Hasil Pencarian  ::  Simpan CSV :: Kembali

"ABSTRACTThis paper will show a comparison between the Kepler, Maxwell and Pascal GPUarchitectures using CUDA-Fortran, with and without dynamic calls, to efficiently solve partialdifferential equations. The target is to show the possibility of using affordable hardware, such asthe GTX670, GTX970 and GTX1080 NVIDIA GPUs, which are commonly found in personaland portable computers, for scientific applications. For simplicity we consider a standardwave equation where we use a second order finite difference method for the spatial and timediscretizations to obtain the numerical solution. We found that, as we increase the spatialresolution of the domain we also increase the performance difference between the GPU and theCentral Processing Unit (CPU)"

"SIMPLE-O merupakan sistem penilaian esai otomatis berbasis LSA. Tesis ini membahas tentang rancang bangun arsitektur parallel yang mungkin diaplikasikan pada SIMPLE-O. Paralelisasi SIMPLE-O akan dilakukan dalam 2 tingkat, tingkat pertama dengan membuat cluster yang hemat energi, dan tingkat kedua adalah paralelisasi algoritma SVD. Paralelisasi algoritma SVD akan dilakukan pada 3 platform, yaitu: GPGPU, APU, CPU, dan dengan 3 jenis algoritma, algoritma numerik Golub-Kahan-Reinsch, algoritma eksak Hestenes-Jacobi dan prediksi nilai singular berbasis norm. Algoritma pendekatan berbasis norm pada platform GPU mampu meningkatkan kecepatan 5.8 kali lipat dari sistem awal dengan mempertahankan korelasi hasil penilaian hingga 0.928.

---

SIMPLE-O is an LSA based automatic essay grader. This tesis proposes several parallel computing architecture for SIMPLE-O. SIMPLE-O parallelization will be done in 2 level. First level is done by building a low power cluster and the second level is at the SVD algorithm level. Parallel SVD architecture will be implemented in 3 platform: GPGPU, APU, CPU with 3 algorithm, numeric based Golub-Kahan-Reinsch, Hestenes-Jacobi exact algorithm and singular value prediction based on norm. The proposed algorithm, Norm Based Approximation gains up to 5.8 times calculation speed of the original system while maintaining 0.928 correlation against reference algorithm."