Intel Corporation

Rahul Saxena

Munsi Haque

John William Phillips

Hitesh Rastogi

Ashwani Oberai

According to some embodiments, an external memory unit stores rows of macroblocks representing an image. A set of macroblocks may be transferred from the external memory unit to a local buffer, the set of macroblocks including fewer macroblocks than a row. A macroblock may then be decoded based on information in the local buffer to generate an image signal.

Local macroblock information buffer

A video decoder cache used for motion compensation data may be dynamically reconfigured. In some embodiments, it may be reconfigured on picture or frame boundaries and in other embodiments it can be reconfigured on sequence boundaries. The cache may be flushed on each boundary to enable such reconfiguration.

Dynamically configuring a video decoder cache for motion compensation

A switch and a process of operating a switch are described where a received data frame is copied one or more times into a memory before being transmitted out of the switch. The switch and method determine how much space in the memory is needed to store all of the copies of the received data frame and then the switch and method determine locations in the memory for storing the copies of the received data frames. The copies of the received data frame are stored until the ports designated as transmitting the copies of the received data frame are ready. When a port is ready, a copy of the received data frame is read out of the memory and the port is instructed where to locate the copy on a bus. When the port has retrieved the copy of the data frame, it transmits the data frame out of the switch.

Multicast and broadcast operations in ethernet switches

A method and system is provided for a multi-level memory. The system includes an internal memory and an external memory. Data packets are received through one or more input ports and initially stored in the internal memory. A control unit determines whether there is congestion of resources within the system and transfers data packets to external memory to ease the congestion. Data packets are eventually transferred from the internal or external memory to one or more output ports.

Method and system for a multi-level memory

A switch and a process of operating a switch are described where a received data frame is stored into memory in a systematic way. In other words, a location is selected in the memory to store the received data frame using a non-random method. By storing the received data frame in this way, switches that employ this system and method increase bandwidth by avoiding delays incurred in randomly guessing at vacant spaces in the memory. The received data frame is stored until a port that is to transmit the received data frame is available. Throughput is further improved by allowing the received data frames to be stored in either contiguous or non-contiguous memory locations.

Systematic memory location selection in Ethernet switches

The present invention is a method and apparatus for rounding a result operand of a floating-point (FP) operation which causes an underflow. The FP operation is recomputed using a truncate rounding mode to generate an underflowed operand. The underflowed operand is denormalized and providing characteristic bits. A rounding bit is generated based on the characteristic bits. The rounding bit is merged with the denormalized operand to generate the rounded result operand.

Optimized rounding in underflow handlers

The present invention is a method and apparatus for reporting exception in a single instruction multiple data (SIMD) processor in computing an arithmetic function for a plurality of argument data. The SIMD processor is configured for processing N elements simultaneously. A sequence of instructions is re-arranged to allocate the plurality of argument data in the N elements. The N elements are processed simultaneously. The exceptions for N elements are detected simultaneously. The detected exceptions are then combined to generate a global exception.

Growing community of inventors

Sunnyvale, CA, United States of America

Rahul Saxena