The concurrency architecture identifies the threads of execution, the allocation of software elements to those threads, specifies how the threads will be scheduled, and defines how resources will be shared among them.
Also, we propose and evaluate "copy-on-thread", an alternative solution that enables the performance to scale up to 25.5× without relying on specialized libraries nor requiring specific data allocation and thread scheduling.
In a real chip, the allocation of threads to cores is "a complex interplay of latencies and scheduling policies," Alistarh says.
Before the process times are measured, the allocation of threads to blocks should be considered in order to maximize parallelism.
Wide Single Instruction, Multiple Thread architectures often require static allocation of thread groups, executed in lockstep.
A fast, wire-efficient synchronization technique is developed that supports dynamic allocation of multiple threads on shared-memory, message-passing, and/or single-chip multiprocessors.
Because of this uncertainty, we decided to allocate various numbers of threads to blocks to find an allocation that minimizes the process time.
With this approach, each application knows exactly how many cores it owns at any given point in time; this allows it to tailor its internal concurrency level and thread placement for its current core allocation.
The computational results showed that the proposed method significantly improved Smith-Waterman algorithm on CUDA-enabled GPUs in proper allocation of block and thread numbers.
The results showed that the proposed method significantly improves Smith-Waterman algorithm on CUDA-enabled GPUs in proper allocation of block and thread numbers.
