What is Blackwell?
NVIDIA's 2024–2025 GPU architecture generation. Features 5th-generation Tensor Cores, GDDR7 memory, and significant AI inference performance improvements over Ada Lovelace (RTX 40 series).
Full Explanation
Blackwell is NVIDIA's GPU microarchitecture codename for the RTX 50 series, succeeding Ada Lovelace (RTX 40 series). Key advances for AI workloads include 5th-generation Tensor Cores with FP4/FP8 native support, GDDR7 memory delivering 40% more bandwidth, and an improved NVLink fabric for multi-GPU setups. The consumer flagship RTX 5090 packs 32 GB GDDR7 at 1.8 TB/s — roughly 3× the bandwidth of the RTX 4090. The RTX 5070 brings GDDR7 bandwidth to the mid-range at ~$600.
Why It Matters for Local AI
Blackwell is the first architecture where mid-range consumer cards ($600–800) provide meaningful LLM acceleration. The RTX 5070's 672 GB/s bandwidth rivals what the RTX 4090 offered at $1,600 just two years earlier. For anyone buying a GPU for local AI in 2025–2026, Blackwell is the minimum bar worth considering.
Hardware Relevant to Blackwell
gpu · Check Price on Amazon · 12 GB VRAM · 672 GB/s
gpu · Check Price on Amazon · 12 GB VRAM · 672 GB/s
Related Terms
Tensor Cores→
Specialized hardware units on NVIDIA GPUs designed for matrix multiplication — the core math operation in neural networks. 5th-gen Tensor Cores (Blackwell) are significantly faster than 4th-gen (Ada Lovelace) for AI inference.
GDDR7→
The latest generation of GPU memory (2024+). Significantly higher bandwidth than GDDR6X at the same capacity tier. Used in NVIDIA Blackwell cards (RTX 5070 series).
VRAM→
Video RAM — dedicated memory on a GPU. Determines the maximum model size you can run with full GPU acceleration. Once a model exceeds VRAM, it spills to system RAM over the slow PCIe bus.
CUDA→
NVIDIA's proprietary parallel computing platform. Industry standard for AI/ML. Nearly every AI framework (PyTorch, Ollama, ComfyUI) supports CUDA natively and first.