Verified65% confidenceFactExact time
In GPU parallel reduction, because floating-point operations do not satisfy the associative law, differences in the execution order of different warps cause deviations in numerical results, which is the root cause of non-determinism.
3
Sources
65%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
Unverified浮点运算不满足结合律，不同线程束的执行顺序差异会导致数值结果出现细微偏差78% similarUnverifiedNon-determinism refers to the phenomenon where the same input may produce slightly different outputs across different runs, which is especially common in models that use GPU parallel computation, dropout layers, or stochastic sampling strategies.76% similarUnverifiedNon-deterministic GPU operations are designed to prioritize performance over determinism; forcing a fixed execution order introduces numerous synchronization barriers that significantly reduce parallel efficiency.76% similarUnverifiedDuring GPU porting, the choice between single precision (float) and double precision (double), the use of atomic operations, and the implementation of reduction operations all affect numerical consistency.72% similarUnverifiedGPU inference deployment faces an impossible triangle of cost, latency, and operational complexity, and it is not possible to optimize all three simultaneously.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/575340API
curl https://kongchang.com/api/v1/knowledge/claims/575340MCP
get_claim(id=575340)