The　dual-camera　compressive　hyperspectral　imager　(DCCHI)　can　capture　3D　hyperspectral　image　(HSI)　with　a　single　snapshot.　However,　due　to　the　high　computation　complexity　of　reconstruction　methods,　DCCHI　cannot　apply　to　the　time-crucial　applications.　In　this　paper,　we　propose　a　GPU　assisted　towards　real-time　reconstruction　framework　for　DCCHI.　First,　leveraging　the　fast　convergence　rate　of　the　alternative　direction　multiplier　method,　we　propose　a　reformative　reconstruction　algorithm　which　can　achieve　a　fast　convergence　rate.　Then,　using　the　interpolation　results　of　a　low　resolution　reconstructed　HSI　as　the　warm　start,　we　propose　a　fast　reconstruction　strategy　to　further　reduce　the　computation　burden.　Last,　a　GPU　parallel　implementation　is　presented　to　achieve　nearly　real-time　reconstruction.　Evaluation　experiments　indicate　our　framework　can　obtain　a　significant　promotion　in　reconstruction　efficiency　with　a　slight　accuracy　loss.　©　Springer　Nature　Switzerland　AG　2018.