public inbox for fortran@gcc.gnu.org
 help / color / mirror / Atom feed
From: Cesar Philippidis <cesar_philippidis@mentor.com>
To: Vikram Singh <vikramsingh001@gmail.com>
Cc: "Thomas Schwinge" <thomas@codesourcery.com>,
	"Salvatore Filippone" <filippone.salvatore@gmail.com>,
	"Vladimír Fuka" <vladimir.fuka@gmail.com>,
	"James Norris" <jnorris@codesourcery.com>,
	"Chung-Lin Tang" <cltang@codesourcery.com>,
	"Fortran List" <fortran@gcc.gnu.org>
Subject: Re: OpenACC-Library-Interoperability
Date: Mon, 29 Aug 2016 16:16:00 -0000	[thread overview]
Message-ID: <84ac193d-5cfe-277c-732e-73bfa89ab6f8@mentor.com> (raw)
In-Reply-To: <CAD0gq3WGtSo7bg3vBo_T1WmVt6pnJOthgkM9s6CGqmmVttEuyg@mail.gmail.com>

On 08/29/2016 08:38 AM, Vikram Singh wrote:

> z from the cublasdgemm call gives 0 everywhere, unlike the actual
> dgemm call. The first line of output is
> 
>  ** On entry to DGEMM  parameter number 1 had an illegal value

Check the documentation for cublas. You may not be calling that function
with the proper arguments.

> In addition, I looked at your test case. You do not do
> 
>  !$acc update self(y)
> 
> Does that mean that the host variable is automatically updated in gfortran.

The host_data clause tells the compiler to use variables that have
already been mapped onto the accelerator via acc data or acc enter enter
data. You need to use the data directives to actually move the data as
appropriate. In host_data-2.f90 example, the value of 'y' was going to
be copyied out by the acc data region, so there was no need to use acc
update host.

You only need to use the update directive if you want to retain that
data on the accelerator.

Cesar

  reply	other threads:[~2016-08-29 16:16 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-04-12 15:00 OpenACC-Library-Interoperability Vikram Singh
2016-04-15  6:16 ` OpenACC-Library-Interoperability Thomas Schwinge
2016-04-15  8:36   ` OpenACC-Library-Interoperability Vikram Singh
2016-04-15  8:58     ` OpenACC-Library-Interoperability Thomas Schwinge
2016-04-15 11:00       ` OpenACC-Library-Interoperability Vikram Singh
2016-05-12 15:35         ` OpenACC-Library-Interoperability Thomas Schwinge
2016-05-12 16:41           ` OpenACC-Library-Interoperability Vikram Singh
2016-05-12 17:54             ` OpenACC-Library-Interoperability Vikram Singh
     [not found]           ` <CAD0gq3VoRWCiXRkgi-bnGLBfSjR-bFc0Mzp19LRr+yWP4MrYLg@mail.gmail.com>
     [not found]             ` <CANSzZf6md-w8SZOJEeawThYbVfH0cLgNRS9r5VaADqZKdy6KtA@mail.gmail.com>
2016-08-02 16:56               ` OpenACC-Library-Interoperability Salvatore Filippone
2016-08-03 15:53             ` OpenACC-Library-Interoperability Vikram Singh
2016-08-29 13:59               ` OpenACC-Library-Interoperability Vikram Singh
2016-08-29 14:05                 ` OpenACC-Library-Interoperability James Norris
2016-08-29 14:15                   ` OpenACC-Library-Interoperability Vikram Singh
2016-08-29 14:34                 ` OpenACC-Library-Interoperability Cesar Philippidis
2016-08-29 15:39                   ` OpenACC-Library-Interoperability Vikram Singh
2016-08-29 16:16                     ` Cesar Philippidis [this message]
2016-08-31  9:43                       ` OpenACC-Library-Interoperability Vikram Singh
     [not found] <CAKe2ite0OWGjtQtHkLY-FDxqJLXmDbKAWOiRLqTP+cUS1-qWog@mail.gmail.com>
2016-04-27 17:16 ` OpenACC-Library-Interoperability Vladimír Fuka
2016-04-15 14:36   ` OpenACC-Library-Interoperabilit Salvatore Filippone
2016-05-09 14:27     ` [PATCH, Fortran, OpenACC] Fix PR70598, Fortran host_data ICE Chung-Lin Tang
2016-05-10 18:58       ` Bernhard Reutner-Fischer
2016-06-07 12:04         ` Chung-Lin Tang
2016-06-21  6:18           ` [PATCH, Fortran, OpenACC] Fix PR70598, Fortran host_data ICE (ping x2) Chung-Lin Tang
2016-07-13 11:53             ` [PATCH, Fortran, OpenACC] Fix PR70598, Fortran host_data ICE (ping x3) Chung-Lin Tang
2016-07-21  9:29               ` [PATCH, Fortran, OpenACC] Fix PR70598, Fortran host_data ICE (ping x4) Chung-Lin Tang
2016-07-21 10:54                 ` Paul Richard Thomas
2016-07-21 11:13       ` [PATCH, Fortran, OpenACC] Fix PR70598, Fortran host_data ICE Jakub Jelinek
2016-07-29 15:47         ` Chung-Lin Tang
2016-08-09 15:30           ` Jakub Jelinek

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=84ac193d-5cfe-277c-732e-73bfa89ab6f8@mentor.com \
    --to=cesar_philippidis@mentor.com \
    --cc=cltang@codesourcery.com \
    --cc=filippone.salvatore@gmail.com \
    --cc=fortran@gcc.gnu.org \
    --cc=jnorris@codesourcery.com \
    --cc=thomas@codesourcery.com \
    --cc=vikramsingh001@gmail.com \
    --cc=vladimir.fuka@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).