From: Shark8 <OneWingedShark@gmail.com>
Subject: Re: "The Machine" and Ada (blog post)
Date: Sun, 13 Jul 2014 13:05:43 -0600
Date: 2014-07-13T13:05:43-06:00 [thread overview]
Message-ID: <cMAwv.205458$nc6.133857@fx27.iad> (raw)
In-Reply-To: <aa0290e4-b4dc-45ac-9c46-dac9e2daa11f@googlegroups.com>
On 13-Jul-14 09:29, Dan'l Miller wrote:
> To which existing pragma are you referring?
It wouldn't *need* to be a pre-existing pragma.
Implementation-defined pragmas [and aspects] are allowed for just this
reason.
e.g. Imagine an Ada compiler targeting CPU + CUDA.
Task type K is
Entry Do_something( Data : in Integer );
Entry Get_Result ( Data : out Integer );
End K with
Target_Processor => GPU;
could be used to signify the task is to be run on one of the GPU cores;
the nice thing about this is that it would still be compilable [and
runnable] w/o the CUDA targeting... though it may raise a unrecognized
pragma/aspect when compiled.
prev parent reply other threads:[~2014-07-13 19:05 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-06-21 12:50 “The Machine” and Ada (blog post) Victor Porton
2014-06-21 14:16 ` "The Machine" " Dan'l Miller
2014-06-21 17:31 ` ³The Machine² " Bill Findlay
2014-06-21 19:43 ` Victor Porton
2014-06-21 20:49 ` Bill Findlay
2014-06-21 21:47 ` Victor Porton
2014-06-21 22:27 ` Bill Findlay
2014-06-22 8:50 ` Georg Bauhaus
2014-06-22 9:59 ` Dmitry A. Kazakov
2014-06-22 12:32 ` Bill Findlay
2014-06-21 21:49 ` Dan'l Miller
2014-06-21 22:28 ` Bill Findlay
2014-06-22 20:56 ` Stephen Leake
2014-07-01 10:02 ` Bill Findlay
2014-07-09 0:43 ` Stephen Leake
2014-06-22 16:08 ` ?The Machine? " Robert Love
2014-06-22 17:25 ` invalid
2014-07-11 21:43 ` "The Machine" " Dan'l Miller
2014-07-12 10:41 ` Victor Porton
2014-07-13 15:29 ` Dan'l Miller
2014-07-13 19:05 ` Shark8 [this message]
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox