Abstract: A programmable motion estimator includes one dual ported memory for storing an image block, the prediction error, and a temporary block used in interpolation, and a pixel-group random access dual ported memory for storing a search window. The two ports of the two memories are selectively applied to an arithmetic logic unit, or ALU, through a multiplexer. One output of the ALU provides an absolute difference, which is furnished to a tree adder. Another output of the ALU provides an average value or a difference value, as selected, which is routed to the inputs of the image memory and the search memory. In motion vector searching, the ALU performs pixel absolute difference arithmetic using the pixel groups from the image memory and from the search memory, and determines a sum of absolute differences in the tree adder. In half pixel interpolation, the ALU performs pixel averaging arithmetic using pixel groups from the search memory, and writes back to the search memory.

Abstract: A multimedia processor contains a general purpose RISC and video processors which operate in parallel to execute software for combined video and audio bit stream coding and decoding. The RISC processor controls operation of the multimedia processor and performs bit stream parsing and coding, audio compression and decompression, and general processing for embedded applications. The video processor performs video encoding and decoding functions such as scaling, filtering, decimation, and DCT transforms. The RISC processor and the video processor each have separate data buses which are interconnected through a portal circuit and a Huffman codec. Each data bus has a DMA controller which transfers data to and from a memory interface to an external memory. DMA channels serve I/O interface resources coupled to the data buses and can form buffers in the external memory. This reduces the need for on-chip FIFO buffers and separate buffers between the multimedia processor and attached devices.

Abstract: A programmable motion estimator includes one dual ported memory for storing an image block, the prediction error, and a temporary block used in interpolation, and a pixel-group random access dual ported memory for storing a search window. The two ports of the two memories are selectively applied to an arithmetic logic unit, or ALU, through a multiplexer. One output of the ALU provides an absolute difference, which is furnished to a tree adder. Another output of the ALU provides an average value or a difference value, as selected, which is routed to the inputs of the image memory and the search memory. Among other tasks, the programmable motion estimator performs motion vector searching, half pixel interpolation, quarter pixel interpolation and error prediction determination.

Abstract: A vision processor includes a control section, a motion estimation section, and a discrete cosine transform ("DCT") section. The motion estimation section includes two memories, an image memory with two read ports and a write port, and a search memory with two read ports and a write port. The DCT section includes a DCT memory configurable as a two read, two write port memory and as a four read, four write port memory. The ports of these memories are selectively applied to various elements in the motion estimation path and the DCT path. In motion vector searching, the ALU performs averaging and difference operations on pixels in the frame and search memories. Data from the search memory is shifted for certain operations, before arithmetic operations in the ALU are performed.

Abstract: A programmable motion estimator includes one dual ported memory for storing an image block, the prediction error, and a temporary block used in interpolation, and a pixel-group random access dual ported memory for storing a search window. The two ports of the two memories are selectively applied to an arithmetic logic unit, or ALU, through a multiplexer. One output of the ALU provides an absolute difference, which is furnished to a tree adder. Another output of the ALU provides an average value or a difference value, as selected, which is routed to the inputs of the image memory and the search memory. In motion vector searching, the ALU performs pixel absolute difference arithmetic using the pixel groups from the image memory and from the search memory, and determines a sum of absolute differences in the tree adder. In half pixel interpolation, the ALU performs pixel averaging arithmetic using pixel groups from the search memory, and writes back to the search memory.

Abstract: A vision processor includes a control section, a motion estimation section, and a discrete cosine transform ("DCT") section. The motion estimation section includes two memories, an image memory with two read ports and a write port, and a search memory with two read ports and a write port. The DCT section includes a DCT memory configurable as a two read, two write port memory and as a four read, four write port memory. The ports of these memories are selectively applied to various elements in the motion estimation path and the DCT path. In motion vector searching, the ALU performs averaging and difference operations on pixels in the frame and search memories. Data from the search memory is shifted for certain operations, before arithmetic operations in the ALU are performed.

Abstract: An improved logic structure and a method for implementing the same to perform division and square-root operations for radix four and higher is disclosed. The divsion and square-root bits are generated by a non-restoring method with the partial remainder, partial radicand, quotient and root all in redundant form. The partial remainder/radicand is stored in a series of sum and carry registers. The upper bits from these registers are supplied to a carry look-ahead adder for conversion to non-redundant form. These upper bits are then used to select a next divisor or root from a prediction programmable logic array (PLA). The output of the prediction PLA is supplied to a quotient/root register and a divisor/root multiple selector. The output of the selector is supplied to a carry save adder which has its output provided back to the input of the partial remainder/radicand sum and carry registers. The system of the present invention allows both division and square root calculations to be done with the same hardware.

Abstract: A method and apparatus for combining the multiply and ALU functions for floating point numbers to enable the completion of a multiply-accumulate operation in a shorter time. The multiplied fraction is left in sum and carry form and is provided in this form to the ALU, eliminating the CP adder from the multiplier. The normalization of the fraction and the corresponding changes to the exponent in the multiplier are also eliminated. The ALU can combine the sum and carry of the product fraction simultaneously if the exponents are sufficiently similar. Otherwise, the sum and carry of the fraction product is combined first and compared with the new fraction, with the smaller of the fractions being right shifted prior to their combination.