Abstract: In an aspect, in general, a programmable computation device performs computations of an inference task specified by a plurality of variables and a plurality of factors, each factor being associated with a subset of the variables. The device includes one or more processing elements. Each processing element includes a first storage for a definition of a factor, a second storage for data associated with the inputs and/or outputs of at least some of the computations, and one or more computation units coupled to the first storage and the second storage for performing a succession of parts of the at least some of the computations that are associated with a factor, the succession of parts defined by data in the storage for the definition of the factor.

Abstract: A two-level skewing architecture is imposed on the memory subsystem of a volume rendering system in which voxel data is stored in mini-blocks assigned to a set of DRAM memory modules, thereby permitting data transfer at the maximum burst rate of the DRAM memory and enabling real-time volume rendering. Within each DRAM module, mini-blocks are assigned to the memory banks so that consecutively accessed mini-blocks are assigned to different banks, thereby avoiding idle cycles during data transfer and increasing DRAM transfer efficiency to nearly 100%. In one embodiment, read-out of voxel data from banks of the DRAM memory proceeds from left to right unless there is a conflict of banks, in which case the read-out order is reversed. A specialized de-skewing network is provided to re-order the voxel data read out from DRAM memory so that the voxels can be processed in the order which they are arranged in the volume data set rather than the order in which they are stored in memory.