AN 831: Intel® FPGA SDK for OpenCL™: Host Pipelined Multithread
ID
683013
Date
11/20/2017
Public
1.4.3.2. Processor Affinity in GZIP
In a high-throughput design, Processor Affinity enables assigning each thread loop to a specific core.
As mentioned earlier, Huffman code generation, kernel execution, and metadata generation, all execute a thread loop that reads or consumes data object from an input queue and delivers the processed data or job item to an output queue that acts as an input queue for the next stage in the pipeline. Along with these thread loops, the frequency counter in Huffman code generation uses up to three threads internally to obtain high-throughput. These threads are also assigned to specific cores for optimal performance, as mentioned earlier.
Related Information