Bolt is a C++ template library optimized for GPUs. Bolt provides high-performance library implementations for common algorithms such as scan, reduce, transform, and sort.
Other
373
stars
65
forks
source link
Bolt_1.2: bolt:cl::sort is hanging for higher odd buffer sizes with 1000 iterations. #222
If we run bolt::cl::sort for 1000 iterations by having higher odd buffer sizes like 2 power 23, 2 power 25,.. for double and float data type it is hanging. No issues with non powers of 2 and even buffer sizes like 2 power 24, 2 power 26..
If we run bolt::cl::sort for 1000 iterations by having higher odd buffer sizes like 2 power 23, 2 power 25,.. for double and float data type it is hanging. No issues with non powers of 2 and even buffer sizes like 2 power 24, 2 power 26..