Greg Smith questions

Greg Smith answers

• gpgpu: Why dont we need branch prediction in fine grain multi-threading?
• Do GPU architectures have Persistent Last-Level Cache Across Kernel Launches?
• Difference on creating a CUDA context
• GPGPU: Consequence of having a common PC in a warp
• How is cudaMemset implemented?
• Which is faster for CUDA shared-mem atomics - warp locality or anti-locality?
• How to use L2 Cache in CUDA
• Is register overflowing a possible cause of a CUDA_EXCEPTION_5, Warp Out-Of-Range Address error?
• NVIDIA Parallel Nsight Vs Visual Profiler
• Interpreting compute workload analysis in Nsight Compute