5-reduce-sum

规约（Reduction）

实验环境

• GPU: NVIDIA GeForce RTX 3080
• CPU: Intel(R) Xeon(R) Silver 4210R CPU @ 2.40GHz x 40
• CUDA/NVCC: 11.7
• OS: Ubuntu 20.04
• Host Compiler: g++ (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
  • Compiler Options: -O3 -std=c++17

性能数据

版本	耗时（us）	加速比
reduce_cpu	22,461	1
reduce_V1	2,127	10.6
reduce_V2	1,261	17.8
reduce_V3	1,203	18.7
reduce_V4	646	34.8
reduce_V5	431	52.1
reduce_V6	391	57.4
reduce_V7	250	89.9

算法说明

TODO

参考

• Harris, Mark. "Optimizing parallel reduction in CUDA." Nvidia developer technology 2.4 (2007): 70.
• NVIDIA/cuda-samples: Samples/2_Concepts_and_Techniques/reduction/reduction.cpp