A bioinformatics workload requires maximum single-threaded CPU performance for a genomic sequencing algorithm, with jobs that run 2–4 hours. Which EC2 instance family fits best?
Show answer and explanation
Answer: B
Compute-optimized C-family instances exist exactly for CPU-bound jobs, and C7g adds Graviton3 cores, which deliver strong per-core performance and very attractive performance per dollar for repeatable batch-style runs in the 2–4 hour range.
GPU instances help parallel matrix work, not single-threaded code, and memory-optimized instances pay for memory capacity the algorithm does not need.
Why the other options are wrong
- A. M-family general-purpose instances strike a balance between CPU and memory, so they cost more per core and give up throughput for memory the genomic workload does not use.
- C. R-family instances are built around large memory capacity — the profile of in-memory databases and caches, not a CPU-bound single-threaded algorithm.
- D. P4d GPUs accelerate massively parallel matrix operations like deep learning training; a single-threaded sequencing loop gains almost nothing from them.