The discovery of novel magnetic quantum materials and topologically protected spin textures is opening new opportunities for energy-efficient information technologies, spin-based computing, and quantum functionalities. Realizing these opportunities requires predictive multiscale modeling that connects electronic structure, magnetic interactions, finite-temperature behavior, and non-equilibrium spin dynamics across vastly different length and time scales. We request allocations on both CPU- and GPU-based computing systems to advance a coordinated research program focused on emergent magnetism in low-dimensional materials, topological spin textures, and complex energy landscapes governing magnetic phase transformations. Our workflow combines state-of-the-art electronic structure methods with large-scale atomistic spin simulations. First-principles calculations using Quantum ESPRESSO, VASP, and the in-house relativistic electronic structure code RSPt will be employed to determine magnetic exchange interactions, magnetic anisotropies, spin-orbit effects, electronic correlations, and spectroscopic signatures in a broad class of magnetic materials. These calculations require substantial CPU resources owing to large supercells, dense Brillouin-zone sampling, relativistic treatments, and beyond-DFT methodologies needed to capture correlation effects and finite-temperature electronic structure. The resulting Hamiltonians will serve as input for extensive atomistic spin simulations on GPU systems. Recent work from our group has demonstrated how accelerator-native approaches enable simulations at previously inaccessible scales. We have developed SpinX, a GPU-native atomistic spin simulation framework capable of deterministic and stochastic spin dynamics, Monte Carlo sampling, static optimization, spectroscopy calculations, and large-scale transition-state searches. By exploiting tensor-convolution formulations of magnetic interactions and heterogeneous accelerator architectures, SpinX achieves throughput exceeding ten billion spin-site operations per second and enables simulations containing more than one billion atomic spins on a single node. During the proposed allocation period, we will use these capabilities to investigate several outstanding problems in contemporary magnetism. These include twist-controlled altermagnetic magnons and spin transport in van der Waals heterostructures, finite-temperature magnetic phase transitions in correlated layered magnets, and the stability and dynamics of two- and three-dimensional topological spin textures such as hopfions, skyrmions, antiskyrmions, and related chiral excitations. Particular emphasis will be placed on identifying transition mechanisms and activation barriers governing creation, annihilation, transformation, and transport processes. To achieve this, we will employ recently developed GPU-based transition-state and energy-landscape exploration algorithms that efficiently map complex networks of metastable states and saddle points without requiring prior knowledge of reaction pathways. The combination of electronic structure calculations and large-scale spin simulations provides a predictive route from quantum-mechanical interactions to experimentally observable magnetic phenomena. CPU resources are essential for generating accurate first-principles Hamiltonians and spectroscopic observables, while GPU resources are required for large-scale finite-temperature simulations, long-time spin dynamics, and transition-state searches involving millions to billions of degrees of freedom. Together, these computational capabilities will enable the discovery of new magnetic functionalities, provide fundamental understanding of emergent magnetic phenomena, and establish scalable computational methodologies for next-generation spintronic and quantum materials research.