An integrated GPU-oriented computational framework is introduced for large-scale, interface-resolved simulations of multiphase turbulence, combining a high-performance flow solver with runtime droplet analysis capabilities. The core solver, MHIT36 [1], performs direct numerical simulation (DNS) of the incompressible Navier–Stokes equations coupled with a phase-field method to capture interfacial dynamics. Simulations are conducted in a triply periodic domain representative of homogeneous isotropic turbulence. Transport of the phase-field variable is described using the Accurate Conservative Diffuse Interface (ACDI) [2] formulation. From a computational perspective, MHIT36 is designed for modern GPU-accelerated architectures. The code employs a two-dimensional domain decomposition with MPI parallelism, using the cuDecomp library [3] for pencil transpositions and halo exchanges, and cuFFT together with OpenACC directives to offload compute-intensive kernels to GPUs. This strategy delivers excellent scalability while retaining a modular and extensible code structure. Building on this solver, we introduce GALENE36, a fully GPU-accelerated and parallel module for runtime droplet identification, tracking, and event detection in diffuse-interface simulations. Droplet identification relies on connected- component labeling (CCL) applied to a thresholded phase field, a well-established approach in image analysis but still challenging to deploy efficiently in large-scale, three-dimensional, distributed GPU simulations. GALENE36 implements a GPU-optimized CCL strategy tailored to structured Cartesian grids and domain-decomposed data, minimizing synchronization and communication overhead while preserving scalability. Temporal tracking is performed using a voxel-overlap criterion between labeled droplets at consecutive time steps, enabling robust association of parent and child structures. This overlap-based approach allows reliable detection and classification of breakup and coalescence events, while naturally accounting for the diffuse-interface representation inherent to phase-field methods. Together, MHIT36 and GALENE36 form an integrated computational application that couples high-fidelity multiphase DNS with native, runtime droplet statistics, avoiding costly post-processing and reducing the storage demands.