By Akhil Bagaria on 01 April 2014

BFS finds use in state space searching, graph partitioning, automatic theorem proving, etc., and is one of the most used graph operation in practical graph algorithms. The BFS problem is, given an undirected, unweighted graph G(V,E) and a source vertex S, find the minimum number of edges needed to reach every vertex V in G from source vertex S. The optimal sequential solution for this problem takes O(V +E) time. CUDA implementation of BFS. We solve the BFS problem using level synchronization. BFS traverses the graph in levels; once a level is visited it is not visited again. The BFS frontier corresponds to all the nodes being processed at the current level. We do not maintain a queue for each vertex during our BFS execution because it will incur additional overheads of maintaining new array indices and changing the grid configuration at every level of kernel execution. This slows down the speed of execution on the CUDA model. For our implementation we give one thread to every vertex. Two boolean arrays, frontier and visited, Fa and Xa respectively, of size |V| are created which store the BFS frontier and the visited vertices. Another integer array, cost, Ca, stores the minimal number of edges of each vertex from the source vertex S. In each iteration, each vertex looks at its entry in the frontier array Fa. If true, it fetches its cost from the cost array Ca and updates all the costs of its neighbors if more than its own cost plus one using the edge list Ea. The vertex removes its own entry from the frontier array Fa and adds to the visited array Xa. It also adds its neighbors to the frontier array if the neighbor is not already visited. This process is repeated until the frontier is empty. This algorithm needs iterations of order of the diameter of the graph G(V,E) in the worst case. Algorithm 1 runs on the CPU while algorithm 2 runs on the 8800 GTX GPU. The while loop in line 5 of Algorithm 1 terminates when all the levels of the graph are traversed and frontier array is empty

Algorithm 1. CUDA BFS (Graph G(V,E), Source Vertex S) 1: Create vertex array Va from all vertices and edge Array Ea from all edges in G(V,E), 2: Create frontier array Fa, visited array Xa and cost array Ca of size V. 3: Initialize Fa, Xa to false and Ca to ∞ 4: Fa[S] ←true, Ca[S]←0 5: while Fa not Empty do 6: for each vertex V in parallel do 7: Invoke CUDA BFS KERNEL(Va,Ea,Fa,Xa,Ca) on the grid. 8: end for 9: end while

AES

Given the flexibility of the memory model, it is possible to efficiently use the four T-lookup tables each one containing 256 entries of 32 bits each. Unlike previous GPU hardware, G80 is a scalar processor, so there is no need to combine instructions in vector operations, in order to get the full processing power. Furthermore, the native availability of the 32 bit logical XOR operation on G80 speeds up by orders of magnitude the execution of that fundamental operation of the cryptographic theory. Thus, a single AES round on a state can be done with 16 table look-ups and 16 32-bit exclusive-or operations. First of all, the input data and the expanded key are stored in the GPU global memory space. As the final results show, moving the data to and back from the device memory may become the slowest operation when doing cryptography on the GPU. It is due to the bandwidth of the PCExpress interface which is only about 3,2 GB/s compared to the 50 GB/s of the onboard memory of the GeForce 8800 graphics card. The precomputed T-boxes are pre-loaded in the specific constant memory of the device. The constant memory is cached and the look-up table fully matches the size of the cache. The input data is then divided in chunks of 1024 bytes which are encrypted or decrypted completely in parallel. One CUDA block of threads is responsible for computing one chunk of the input. The block is composed of 256 GPU threads. An exhaustive research has indicated that size to lead the fastest implementation. When running the kernel code on the device, performance improvement can be observed as the number of blocks increases. So bigger input sizes lead to better performance.

Every GPU thread of the block computes four output bytes in each AES round, so four threads encrypt/decrypt the whole input state. Threads from the same block need to share and to frequently access information of the AES expanded key. For this reason it is loaded in shared memory together with the portion of input processed by that block. More precisely, two arrays of 1KB shared memory are used for the input, reading data from the first and saving results of each AES round to the second one. Then the arrays are swapped for the successive round. This strategy allows to complete the encryption of the input chunk without exiting the kernel. So it is done without using the CPU to manage an external for loop to launch sequentially all the AES rounds. At the end of the computation, the resulted output data is written again in the global device memory and then returned to the CPU.

CP

Computes the short-range component of Coulombic potential at each grid point over a 3D grid containing point charges representing an explicit-water biomolecular model.

MMU

MUMmerGPU builds multiple suffix trees of the reference and partitions the query sequences into sets, called QueryBlocks, depending on the memory available on the GPU. Sequences within a given QueryBlock are aligned in parallel on the GPU.

The algotithm used in MMU is shown here:

http://www.biomedcentral.com/1471-2105/8/474/figure/F4



blog comments powered by Disqus