Abstract: A system and method for implementing multi-level resolution conversion (MLRC) of a set of image pixels efficiently using linear interpolation. The system automatically generates equations for determining the values of output pixels in a real-time manner as a function of the input pixels and a desired resolution conversion. As the input pixel values are received, the MLRC system generates the pixel values in real-time at a rate required for the output resolution without requiring a large number of intermediate complex calculations to be stored in memory. The pixel values utilized in the equations are normalized and scaled to values within a predetermined normalization range in order to minimize the complexity of the calculations performed by the MLRC system.

Abstract: A status queue in a host and a control queue in a segmentation and reassembly (SAR) subsystem are on opposite sides of a host bus in a control plane. Buffer descriptors in the host and the SAR and buffers in the host are in a data plane. To transfer cell payloads to a first line interfacing the SAR, the host writes the SAR that it has such cell payloads. The host writes the host buffer descriptors into the control queue to obtain the transfer of the buffer payload to the first line. The SAR writes the status queue when the transfer has been completed. To transfer cell payloads to the host memory, the host writes into the control queue the address of the buffers to receive the payload from the SAR. The SAR then writes the buffer descriptors to the status queue to obtain the transfer of the cell payloads to the buffers. Each of the control and status queues may be respectively considered to constitute two (2) control queues and two (2) status queues.

Abstract: A system and method for providing sound in a computer are disclosed. An audio module for controlling digitized sound I/O is included in a media stream controller. The media stream controller may also coordinate graphics and video which allows multiple media subsystems to be supported from a single bus device. A software application may initiate sound data transfer by sending a conventional DMA mode command to the media stream controller. The media stream controller activates an audio interrupt service routine which processes the request without using a conventional DMA controller. Digital sound data is transferred across a local bus using high speed burst mode block transfer commands and is buffered by the media stream controller in a display memory. Concurrently, the media stream controller may output sound data from the display memory to a sound output device using a double buffering method.

Abstract: A header and a payload in a cell are separated for transfer between a cell interface and a host memory. The header is transferred to a control memory. For transfer to the host memory, the control memory initially provides a host-memory region address and the region length. The payload is recorded in such region. The control memory also provides a second host-memory region address, and length, when the payload length exceeds the payload length in the first address region. For transfer from the host memory to the cell interface, the control memory provides a host memory region address and the header combines the header and the payload and passes the combination to the cell interface. Cells from different sources (i.e. terminals) are scheduled at table positions dependent upon their individual transfer rates. The cells at the scheduled positions are normally transferred in time slots corresponding to such positions.

Abstract: Multimedia information (e.g. graphics, video, sound, control information) passes through a system bus from a CPU main memory to a display memory in accordance with CPU commands. The information may be packetized with associated packet types identifying the different media. A media stream controller processes the information and passes the processed information to the display memory. Controllers in the media stream controller individually pass multimedia information to the display memory. A PACDAC controller in the media stream controller causes media (e.g. graphics, video) in the display memory to be transferred to a PACDAC for display. The format, sequence, and rate of this transfer may be flexibly controlled by software on a frame by frame basis. Arbitration logic establishes priorities for the different controllers in the media stream controller so they may share a single bus for accessing the display memory. A single interrupt controller coordinates interrupts (e.g.

Abstract: A system and method for generating NTSC and PAL formatted composite video signals in a computer. Digital pixel data may be processed by software to form component video pixel data, which may include luminance and chrominance components. A chrominance look-up table is provided in a display memory and is used for modulation of the chrominance components in accordance with NTSC and/or PAL formats. The modulated components are then combined to form digital composite video pixel data which may be stored in a frame buffer in the display memory. Alternatively, a pixel clock frequency equal to an integer multiple of the carrier frequency may be used and modulation in accordance with the NTSC and PAL formats may be accomplished by inverting, setting to zero or leaving unmodified the chrominance components. The architecture of this system greatly reduces hardware complexity and bandwidth requirements.

Abstract: Aspects of the present invention provide a system for selectively processing a video signal in accordance with instructions from application software. The system contains a video decoder for converting an analog video signal to digital video data, and a controller for formatting and routing the digital video data. A list of control structures may be loaded into a memory associated with the controller. The control structures contain formatting and routing information used by the controller to process different portions of the video stream. The number and content of control structures as well as the correlation between the control structures and selected portions of the video stream may be flexibly determined by application software. In particular, the control structures may be configured such that certain portions of the video stream are routed to the CPU for processing, while other portions are routed to a display driver and output on a display device.

