TPC GPU Decoding

Filter TPC clusters at decoding time (timebin cutoff) [work in progress]

• Makes the final #clusters unknown
• Requires small redesign of buffer allocations in the workflow
• If parameter for cutoff is not set, normal workflow
• If it is set, allocate tmp buffer and tmp ClusterNativeAccess
  1. Decode all clusters in tmp buffer
  2. Scan all clusters, count how many remains after filtering (done in GPU kernel)
  3. Allocate final buffer with correct size
  4. Move the filtered clusters to the final buffer (done in GPU kernel)

New projects

Familiar with simulation script of FST, will start to simulate datasets with different beams and interaction rates