Hi, I am somewhat new to Boost.uBlas, but having examined it a bit, it is not clear to me that it supports block storage organization, as opposed to row major or column major. The intent of this capability is to  take into account the cache size of the architecture to optimize sub-matrix operations.

The usual way of representing matrix in memory does not always take full advantage of the cache line. For example, matrix multiplication could be made faster if all elements of certain sub-matrixes are near-by (in a contiguous address range). This can be achieved by using a blocked storage organization for matrix. Is there interest to take a look at this in uBlas  or are there already existing ways to support this?

Thanks.

Michael Wong
XL C++ Compiler kernel Development
IBM Canada Ltd., C2/KD2/8200/MKM
8200 Warden Avenue
Markham, Ontario  L6G 1C7
W:905-413-3283 F:905-413-4839
Boost test results   http://www-1.ibm.com/support/docview.wss?rs=2239&context=SSJT9L&uid=swg27006911
C/C++ Compilers Support Page   http://www.ibm.com/software/awdtools/ccompilers/support/
C/C++ Feature Request Interface     http://www.ibm.com/support/docview.wss?uid=swg27005811
XL Fortran Compiler Support Page     http://www.ibm.com/software/awdtools/fortran/xlfortran/support/
XL Fortran Feature Request Interface     http://www.ibm.com/support/docview.wss?uid=swg27005812