Is there a general survey on the basic algorithms for the compressed sparse rows format (like transposition, multiplication, addition, ...)? While it is not hard to write effective algorithms for that, I am not aware of any comprehensive piece of writing on these algorithmic tasks.
This does not defy that many libraries already in existence perform these operations reasonable well.
addendum: The most efficient algorithms that I am aware of generally boil down to assembling (row,column,value) triplets from the input, and building a CSR structure from this.