Example of MultiGPU timeline: link
####Workload / Data:
#####Human Reference Files: | Reference | GEM FM-index | Myers Binary Reference |
#####Candidates of 1Million queries (GEM output: .prof)
Synthetic data (1M queries): | 100 | 200 | 400 | 600 | 800 | 1000 |
Real data queries: | Illumina | Roche 454 | Pacbio | ION Torrent |
#####Example: Input (candidates) and Ouput (scores) Files: | Candidates: 1000 Bases - 100K Queries - 11 Candidates per Query | Results |
#####Compile make gcandidates-GEM-1st make sample-0
#####Generate input files head -100000 1000.prof > 1000.100K.api.prof ./gcandidates-GEM-1st ../../data/1000.100K.api.prof
#####Run Set the visible GPUs:
export CUDA_VISIBLE_DEVICES=0,1,2
Run Myers in 3 GPUs and 6 buffers:
./TEST 0.2 ../../data/HG.4bits.ref ../../data/1000.100K.api.prof.100000.11.0.gem.qry ../../data/1000.100K.api.prof.100000.11.0.gem.qry ../../data/1000.100K.api.prof.100000.11.0.gem.qry ../../data/1000.100K.api.prof.100000.11.0.gem.qry ../../data/1000.100K.api.prof.100000.11.0.gem.qry ../../data/1000.100K.api.prof.100000.11.0.gem.qry