Skip to main content
. 2017 Dec 26;16(1):95–115. doi: 10.1007/s12021-017-9352-y

Fig. 2.

Fig. 2

Workflow of concurrent list-mode (LM) processing. The LM data is divided into data chunks and processed by 32 CUDA streams at any given time. On processing completion by any stream, a new LM data chunk is read from the disk and processed asynchronously until all LM data is read and processed