Closed sepehrMSP closed 2 years ago
This condition is used for conventional mapping. We assume a memory hierarchy from sub-array to processing element and to tile. To keep the hierarchy, there are at least 2 sub-arrays in one PE and at least 2 PEs in one tile, therefore, there are at least 4 sub-arrays in one tile.
And for novel mapping, the condition is maxPESizeNM < 2*param->numRowSubArray
to make sure at least 2 subarrays in one PE, and there are normally 3*3 PEs in one tile (corresponding to the kernel size).
Hi, can you explain the reason of this condition (used in
chipFloorPlan
)?XTileSize > 4 * param->numRowSubArray
It is logical thatTileSize
be greater thannumRowSubArray
, however I cannot understand why4
is used? it could be a meaningful parameter. @neurosim