This code is implementation of algorithm described in the following paper:
Yan-Hao Chen, Fei Hua, C. Huang, Jeremy Bierema, Chi Zhang, and E. Z. Zhang. 2019. Accelerating Concurrent Heap on GPUs.ArXivabs/1906.06504 (2019). https://arxiv.org/abs/1906.06504
This was my project for the CSE-560 course on GPU computing offered by IIIT-D under the guidance of Dr. Ojaswa Sharma.
How to run:
make in src/ directory and it will compile, link and run the test file automatically.
Excellent speedups were obtained in comparison to CPU heap implementation and speedups are as high as 130x when compared STL-heap on inserted approximately 130 million keys in the heap.
Time graph on varying number of keys inserted:
Speedup graph on varying number of keys inserted. Speedup is with respect to my naive CPU implementation of heap:
Speedup graph on varying number of keys inserted. Speedup is with respect to STL-HEAP:
