PPT-CS 179: GPU Programming Lecture 7 Last Week Memory optimizations using different GPU

CS 179 GPU Programming Lecture 7 Last Week Memory optimizations using different GPU caches Atomic operations Synchronization with syncthreads Week 3 Advanced GPUaccelerable algorithms Reductions to parallelize problems that dont seem intuitively parallelizable. Avg Access Time 2 Tokens Number of Controllers Average Access Time clock cyles brPage 16br Number of Tokens vs Avg Access Time 9 Controllers Number of Tokens Average Access Time clock cycles brPage 17br brPage 18br m pm am pm am pm am pm am pm am pm am pm am pm am pm am pm am pm am pm Meal 1 Meal 2 Meal 3 Meal 4 Meal 5 Meal 6 NOTES brPage 3br The Training for LIFE Experience Daily Progress Report ACTUAL Upper Body Muscle Groups Chest Shoulders Back Triceps Bice Dr A . Sahu. Dept of Comp Sc & . Engg. . . IIT . Guwahati. 1. Outline. Graphics System . GPU Architecture. Memory Model. Vertex Buffer, Texture buffer. GPU Programming Model. DirectX. , OpenGL, . Lecture 2: more basics. Recap. Can use GPU to solve highly parallelizable problems. Straightforward extension to C++. Separate CUDA code into .cu and .. cuh. files and compile with . nvcc. to create object files (.o files). Cache overview. 4 Hierarchy questions. More on Locality. Please bring these slides to the next lecture!. Projects 2 and 3. Regrade. issues for 3. Please resubmit and come to office hours with a diff.. Host-Device Data Transfer. 1. Moving data is slow. So far we’ve only considered performance when the data is already on the GPU. This neglects the slowest part of GPU programming: getting data on and off of GPU. Lecture 5: GPU Compute . Architecture. 1. Last time.... GPU Memory System. Different kinds of memory pools, caches, . etc. Different optimization techniques. 2. Warp Schedulers. Warp schedulers find a warp that is ready to execute its next instruction and available execution cores and then start execution. Linchuan. . Chen. Advisor: Dr. . Gagan. Agrawal. Motivation - Platforms. Accelerators are Evolving. General Purpose Graphics Processing Units (GPGPU). Extreme-scale, cost-effective, power efficient. Lecture 7. Last Week. Memory optimizations using different GPU caches. Atomic operations. Synchronization with __. syncthreads. (). Week 3. Advanced GPU-accelerable algorithms. “Reductions” to parallelize problems that don’t seem intuitively parallelizable. T. Chen, T. Moreau, Z. Jiang, L. Zheng, S. Jiao, E. Yan, H. Shen, M. Cowan, L. Wang, Y. Hu, L. . Ceze. , C. . Guestrin. , and A. Krishnamurthy . Presentation by Grzegorz . Nuno Lopes . and. José Monteiro. Deriving preconditions by hand is hard; WPs are often non-trivial. WPs derived by hand are often wrong!. Weaker preconditions expose more optimization opportunities. Scientific Computing and Visualization. Boston . University. GPU Programming. GPU – graphics processing unit. Originally designed as a graphics processor. Nvidia's. GeForce 256 (1999) – first GPU. with Application-Transparent Support . for Multiple Page Sizes. Rachata. . Ausavarungniru. n. ,. . Joshua Landgraf, Vance Miller. Saugata. . Ghose. , . Jayneel. Gandhi, Christopher J. . Rossbach. Recap. Some algorithms are “less obviously parallelizable”:. Reduction. Sorts. FFT (and certain recursive algorithms). Parallel FFT structure (radix-2). Bit-reversed access. http://staff.ustc.edu.cn/~csli/graduate/algorithms/book6/chap32.htm.