Abstract: A system and method for generating composite video signals in a computer. Digital pixel data may be processed by software to form component video pixel data, which may include luminance and chrominance components. A chrominance look-up table is provided in a display memory and is used for modulation of the chrominance components. The modulated components are then combined to form digital composite video pixel data which may be stored in a frame buffer in the display memory. Video control information is precalculated and stored in the display memory in advance. The digital composite video pixel data and video control information are then recovered from the display memory to produce a formatted stream of video data. The architecture of this system greatly reduces hardware complexity and bandwidth requirements. In addition, the process may be controlled by a media stream controller which is also adapted for audio and graphics processing.

Abstract: An inverter receives an energizing voltage with a particular magnitude (e.g. 1.5 V) at a first terminal and produces the voltage at a second terminal. The received and produced voltages are differentially introduced to a stage which produces a single-ended bias voltage (e.g. 3 V) related to the second terminal voltage. The bias voltage is servoed to regulate the second terminal voltage (e.g. 1.5 V) and the bias voltage (e.g. 3 V) regardless of energizing voltage variations. In response to the bias voltage and a variable input voltage, a pass transistor in a buffer produces a first control voltage different from the bias voltage by the pass transistor threshold voltage for input voltages greater than the bias voltage less the pass transistor threshold. The first control voltage corresponds to the input voltage for input voltages less than the bias voltage less the pass transistor threshold.

Abstract: The magnitudes of an input voltage and individual ones of progressive fractions of a reference voltage are compared to produce first and second output voltages. Each of the elements in a first logical network receives the first output voltage from an individual one of the comparators and the second output voltage from a comparator non-consecutive with (preferably 2 comparators removed from) such individual comparator. Signals from these elements pass to latches. The latches have assertion and negation outputs which pass to elements in a second logical network. When an individual one of the elements in the second logical network provides a particular output, it prevents the elements receiving outputs from comparators responsive to lower reference voltage fractions from providing the particular output.

Abstract: A system and method for generating composite video signals in a computer. Digital pixel data may be processed by software to form component video pixel data, which may include luminance and chrominance components. A chrominance look-up table is provided in a display memory and is used for modulation of the chrominance components. The modulated components are then combined to form digital composite video pixel data which may be stored in a frame buffer in the display memory. Video control information is precalculated and stored in the display memory in advance. The digital composite video pixel data and video control information are then recovered from the display memory to produce a formatted stream of video data. The architecture of this system greatly reduces hardware complexity and bandwidth requirements. In addition, the process may be controlled by a media stream controller which is also adapted for audio and graphics processing.

Abstract: Analog signals representing individual digital values (.+-.1, .+-.3) pass through a telephone line to a receiver. These signals may be first provided in a pseudo random sequence. A linear echo canceller and a first adder eliminate, to an extent, echo signals resulting from second analog signals transmitted on the same telephone line by the receiver. A non-linear echo canceller and a second adder further reduce the echo signals and specifically reduce non-linear components in the echo signals. Adjustable signal delays achieve optimal performance of the linear and non-linear echo cancellers. An equalizer containing four (4) different modules then compensates for signal distortions introduced by the telephone line and minimizes the effect of noise in the telephone line. The equalizer modules are a digital gain control element, a feed forward digital filter and two (2) feedback digital filters. A detector module produces in one of several different ways at the receiver an estimate of the digital data (.+-.1, .+-.

Abstract: A header and a payload in a cell are separated for transfer between a cell interface and a host memory. The header is transferred to a control memory. For transfer to the host memory, the control memory initially provides a host-memory region address and the region length. The payload is recorded in such region. The control memory also provides a second host-memory region address, and length, when the payload length exceeds the payload length in the first address region. For transfer from the host memory to the cell interface, the control memory provides a host memory region address and the header combines the header and the payload and passes the combination to the cell interface. Cells from different sources (i.e. terminals) are scheduled at table positions dependent upon their individual transfer rates. The cells at the scheduled positions are normally transferred in time slots corresponding to such positions.

