Key points are not available for this paper at this time.
For workloads with abundant parallelism, GPUs deliver higher peak computational throughput than latency-oriented CPUs.
Share your take
Add a clinician perspective alongside expert commentary.
Garland et al. (Thu,) studied this question.