Fix a bug in a packed block type in TensorContractionThreadPool

This commit is contained in:
Eugene Zhulenev 2019-09-24 16:54:36 -07:00
parent d38e6fbc27
commit f35b9ab510

View File

@ -750,7 +750,7 @@ struct TensorEvaluator<const TensorContractionOp<Indices, LeftArgType, RightArgT
template <typename EvalCtx>
struct ThreadLocalBlocksAllocator</*pack_rhs=*/false, EvalCtx> {
static void allocate(EvalCtx& ctx, Blocks& blocks) {
std::vector<RhsBlock> lhs_blocks;
std::vector<LhsBlock> lhs_blocks;
BlockMemHandle mem_handle = ctx.kernel_.allocateSlices(
ctx.device_,
/*num_lhs=*/ctx.gm_,
@ -758,7 +758,7 @@ struct TensorEvaluator<const TensorContractionOp<Indices, LeftArgType, RightArgT
/*num_slices=*/1,
/*lhs_blocks=*/&lhs_blocks, /*rhs_blocks=*/nullptr);
blocks = ThreadLocalBlocks<RhsBlock>(std::move(mem_handle),
blocks = ThreadLocalBlocks<LhsBlock>(std::move(mem_handle),
std::move(lhs_blocks));
}