[clblas] 01/54: Fix kernel crash on nvidia, caused by float4 alignemtn error, see https://github.com/clMathLibraries/clBLAS/issues/108 for more details
Ghislain Vaillant
ghisvail-guest at moszumanska.debian.org
Thu Jan 14 20:07:32 UTC 2016
This is an automated email from the git hooks/post-receive script.
ghisvail-guest pushed a commit to branch debian/sid
in repository clblas.
commit dff63b437da37ffcf23af2f18d27c42889f49a74
Author: Hugh Perkins <hughperkins at gmail.com>
Date: Thu Jun 18 18:15:53 2015 +0800
Fix kernel crash on nvidia, caused by float4 alignemtn error, see https://github.com/clMathLibraries/clBLAS/issues/108 for more details
---
src/library/blas/gens/clTemplates/ger.cl | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/src/library/blas/gens/clTemplates/ger.cl b/src/library/blas/gens/clTemplates/ger.cl
index 0006087..99bbcbd 100644
--- a/src/library/blas/gens/clTemplates/ger.cl
+++ b/src/library/blas/gens/clTemplates/ger.cl
@@ -52,7 +52,8 @@ __kernel void %PREFIXger_C_kernel( __global %TYPE const* restrict _X, __global %
}
// create local memory
- __local %TYPE localX[ BH * %V ];
+ __local %TYPE%V localXV[ BH ];
+ __local %TYPE *localX = (__local %TYPE *)localXV;
__local %TYPE localY[ BW ];
uint lID = get_local_id( 0 );
@@ -193,7 +194,8 @@ __kernel void %PREFIXger_R_kernel( __global %TYPE const* restrict _X, __global %
}
__local %TYPE localX[ BH ];
- __local %TYPE localY[ BW * %V ];
+ __local %TYPE%V localYV[ BW ];
+ __local %TYPE *localY = (__local %TYPE *)localYV;
uint lID = get_local_id( 0 );
uint gID = get_group_id( 0 );
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-science/packages/clblas.git
More information about the debian-science-commits
mailing list