Hi!
On 2019-12-13T17:43:57+0000, Andrew Stubbs <[email protected]> wrote:
> On 19/11/2019 12:21, Andrew Stubbs wrote:
>> This patch adds GCN special casing for most of the OpenACC libgomp tests
>> that require it.
>
> [...] I've gone ahead and committed the attached.
> * testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c: [Handle
> gcn.]
> --- a/libgomp/testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c
> +++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c
> @@ -106,6 +106,8 @@ static void cb_enqueue_launch_start (acc_prof_info
> *prof_info, acc_event_info *e
> assert (event_info->launch_event.vector_length >= 1);
> else if (acc_device_type == acc_device_nvidia) /* ... is special. */
> assert (event_info->launch_event.vector_length == 32);
> + else if (acc_device_type == acc_device_gcn) /* ...and so is this. */
> + assert (event_info->launch_event.vector_length == 64);
> else
> {
> #ifdef __OPTIMIZE__
> @@ -118,6 +120,8 @@ static void cb_enqueue_launch_start (acc_prof_info
> *prof_info, acc_event_info *e
>
> if (acc_device_type == acc_device_host)
> assert (api_info->device_api == acc_device_api_none);
> + else if (acc_device_type == acc_device_gcn)
> + assert (api_info->device_api == acc_device_api_other);
> else
> assert (api_info->device_api == acc_device_api_cuda);
> assert (api_info->valid_bytes == _ACC_API_INFO_VALID_BYTES);
(Someone please scold me for making 'acc_device_api_cuda' the default
'else' case here, without 'if (acc_device_type == acc_device_nvidia)'...)
To make this testcase work in the current GCN setting, I've pushed "Fix
'libgomp.oacc-c-c++-common/acc_prof-kernels-1.c' for 'acc_device_radeon'"
to master branch in commit 984df1e1630f262d782c00cefad2643b8e8469f8, see
attached. (... to be adjusted again, later...)
Grüße
Thomas
-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstrasse 201, 80634 München
Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Frank
Thürauf
>From 984df1e1630f262d782c00cefad2643b8e8469f8 Mon Sep 17 00:00:00 2001
From: Thomas Schwinge <[email protected]>
Date: Sun, 6 Jun 2021 10:41:18 +0200
Subject: [PATCH] Fix 'libgomp.oacc-c-c++-common/acc_prof-kernels-1.c' for
'acc_device_radeon'
... on top of r279378 (commit 26b74ed0223d108d7d7818c3c860f20cfe81a4af)
"Update OpenACC tests for amdgcn".
libgomp/
* testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c: Fix
for 'acc_device_radeon'.
---
.../testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c
index ad33f72e2fb..7f74ee922b7 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c
@@ -93,6 +93,11 @@ static void cb_enqueue_launch_start (acc_prof_info *prof_info, acc_event_info *e
}
if (num_workers < 1)
assert (event_info->launch_event.num_workers >= 1);
+ /* GCN currently enforces 'num_workers (1)'. */
+ else if (acc_device_type == acc_device_radeon
+ /*TODO ... just not in the "Parallelism dimensions: variable" case. */
+ && /*TODO*/ num_gangs != 22)
+ assert (event_info->launch_event.num_workers == 1);
else
{
#ifdef __OPTIMIZE__
--
2.30.2
