<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<p>On 01/13/2017 12:11 PM, Mehdi Amini wrote:<br>
</p>
<blockquote
cite="mid:72AD9CA2-1C0F-404E-A358-E293C42EE1A2@apple.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<br class="">
<div>
<blockquote type="cite" class="">
<div class="">On Jan 13, 2017, at 9:41 AM, Hal Finkel <<a
moz-do-not-send="true" href="mailto:hfinkel@anl.gov"
class="">hfinkel@anl.gov</a>> wrote:</div>
<br class="Apple-interchange-newline">
<div class="">
<div bgcolor="#FFFFFF" text="#000000" class="">
<p class=""><br class="">
</p>
<div class="moz-cite-prefix">On 01/13/2017 12:29 AM, Mehdi
Amini wrote:<br class="">
</div>
<blockquote
cite="mid:5ADA0551-E374-48A7-8311-005E3AA947E4@apple.com"
type="cite" class=""> <br class="">
<div class="">
<blockquote type="cite" class="">
<div class="">On Jan 12, 2017, at 5:02 PM, Hal
Finkel <<a moz-do-not-send="true"
href="mailto:hfinkel@anl.gov" class="">hfinkel@anl.gov</a>>
wrote:</div>
<div class="">
<div bgcolor="#FFFFFF" text="#000000" class="">
<p class="">On 01/12/2017 06:20 PM, Reid
Kleckner via llvm-dev wrote:</p>
<blockquote
cite="mid:CACs=tyKWmdPAQGJkJdrF9xdov-TnV02ofpPURjMGo7rwo28V8w@mail.gmail.com"
type="cite" class="">
<div dir="ltr" class="">
<div class="gmail_extra">
<div class="gmail_quote">On Wed, Jan 11,
2017 at 8:13 PM, Mehdi Amini <span
dir="ltr" class=""><<a
moz-do-not-send="true"
href="mailto:mehdi.amini@apple.com"
target="_blank" class="">mehdi.amini@apple.com</a>></span>
wrote:
<blockquote class="gmail_quote"
style="margin:0 0 0
.8ex;border-left:1px #ccc
solid;padding-left:1ex">
<div style="word-wrap:break-word"
class="">
<div class="">
<div class="">Can you elaborate
why? I’m curious.</div>
</div>
</div>
</blockquote>
<div class=""><br class="">
</div>
<div class="">The con of proposal c was
that many passes would need to learn
about many region intrinsics. With
tokens, you only need to teach all
passes about tokens, which they should
already know about because WinEH and
other things use them.</div>
<div class=""><br class="">
</div>
<div class="">With tokens, we can add as
many region-introducing intrinsics as
makes sense without any additional
cost to the middle end. We don't need
to make one omnibus region intrinsic
set that describes every parallel loop
annotation scheme supported by LLVM.
Instead we would factor things
according to other software design
considerations.</div>
</div>
</div>
</div>
</blockquote>
<br class="">
I think that, unless we allow frontends to add
their own intrinsics without recompiling LLVM,
this severely restricts the usefulness of this
feature.</div>
</div>
</blockquote>
<div class=""><br class="">
</div>
<div class="">I’m not convinced that “building a
frontend without recompiling LLVM while injecting
custom passes” is a strong compelling use-case, i.e.
can you explain why requiring such
use-case/frontends to rebuild LLVM is so limiting?</div>
</div>
</blockquote>
<br class="">
I don't understand your viewpoint. Many frontends either
compose their own pass pipelines or use the existing
extension-point mechanism. Some frontends, Chapel for
example, can insert code using custom address spaces and
then insert passes later to turn accesses using pointers
to those address spaces into runtime calls. This is the
kind of design we'd like to support, without forcing
frontends to use custom versions of LLVM, but with
annotated regions instead of just with address spaces.<br
class="">
</div>
</div>
</blockquote>
<div><br class="">
</div>
<div>I think we’re talking about two different things here: you
mentioned originally “without recompiling LLVM”, which I don’t
see as major blocker, while now you’re now clarifying I think
that you’re more concerned about putting a requirement on a
*custom* LLVM, as in “it wouldn’t work with the source from a
vanilla upstream LLVM”, which I agree is a different story.</div>
<div><br class="">
</div>
<div>That said, it extends the point from the other email (in
parallel) about the semantics of the intrinsics: while your
solution allows these frontend to reuse the intrinsics, it
means that upstream optimization have to consider such
intrinsics as optimization barrier because their semantic is
unknown.</div>
</div>
</blockquote>
<br>
I see no reason why this needs to be true (at least so long as
you're willing to accept a certain amount of "as if" parallelism).
Moreover, if it is true, then we'll lose the benefits of, for
example, being able to hoist scalar loads out of parallel loops. We
might need to include dependencies on "inaccessible memory", so
cover natural runtime dependencies by default (this can be refined
with custom AA logic), but that is not a complete code-motion
barrier. Memory being explicitly managed will end up as arguments to
the region intrinsics, so we'll automatically get more-fine-grained
information.<br>
<br>
-Hal<br>
<br>
<blockquote
cite="mid:72AD9CA2-1C0F-404E-A358-E293C42EE1A2@apple.com"
type="cite">
<div>
<div><br class="">
</div>
<div><br class="">
</div>
<div>— </div>
<div>Mehdi</div>
</div>
</blockquote>
<br>
<pre class="moz-signature" cols="72">--
Hal Finkel
Lead, Compiler Technology and Programming Languages
Leadership Computing Facility
Argonne National Laboratory</pre>
</body>
</html>