From: Tom de Vries <tdevries@suse.de>
To: Jakub Jelinek <jakub@redhat.com>
Cc: Tobias Burnus <tobias@codesourcery.com>,
GCC Patches <gcc-patches@gcc.gnu.org>,
Kwok Cheung Yeung <kcy@codesourcery.com>,
Thomas Schwinge <thomas@codesourcery.com>,
Segher Boessenkool <segher@linux.ibm.com>
Subject: Re: [RFC][nvptx, libgomp] Add 128-bit atomic support
Date: Wed, 2 Sep 2020 13:48:15 +0200 [thread overview]
Message-ID: <898949a8-7f24-6606-1d8d-af0df3914b9a@suse.de> (raw)
In-Reply-To: <20200902104415.GF18149@tucnak>
On 9/2/20 12:44 PM, Jakub Jelinek wrote:
> On Wed, Sep 02, 2020 at 12:22:28PM +0200, Tom de Vries wrote:
>> And test-case passes on x86_64 with this patch (obviously, in
>> combination with trigger patch above).
>>
>> Jakub, WDYT?
>
> I guess the normal answer would be use libatomic, but it isn't ported for
> nvptx.
Ah, I was not aware of that one, filed
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96898 to look into that.
> I guess at least temporarily this is ok, though I'm wondering why
> you need __sync_*_16 rather than __atomic_*_16,
That's what omp-expand.c uses in expand_omp_atomic_pipeline:
BUILT_IN_SYNC_VAL_COMPARE_AND_SWAP_N .
Thanks,
- Tom
> or perhaps both __sync_* and
> __atomic_*.
>
> What happens if you try
> unsigned __int128 v;
> #pragma omp declare target (v)
> int
> main ()
> {
> #pragma omp target
> {
> __atomic_add_fetch (&v, 1, __ATOMIC_RELAXED);
> __atomic_fetch_add (&v, 1, __ATOMIC_RELAXED);
> unsigned __int128v exp = 2;
> __atomic_compare_exchange_n (&v, &expected, 7, 0, __ATOMIC_RELEASE, __ATOMIC_ACQUIRE);
> }
> }
> etc. (see some gcc.dg/atomic* tests, ditto for __sync_*)?
> I guess better not to throw everything into one test, because not every
> target supports them all (e.g. I think x86_64 doesn't really do 128-bit
> atomic loads because the cmpxchg16b insn are not appropriate for .rodata
> locations).
>
> Jakub
>
next prev parent reply other threads:[~2020-09-02 11:48 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-15 20:28 [PATCH] nvptx: Add support for subword compare-and-swap Kwok Cheung Yeung
2020-06-23 16:44 ` Thomas Schwinge
2020-06-23 16:51 ` Jakub Jelinek
2020-06-30 16:35 ` Kwok Cheung Yeung
2020-06-24 11:13 ` Kwok Cheung Yeung
2020-06-30 14:37 ` Tom de Vries
2020-07-01 14:28 ` Tom de Vries
2020-07-15 19:08 ` Kwok Cheung Yeung
2020-07-20 13:19 ` Kwok Cheung Yeung
2020-08-04 14:56 ` [PING] " Kwok Cheung Yeung
2020-08-13 9:27 ` Tom de Vries
2020-09-01 11:41 ` [patch][nvptx] libgomp: Split testcase in order to XFAIL __sync_val_compare_and_swap_16 (was: [PATCH] nvptx: Add support for subword compare-and-swap) Tobias Burnus
2020-09-01 12:58 ` Tom de Vries
2020-09-02 7:56 ` Tom de Vries
2020-09-02 10:22 ` [RFC][nvptx, libgomp] Add 128-bit atomic support Tom de Vries
2020-09-02 10:44 ` Jakub Jelinek
2020-09-02 11:30 ` Tobias Burnus
2020-09-02 11:48 ` Tom de Vries [this message]
2020-09-11 14:24 ` Tom de Vries
2020-09-11 14:25 ` Tom de Vries
2020-09-11 14:48 ` Andrew Stubbs
2020-09-11 15:03 ` tdevries
2020-09-11 15:29 ` Tobias Burnus
2020-09-11 14:37 ` 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=898949a8-7f24-6606-1d8d-af0df3914b9a@suse.de \
--to=tdevries@suse.de \
--cc=gcc-patches@gcc.gnu.org \
--cc=jakub@redhat.com \
--cc=kcy@codesourcery.com \
--cc=segher@linux.ibm.com \
--cc=thomas@codesourcery.com \
--cc=tobias@codesourcery.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).