NVIDIA-Genomics-Research / GenomeWorks

SDK for GPU accelerated genome assembly and analysis
https://clara-parabricks.github.io/GenomeWorks/
Apache License 2.0
284 stars 76 forks source link

[cudapoa] make maximum number of edges per node configurable #607

Open r-mafi opened 3 years ago

r-mafi commented 3 years ago

CUDAPOA_MAX_NODE_EDGES and CUDAPOA_MAX_NODE_ALIGNMENTS with default value of 50 defined in cudapoa_structs.cuh have a big impact on memory usage per POA. Consequently maximum number of POAs residing on GPU can increase by using smaller values for these two parameters. Changing these macros to template parameters allows achieving higher parallelism, particularly for processing long reads. As a rule of thumb, the value for these parameters should not exceed maximum number of reads per window.