Contents:
---
|-- `histogram`: Parallel histogram implementation.
|
|-- `infoli`: A parallelized version of C code that implements a model of a biological
| neural network (the inferior-olivary nucleus). The C code was written by someone
| else, and the challenge was to make it run faster by parallelizing it using CUDA.
|
|-- `parallel-prefix-sum`: Optimized version of the parallel prefix sum algorithm,
| which uses balanced trees using a hierarchical approach, and has O(N) runtime.
| While some other implementations such as Kogge-Stone run in O(N*logN), this version
| is work-efficient (which, should be noted, doesn't mean it's necessarily fastest).
|
|-- `sgemm-tiled`: Optimized SGEMM kernel, using tiling. Makes use of common techniques
| such as coalesced access and using shared memory to minimize global memory reads.
|
|-- `sgemm`: Single-Precision General Matrix Multiplication (SGEMM) kernel.
|
|-- `sum-reduction`: Parallel sum reduction, i.e. computes the sum of an array of values.