Ld

Abstract

Data race detection has become an important problem in GPU programming. Previous designs of CPU racechecking tools are mainly task parallel and incur high overhead on GPUs due to access instrumentation, especially when monitoring many thousands of threads routinely used by GPU programs. This article presents a novel data-parallel solution designed and optimized for the GPU architecture. It includes compiler support and a set of runtime techniques. It uses value-based checking, which detects the races reported in previous work, finds new races, and supports race-free deterministic GPU execution. More important, race checking is massively data parallel and does not introduce divergent branching or atomic synchronization. Its slowdown is less than 5× for over half of the tests and 10× on average, which is orders of magnitude more efficient than the cuda-memcheck tool by Nvidia and the methods that use fine-grained access instrumentation.

Checking Data-Race Freedom of GPU Kernels, Compositionally

“…Such techniques are better suited for highly detailed analysis in smaller kernels, and typically are unable to prove DRF. Dynamic data-race detection executes a kernel to find data-races on a fixed input, e.g., [14,18,19,28,32,38,39]. This technique only reports real data-races, but suffers from a slowdown of at least 10× compared to the non-instrumented program, and requires the kernel input data, which might be unavailable or unknown.…”

CL-Vis: Visualization Platform for Understanding and Checking the OpenCL Programs

“…LDetector is a static detector and uses a two-pass approach to detect write-write/read-write races [11]. LD statically detects a race with a two-pass detection algorithm that is atomic free and a memory-adaptive solution [12]. Oclgrind is a dynamic simulator, which can identify the race through monitoring the memory accesses [18].…”