Loading src/TNL/Algorithms/Segments/SlicedEllpackView.hpp +17 −1 Original line number Diff line number Diff line Loading @@ -244,7 +244,15 @@ forSegments( IndexType first, IndexType last, Function& f, Args... args ) const IndexType localIdx( 0 ); bool compute( true ); for( IndexType globalIdx = begin; globalIdx < end && compute; globalIdx++ ) { // The following is a workaround of a bug in nvcc 11.2 #if CUDART_VERSION == 11020 f( segmentIdx, localIdx, globalIdx, compute, args... ); localIdx++; #else f( segmentIdx, localIdx++, globalIdx, compute, args... ); #endif } }; Algorithms::ParallelFor< Device >::exec( first, last, l, args... ); } Loading @@ -259,7 +267,15 @@ forSegments( IndexType first, IndexType last, Function& f, Args... args ) const IndexType localIdx( 0 ); bool compute( true ); for( IndexType globalIdx = begin; globalIdx < end && compute; globalIdx += SliceSize ) { // The following is a workaround of a bug in nvcc 11.2 #if CUDART_VERSION == 11020 f( segmentIdx, localIdx, globalIdx, compute, args... ); localIdx++; #else f( segmentIdx, localIdx++, globalIdx, compute, args... ); #endif } }; Algorithms::ParallelFor< Device >::exec( first, last, l, args... ); } Loading Loading
src/TNL/Algorithms/Segments/SlicedEllpackView.hpp +17 −1 Original line number Diff line number Diff line Loading @@ -244,7 +244,15 @@ forSegments( IndexType first, IndexType last, Function& f, Args... args ) const IndexType localIdx( 0 ); bool compute( true ); for( IndexType globalIdx = begin; globalIdx < end && compute; globalIdx++ ) { // The following is a workaround of a bug in nvcc 11.2 #if CUDART_VERSION == 11020 f( segmentIdx, localIdx, globalIdx, compute, args... ); localIdx++; #else f( segmentIdx, localIdx++, globalIdx, compute, args... ); #endif } }; Algorithms::ParallelFor< Device >::exec( first, last, l, args... ); } Loading @@ -259,7 +267,15 @@ forSegments( IndexType first, IndexType last, Function& f, Args... args ) const IndexType localIdx( 0 ); bool compute( true ); for( IndexType globalIdx = begin; globalIdx < end && compute; globalIdx += SliceSize ) { // The following is a workaround of a bug in nvcc 11.2 #if CUDART_VERSION == 11020 f( segmentIdx, localIdx, globalIdx, compute, args... ); localIdx++; #else f( segmentIdx, localIdx++, globalIdx, compute, args... ); #endif } }; Algorithms::ParallelFor< Device >::exec( first, last, l, args... ); } Loading