Download presentation
Presentation is loading. Please wait.
Published byLuke Marsh Modified over 8 years ago
1
Matthew Royle Supervisor: Prof Shaun Bangay
4
How do we implement OpenCL for CPUs Differences in parallel architectures Is our CPU implementation efficient Is parallel faster than sequential Is GPU execution faster than CPU
6
Devices and Context Buffers Kernel Execution Model
12
Devices and Context Buffers Kernel Execution Model
20
int (a[1024]) 2 int a[1024] int (a[1024]) 2
22
We can implement OpenCL for CPUs OpenCL design doesn’t take advantage of CPU architecture Executing kernels on CPUs not efficient Parallel execution faster than sequential GPUs more suited to parallel processing
23
Heterogeneous Parallel Processing Language Design tailored for GPUs Implement for CPUs Variance in parallel execution times Parallel execution faster than sequential NVIDIA’s Implementation for GPUs faster than our implementation for CPUs
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.