eric-haibin-lin commented on a change in pull request #10371: [MXNET-263]
Support for dot(dns, csr) = dns and dot(dns, csr.T) = dns on GPU
URL: https://github.com/apache/incubator-mxnet/pull/10371#discussion_r180481585
##########
File path: src/operator/tensor/dot-inl.cuh
##########
@@ -442,6 +442,99 @@ struct DotCsrRspDnsScalarKernel {
}
};
+/*!
+ * \brief GPU Kernel to re-arrange nnz elements to csc order
+ * Parallelization by output elements: 1 thread/row of csr
+ */
+struct CscDataIndicesKernel {
+ template<typename DType, typename IType, typename CType>
+ __device__ __forceinline__ static void Map(int tid,
+ const DType* csr_data,
+ const IType* csr_indices,
+ const CType* csr_indptr,
+ DType* csc_data,
+ unsigned long long* csc_indices,
+ unsigned long long* csc_indptr,
+ unsigned long long* workspace,
Review comment:
Rename "workspace" to "col_counters"? Better document what this space is
used for:
e.g. used to count the offset of column indices atomically
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
