You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello! I can help you with the translation of your question into English. Here's the translation:
"Hello, I still don't understand some details about designing SpMM, so I would like to ask specifically. When v = 2, the LHS matrix is sparse. How can it satisfy the data layout of m8n32k16? Moreover, the matrix is sparse, and some columns have values of 0. How can we obtain the correct result in this case?"
The text was updated successfully, but these errors were encountered:
Hello! I can help you with the translation of your question into English. Here's the translation:
"Hello, I still don't understand some details about designing SpMM, so I would like to ask specifically. When v = 2, the LHS matrix is sparse. How can it satisfy the data layout of m8n32k16? Moreover, the matrix is sparse, and some columns have values of 0. How can we obtain the correct result in this case?"

The text was updated successfully, but these errors were encountered: