University:
Boston University
Course:
MET CS 201 | Introduction to Programming
Academic year:

2023
Views:

414

Pages:

13
Author:

Gironetzbk3n

Atomic Operations Collaborative (inter-block) Global Memory Update: Read Increment Write Atomic Operations Collaborative (inter-block) Global Memory Update: Read Increment Interruptible! Write Atomic Operations Collaborative (inter-block) Global Memory Update: Read Increment Interruptible! Write Interruptible! Atomic Operations Collaborative (inter-block) Global Memory Update: Read Increment Interruptible! Write Interruptible! Atomic Global Memory Update: Read Increment Write Atomic Operations Collaborative (inter-block) Global Memory Update: Read Increment Interruptible! Write Interruptible! Atomic Global Memory Update: Read Increment Protected Write Atomic Operations Collaborative (inter-block) Global Memory Update: Read Increment Interruptible! Write Interruptible! Atomic Global Memory Update: Read Increment Protected Write Protected Atomic Operations Collaborative (inter-block) Global Memory Update: Read Increment Interruptible! Write Interruptible! Atomic Global Memory Update: Read Increment Protected Write Protected How? atomic {add,inc,cmpxchg,. . . }(int *global, int value); Outline 1 The OpenCL Runtime 2 Device Language Synchronization Extensions 3 OpenCL implementations Extensions: Future-proof CL Similar extensions mechanism to OpenGL. Two mechanisms: Runtime: cl ext.h header availability checkable via #ifdef device.extensions Device language: #pragma OPENCL EXTENSION name : enable Important extension: cl khr fp64 Vendor and ‘official’ extensions. Outline 1 The OpenCL Runtime 2 Device Language 3 OpenCL implementations The Nvidia CL implementation Targets only GPUs Notes: Nearly identical to CUDA No native C-level JIT in CUDA (→ PyCUDA) Page-locked memory: Use CL MEM ALLOC HOST PTR. (Careful: double meaning) No linear memory texturing CUDA device emulation mode deprecated → Use AMD CPU CL (faster, too!) The Apple CL implementation Targets CPUs and GPUs General notes: Different header name OpenCL/cl.h instead of CL/cl.h Use -framework OpenCL for C access. Beware of imperfect compiler cache implementation (ignores include files) CPU notes: One work item per processor GPU similar to hardware vendor implementation. (New: Intel w/ Sandy Bridge) The AMD CL implementation Targets CPUs and GPUs (from both AMD and Nvidia) GPU notes: Wide SIMD groups (64) Native 4/5-wide vectors But: very flop-heavy machine, may ignore vectors for memory-bound workloads → Both implicit and explicit SIMD CPU notes: Many work items per processor (emulated) cl amd printf