Scalable method and system for streaming high-resolution media

A pipelined, multithreaded system for prioritizing movement of data, comprising:

• one or more central processing units (CPUs), each CPU executing one or more threads;
• a global clock for defining a current target image number;
• a plurality of media processors;
• and
• a shared memory;
• wherein the one or more threads include:
• a read controller thread that controls a plurality of reader threads, wherein each reader thread receives a piece of source image data and loads the piece into the shared memory, the piece carrying an associated image number, and
• a rendering controller thread that controls a plurality of renderer threads, each renderer thread receiving a piece with its associated image number from a corresponding reader thread, wherein each renderer thread downloads the piece to an appropriate media processor for synchronized output of the pieces, wherein synchronization and selection of the appropriate media processor is based on the associated image number and the current target image number defined by the global clock.

A pipelined, multithreaded method for prioritizing movement of data, comprising:

• (a) receiving source image data as one or more pieces, wherein for each piece a reader thread is generated that loads the piece into a shared memory and each piece is tagged with an associated image number;
• (b) processing the pieces on appropriate central processing units (CPUs), wherein one or more processing controller threads receive the pieces from corresponding reader threads, wherein each processing controller thread generates a plurality of processing threads that extract a plurality of sub-pieces for each piece, each sub-piece retaining the image number associated with the piece from which it was extracted, and wherein the processing threads move the sub-pieces to appropriate rendering nodes;
• (c) downloading the sub-pieces and their associated image numbers to corresponding media processors, wherein a plurality of renderer threads executing on the CPUs receive the sub-pieces from corresponding processing threads;
• (d) synchronizing the reader, processing, and renderer threads to a global clock that defines a current target image number:
• and
• (e) outputting synchronized sub-pieces as a composite image based on their associated image numbers and the current target image number defined by the global clock.

A pipelined, multithreaded method for prioritizing movement of data, comprising:

• (a) receiving source image data as a plurality of pieces, wherein for each piece a reader thread is generated that loads the piece into a shared memory and each piece is tagged with an associated image number;
• (b) moving the pieces to appropriate rendering nodes;
• (c) downloading the pieces to appropriate media processors based on their associated image numbers, wherein a plurality of renderer threads receive the pieces from corresponding reader threads;
• (d) synchronizing the reader and renderer threads to a global clock that defines a current target image number;
• and
• (e) outputting synchronized pieces as a composite image based on their associated image numbers and the current target image number defined by the global clock.

A pipelined, multithreaded method for prioritizing movement of data, comprising:

• (a) receiving source image data as one or more pieces, wherein for each piece a reader thread is generated that loads the piece into a shared memory and each piece is tagged with an associated image number;
• (b) processing the pieces on appropriate central processing units (CPUs), wherein one or more processing controller threads receive the pieces from corresponding reader threads, wherein each processing controller thread generates a plurality of processing threads that extract a plurality of sub-pieces for each piece, each sub-piece retaining the image number associated with the piece from which it was extracted, and wherein the processing threads move the sub-pieces to appropriate rendering nodes;
• (c) downloading a quantity of sub-pieces and their associated image numbers to an equal quantity of graphics processors, wherein a plurality of renderer threads receive the sub-pieces from corresponding processing threads, and wherein a total quantity of sub-pieces exceeds the quantity of graphics processors;
• (d) synchronizing the reader, processing, and renderer threads to a global clock that defines a current target image number;
• and
• (e) processing the sub-pieces on the graphics processors;
• and
• (f) repeating steps (c), (d) and (e) until the total quantity of sub-pieces is processed.