Abstract: A CNN inference engine that convolves an input data set with a weight data set is disclosed together with components that facilitate such computation. The engine includes a plurality of multiply and accumulate processors (MACs), each MAC causing a value in the accumulator to be augmented by a product of a data value received on an input data port, a weight value received on a weight port. The engine also includes a slice buffer having a plurality of output ports, each output port being connected to one of the MAC input data value ports. The engine causes the slice buffer to connect one of the slices to the plurality of slice buffer output ports, and causes a weight received on an inference engine weight port to be input to each MAC weight port. The MACs process the input data values on the output ports in the slice in parallel.

Abstract: The present invention includes a method for operating a data processing system to compute an approximation to a scalar product between first and second vectors in which each vector is characterized by N components. The method includes replacing the first vector by a third vector that is a pyramid integer vector characterized by N components and an integer K equal to the sum of the absolute values of the N components, and computing a scalar product of the third vector with the second vector to provide the approximation to the scalar product between the first and second vectors. Computing the scalar product of the second and third vectors can be carried out by K additions followed by one floating point multiply.

Abstract: The present invention includes a method for operating a data processing system to compute an approximation to a scalar product between first and second vectors in which each vector is characterized by N components. The method includes replacing the first vector by a third vector that is a pyramid integer vector characterized by N components and an integer K equal to the sum of the absolute values of the N components, and computing a scalar product of the third vector with the second vector to provide the approximation to the scalar product between the first and second vectors. Computing the scalar product of the second and third vectors can be carried out by K additions followed by one floating point multiply.

Abstract: An encoder and method for using the same to encode a video stream is disclosed. The encoder includes an input port that receives a video stream that includes a plurality of frames. A first compression sub-system generates a plurality of compressed reference frames using a lossy compression method. The compressed reference frames are communicated to a remote device. A reference frame storage sub-system stores one of the compressed reference frames in a compressed format for use in compressing a subsequent frame in the video stream using a lossy compression method. A second compression sub-system encodes a second plurality of frames using the compressed reference frame stored in the reference frame storage sub-system as a reference, the second compression sub-system generating a compressed video frame. Each compressed video frame that is not a reference frame is coded directly from one of the reference frames.

Abstract: A method for operating a data processing system to compare a first block, B1, of pixels in a current frame to a second block, B2, of pixels in the reference frame is disclosed. First and second signature vectors, V1 and V2, respectively, are generated for the first and second blocks. The distance between first and second signature vectors using a distance function D(V1,V2) is measured to provide a comparison of the similarity of the blocks. The signature vectors are chosen such that D(B1,B2)<D(B1,B3) then D(V1,V2)<D(V1,V3), where B3 is a third block of pixels in the reference frame. In addition, the computational workload of comparing the two blocks, on average, using the signature vectors is less than that imposed by directly comparing the blocks.

Abstract: An encoder and method for using the same to encode a video stream is disclosed. The encoder includes an input port that receives a video stream that includes a plurality of frames. A first compression sub-system generates a plurality of compressed reference frames using a lossy compression method. The compressed reference frames are communicated to a remote device. A reference frame storage sub-system stores one of the compressed reference frames in a compressed format for use in compressing a subsequent frame in the video stream using a lossy compression method. A second compression sub-system encodes a second plurality of frames using the compressed reference frame stored in the reference frame storage sub-system as a reference, the second compression sub-system generating a compressed video frame. Each compressed video frame that is not a reference frame is coded directly from one of the reference frames.

Abstract: A method of calculating a color blend for an output image is disclosed. In a fundamental form, the method comprises the steps of specifying a plurality of non-coincident points (x1, y1, z1), (x2, y2, z2), (x3, y3, z3) on a surface (1) to be subject to color blending. Next, a surface equation (Intensity=Ax+By+C) is derived for each color component from a pixel value (z1, z2, z3) of each of the points. The method then determines each output color component from the corresponding surface equation and a position of the pixel in the output image.

Abstract: A method of calculating a color blend for an output image. In a fundamental form, the method comprises the steps of specifying a plurality of non-coincident points (x.sub.1, y.sub.1, z.sub.1), (x.sub.2, y.sub.2, z.sub.2), (x.sub.3, y.sub.3, z.sub.3) on a surface (1) to be subject to color blending. Next, a surface equation (Intensity=Ax+By+C) is derived for each color component from a pixel value (z.sub.1,z.sub.2,z.sub.3) of each of the points. The method then determines each output color component from the corresponding surface equation and a position of the pixel in the output image.

Abstract: A graphical object rendering system for the rendering of images from edge based object descriptions, into pixel based data, is disclosed. The system includes a buffer that comprises an array of storage structures to hold pixel level data, with each pixel level data structure further comprising a series of setable level indicators. A separate storage arrangement is provided for storing an edged based object description of the desired image, each edge of each object having a level value corresponding to the level of the object. An edge insertion means is connected to the buffer and the storage arrangement, to alter the setable level indicators at locations corresponding to any intersection of the edges with the pixels.

Abstract: Method and apparatus for producing an image made up of pixels is disclosed. The apparatus in one broad configuration comprises a scanning means for inputting scanned data, storage means connected to the scanning means for storing outlines of objects and associated object color data, at least one of the objects having associated object color data defined to be a component of corresponding the input scanned data, and rendering means, connected to the storage means and the scanning means, for rendering the objects utilising the associated object color data.