dGPU

A discrete GPU, or dGPU, is a separate graphics processing unit deployed as a distinct graphics card or module, rather than being integrated into the CPU or motherboard. It carries its own video memory (VRAM) and GPU die, allowing independent operation from system RAM. dGPUs connect to the host system over PCI Express and typically require additional power via PCIe power connectors. This separation enables higher performance, greater memory bandwidth, and more specialized hardware units than integrated graphics solutions.

In daily use, dGPUs provide higher frame rates and better rendering quality for gaming, professional visualization, and compute-heavy tasks. They feature many shader units, high memory bandwidth, and often dedicated features such as ray tracing cores and AI accelerators. Common VRAM types include GDDR6, GDDR6X, and HBM2; memory size ranges widely by target market. Software support includes graphics APIs such as DirectX, OpenGL, and Vulkan, and compute frameworks such as CUDA (NVIDIA), OpenCL, and ROCm (AMD).

Market segments include consumer gaming cards (NVIDIA GeForce RTX, AMD Radeon RX), professional/workstation GPUs (NVIDIA RTX A-series, AMD Radeon Pro), and data-center or cloud GPUs. Form factors include PCIe desktop cards and mobile discrete GPUs in laptops, with varying power envelopes from roughly 100 watts to above 350 watts for high-end models. They require adequate cooling and a capable power supply. The choice between a dGPU and an iGPU depends on performance needs, power budgets, and software requirements; some systems use both for different workloads.