we're usually generating NaCl code and the same as the above applies.</font><br>
<br>
<font size="3" face="serif">The same as PNaCl would apply to JavaScript once it supports SharedArrayBuffer if the current proposal goes forward. One small change would be to rely on synchronic instead of futex.</font><br>
<br>
<font size="3" face="serif">I'm not familiar enough with taskwait, omp_set_lock / omp_unset_lock to provide a useful answer. I assume that we can implement them with atomics and futex if QOI isn't an issue?</font><br>
<br>
<font size="3" face="serif">On reordering: yes, as much as C++ allows. Good timing for this:</font>
<ul style="padding-left: 24pt"><a href="https://github.com/jfbastien/papers/blob/master/source/N4455.rst"><font size="3" color="#0000FF" face="serif"><u>https://github.com/jfbastien/papers/blob/master/source/N4455.rst</u></font></a></ul>
<br>
<font size="3" face="serif">On Fri, Apr 10, 2015 at 10:12 AM, Hal Finkel <</font><a href="mailto:hfinkel@anl.gov" target="_blank"><font size="3" color="#0000FF" face="serif"><u>hfinkel@anl.gov</u></font></a><font size="3" face="serif">> wrote:</font>
<ul style="padding-left: 9pt"><font size="3" face="serif">Hi everyone,<br>
<br>
The OpenMP standards committee has begun work to formalize their memory model, and define its relationship to the C/C++ memory models. A questionnaire has been put together (pasted below), and I'd like everyone's help in composing detailed answers to inform their decision-making process. While our OpenMP support is still in active development, many of these questions apply equally to C/C++ atomics, and a lot of work has certainly been done here on that front.<br>
<br>
* Which processor architectures does your compiler target (e.g. x86, Power, ARM, ARM v8, Xeon Phi, Nvidia GPUs, etc.)?<br>
[I'll just answer "yes" for that one ;)]<br>
* What is a flush lowered to in assembly for each of the supported architectures? For instance, a flush might be implemented as an MFENCE on the x86 architecture in some compilers.<br>
* What are non-seq_cst atomic read, write, update and capture lowered to for each of your targets?<br>
* What are seq_cst atomic read, write, update and capture lowered to for each of your targets?<br>
* What is the taskwait construct lowered to for each of your targets?<br>
* What are omp_set_lock and omp_unset_lock lowered to for each of your targets?<br>
* What is a barrier lowered to for each of your targets?<br>
* Are any optimisations allowed to reorder, change or remove code that uses any of the synchronisation constructs above, or any of the other synchronisation constructs in section 2.12 of the OpenMP 4.0 specification?<br>
<br>
I'll be happy to collate answers to send back to the committee; please provide as much feedback as you can.<br>
<br>
Thanks in advance,<br>
Hal</font><font size="3" color="#888888" face="serif"><br>
<br>
--<br>
Hal Finkel<br>
Assistant Computational Scientist<br>
Leadership Computing Facility<br>
Argonne National Laboratory</font></ul>
<br>
</body></html>