A GPU-accelerated extension alignment engine for long-read sequencing that delivers high speed with near-lossless accuracy.
make -j$(nproc) GPUARCH=sm_80Optional flags:
DPX=1— use the Hopper DPX kernel (requiresGPUARCH=sm_90).PRINT=1— verbose logging.RAW=1/KERNEL_ONLY=1— ablation builds.
./minimap2 -a -t $(nproc) --gpu --gpu-cfg gpu/gpu_config.json ref.mmi reads.fastq.gz > out.samMIT (see LICENSE.txt), inheriting from upstream minimap2.