Data transfer unit with support for multiple coherency granules

A method for transferring a data block, comprising:

• determining a first and a second coherency granule from a user programmable coherency granule field;
• determining a data block size corresponding to the data block;
• requesting a first receive sub-block of the data block from a source corresponding to the first coherency granule, wherein the first receive sub-block has a size that is at most the first coherency granule;
• receiving the first receive sub-block from the source;
• and
• providing a first transmit sub-block of the data block to a destination corresponding to the second coherency granule, wherein the first transmit sub-block has a size that is at most the second coherency granule, and wherein at least a portion of the first receive sub-block and a portion of the first transmit sub-block are a same portion of the data block.

A system, comprising:

• a first processor having a first cache memory and a first coherency granule corresponding to the first cache memory;
• a second processor having a second cache memory and a second coherency granule corresponding to second cache memory, the second coherency granule different from the first coherency granule;
• and
• a data transfer unit, coupled to the first processor and the second processor, for transferring data between the first processor and the second processor, wherein the data transfer unit comprises:
• attribute storage circuitry having a first coherency granule field for storing the first coherency granule and a second coherency granule field for storing the second coherency granule.