Abstract: The noise from the effects of currents through distributed capacitances between electrical circuitry on an integrated circuit chip and the chip substrate is minimized, especially for analog circuitry mixed on the chip with digital circuitry. The invention separates a plurality of bits in each digital word into a plurality (e.g. 2) of segments. A first register off the chip latches the first bits in each word with a first clock signal. A second register off the chip latches the second bits in each word with a second clock signal delayed from the first clock signal. The first register bits are latched on the chip with the first clock signal by a third register. The delayed second register bits are latched on the chip by a fourth register with the second clock signal or with a delayed first clock signal having the same delay as the second clock signal. Substrate ties for the third and fourth registers may be connected to at least one, preferably a plurality, of bonding pads on the chip.

Abstract: A multiplexer, preferably on an integrated circuit chip, receives a plurality of buses each having a plurality of lines responsive to binary indications and passes the binary indications in the lines of a particular one of the buses. The multiplexer includes a plurality of circuit blocks each responsive to the binary indications in the lines of an individual one of the buses. Each block has a plurality of recursive circuits each having first and second stages. The second stages of the recursive circuits in an individual one of the circuit blocks receive an individual one of a plurality of control indications at a first side of the block to activate the first stages in such recursive circuits. The first stage in each recursive circuit in each individual circuit block receives at a second side of the block the binary indications in an individual line in an individual one of the buses to obtain a signal from such first stage in accordance with such binary indication upon the activation of such first stage.

Abstract: A system and method for providing sound in a computer are disclosed. An audio module for controlling digitized sound I/O is included in a media stream controller. The media stream controller may also coordinate graphics and video which allows multiple media subsystems to be supported from a single bus device. A software application may initiate sound data transfer by sending a conventional DMA mode command to the media stream controller. The media stream controller activates an audio interrupt service routine which processes the request without using a conventional DMA controller. Digital sound data is transferred across a local bus using high speed burst mode block transfer commands and is buffered by the media stream controller in a display memory. Concurrently, the media stream controller may output sound data from the display memory to a sound output device using a double buffering method.

Abstract: An apparatus and a method are provided to obtain oscillations from a crystal at a particular frequency by introducing real and imaginary components of voltage to the crystal. The imaginary component of voltage is different from the real component of voltage by a particular phase angle such as 90.degree.. The voltage introduced to the crystal is processed to produce a first current having characteristics corresponding to such voltage and to produce a second current having characteristics related to the imaginary component of such voltage. The first and second currents are combined to produce a first current corresponding to the real component of the voltage introduced to the crystal. This current is shifted through a phase angle of 90.degree. to produce a second current corresponding to the imaginary component of the voltage introduced to the crystal. The first current is converted to a first voltage which is regulated to provide a particular gain.

Abstract: A system for processing a stream of data and automatically selecting a portion or all of the data stream for block writing to a memory. The memory is capable of storing data in response to a block write command and a normal write command. The system contains a first data register and a second data register having the same data width. The first data register accepts data from the data stream in accordance with the its data width. Data stored in the first data register is transferred to the second data register. The first data register is then loaded with a portion of the data stream which is contiguous to the data stored in the first data register prior to the transferring. The data in the first and the second data registers is then compared. If the data in the first and the second registers is the same, then the content of a data counter is increased by one. When the content of the data counter exceeds a predetermined value, the system executes a block write command.

Abstract: A CPU introduces software commands to a first limited capacity memory (e.g. FIFO), on an integrated circuit chip. Data (e.g. graphics) from a first portion of a second memory (off chip) is processed in accordance with such commands. A second portion (e.g. FIFO) of the second memory may also store commands normally passing from the CPU through the first memory. When the first memory becomes full, the commands may pass from the CPU through the second portion of the second memory (which may have a storage capacity considerably greater than that of the first memory) and then through the first memory. The commands may continue to flow in this auxiliary path until the second portion of the second memory becomes empty. A third memory of a limited capacity on the chip may pass the commands from the CPU to the first memory in the normal operation or to the second portion of the second memory when the first memory becomes full. The CPU may also pass commands to other peripheral equipment while a ready line is high.

Abstract: Words of different types of digital information, including standard interframe video (SIF), graphics, television and audio are transferred preferably in packets between a controller, storage memory and shift registers (e.g. FIFO's) individually associated with the different information types. For a VRAM memory, information is transferred in parallel, controlled by tag bus information, from the controller to the memory and then serially to the FIFO's, all at a frequency higher than a clock frequency in a monitor raster scan. The tag bus information is decoded and introduced to an additional FIFO. A state machine processes such additional FIFO information and transfers the digital information to the different FIFO's at times controlled in each line by such additional FIFO--e.g. particular times in each line for the SIF and graphics and, thereafter, for television and audio, at times unrelated to any times in such line.