Lines Matching refs:chunksperblock
17 PetscInt chunksperblock; member
185 …seqsell_tiled_kernel8(PetscInt nrows, PetscInt sliceheight, PetscInt chunksperblock, PetscInt tota… in matmult_seqsell_tiled_kernel8() argument
196 for (int iter = 0; iter < chunksperblock; iter++) { in matmult_seqsell_tiled_kernel8()
197 cid = blockIdx.x * chunksperblock + iter; /* chunk id */ in matmult_seqsell_tiled_kernel8()
217 …if (iter == chunksperblock - 1 || (cid + 2) * BLOCKY * 32 > sliidx[start_slice + 1]) { /* last ite… in matmult_seqsell_tiled_kernel8()
240 …seqsell_tiled_kernel8(PetscInt nrows, PetscInt sliceheight, PetscInt chunksperblock, PetscInt tota… in matmultadd_seqsell_tiled_kernel8() argument
251 for (int iter = 0; iter < chunksperblock; iter++) { in matmultadd_seqsell_tiled_kernel8()
252 cid = blockIdx.x * chunksperblock + iter; /* chunk id */ in matmultadd_seqsell_tiled_kernel8()
272 …if (iter == chunksperblock - 1 || (cid + 2) * BLOCKY * 32 > sliidx[start_slice + 1]) { /* last ite… in matmultadd_seqsell_tiled_kernel8()
579 PetscInt chunksperblock, nchunks, *chunk_slice_map; in MatMult_SeqSELLCUDA() local
662 …chunksperblock = cudastruct->chunksperblock ? cudastruct->chunksperblock : 1 + (cudastruct->total… in MatMult_SeqSELLCUDA()
663 nblocks = 1 + (nchunks - 1) / chunksperblock; in MatMult_SeqSELLCUDA()
666 …qsell_tiled_kernel8<2><<<nblocks, dim3(32, 2)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMult_SeqSELLCUDA()
668 …qsell_tiled_kernel8<4><<<nblocks, dim3(32, 4)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMult_SeqSELLCUDA()
670 …qsell_tiled_kernel8<8><<<nblocks, dim3(32, 8)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMult_SeqSELLCUDA()
672 …ell_tiled_kernel8<16><<<nblocks, dim3(32, 16)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMult_SeqSELLCUDA()
674 …ell_tiled_kernel8<32><<<nblocks, dim3(32, 32)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMult_SeqSELLCUDA()
676 …qsell_tiled_kernel8<2><<<nblocks, dim3(32, 2)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMult_SeqSELLCUDA()
720 PetscInt chunksperblock, nchunks, *chunk_slice_map; in MatMultAdd_SeqSELLCUDA() local
758 …chunksperblock = cudastruct->chunksperblock ? cudastruct->chunksperblock : 1 + (cudastruct->total… in MatMultAdd_SeqSELLCUDA()
759 nblocks = 1 + (nchunks - 1) / chunksperblock; in MatMultAdd_SeqSELLCUDA()
762 …qsell_tiled_kernel8<2><<<nblocks, dim3(32, 2)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
764 …qsell_tiled_kernel8<4><<<nblocks, dim3(32, 4)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
766 …qsell_tiled_kernel8<8><<<nblocks, dim3(32, 8)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
768 …ell_tiled_kernel8<16><<<nblocks, dim3(32, 16)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
770 …ell_tiled_kernel8<32><<<nblocks, dim3(32, 32)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
772 …qsell_tiled_kernel8<2><<<nblocks, dim3(32, 2)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
823 …chunksperblock = cudastruct->chunksperblock ? cudastruct->chunksperblock : 1 + (cudastruct->total… in MatMultAdd_SeqSELLCUDA()
824 nblocks = 1 + (nchunks - 1) / chunksperblock; in MatMultAdd_SeqSELLCUDA()
827 …qsell_tiled_kernel8<2><<<nblocks, dim3(32, 2)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
829 …qsell_tiled_kernel8<4><<<nblocks, dim3(32, 4)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
831 …qsell_tiled_kernel8<8><<<nblocks, dim3(32, 8)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
833 …ell_tiled_kernel8<16><<<nblocks, dim3(32, 16)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
835 …ell_tiled_kernel8<32><<<nblocks, dim3(32, 32)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
837 …qsell_tiled_kernel8<2><<<nblocks, dim3(32, 2)>>>(nrows, sliceheight, chunksperblock, nchunks, chun… in MatMultAdd_SeqSELLCUDA()
890 …ptionsGetInt(NULL, NULL, "-mat_sell_spmv_cuda_chunksperblock", &cudastruct->chunksperblock, &flg)); in MatSetFromOptions_SeqSELLCUDA()