Parametric flows: Automated behavior equivalencing for symbolic analysis of races in CUDA programs

Abstract

Abstract-The growing scale of concurrency requires automated abstraction techniques to cut down the effort in concurrent system analysis. In this paper, we show that the high degree of behavioral symmetry present in GPU programs allows CUDA race detection to be dramatically simplified through abstraction. Our abstraction techniques is one of automatically creating parametric flows-control-flow equivalence classes of threads that diverge in the same manner-and checking for data races only across a pair of threads per parametric flow. We have implemented this approach as an extension of our recently proposed GKLEE symbolic analysis framework and show that all our previous results are dramatically improved in that (i) the parametric flowbased analysis takes far less time, and (ii) because of the much higher scalability of the analysis, we can detect even more data race situations that were previously missed by GKLEE because it was forced to downscale examples to limit analysis complexity. Moreover, the parametric flow-based analysis is applicable to other programs with SPMD models.

Efficient search for inputs causing high floating-point errors

“…First, we assume the absence of data races (checking for races using tools such as [30] is future work). Second, we have manually verified that there are no schedule-dependent reduction operations (automation of these checks is future work).…”

Formal Analysis of GPU Programs with Atomics via Conflict-Directed Delay-Bounding

“…In addition to detecting subtle concurrency bugs, CD proves to be a light-weight and tailorable approximation to more complete (but also more expensive) algorithms such as DPOR. Our future work will include exploiting thread symmetry [18] and informing concolic verification through static analysis.…”

“…In particular, one such technique that enables scalability to a large number of threads is described in our recent work [18].…”

The Design and Implementation of a Verification Technique for GPU Kernels

“…This allows for precise defect checking, but limits scalability. A recent extension to GKLEE by Li et al [2012a] uses the notion of parametric flows to soundly restrict defect checking to consider only certain pairs of threads. This is similar to the two-thread abstraction employed by GPUVerify and PUG, and leads to scalability improvements over standard GKLEE, at the expense of a loss in precision for kernels that exhibit inter-thread communication (GPUVerify and PUG also suffer from this loss in precision, as discussed under Incompleteness of the Two-Thread Reduction in Section 4.1).…”