Ada Lovelace (microarchitecture)
The Ada Lovelace microarchitecture is a graphics processing unit (GPU) microarchitecture developed by Nvidia as the successor to Ampere. It is named after Ada Lovelace, an English mathematician and writer, considered by some to be the first computer programmer.
Ada Lovelace features several key improvements over its predecessor. These include, but are not limited to:
-
New Streaming Multiprocessor (SM) Architecture: This is the fundamental building block for executing CUDA cores, tensor cores, and ray tracing cores. Ada Lovelace introduces advancements in SM architecture aimed at improving performance and efficiency.
-
Next-Generation Ray Tracing Cores: These cores are specialized hardware units dedicated to accelerating ray tracing calculations, which are essential for realistic lighting and shadows in games and other applications. Ada Lovelace introduces improvements to these cores to further enhance ray tracing performance.
-
Next-Generation Tensor Cores: These cores are specialized hardware units designed to accelerate matrix multiplication operations, which are fundamental to deep learning and artificial intelligence. Ada Lovelace features enhanced Tensor Cores for increased AI performance.
-
Shader Execution Reordering (SER): A technology that dynamically reorganizes the order of shader execution to improve efficiency and reduce latency.
-
DLSS 3 (Deep Learning Super Sampling): A new version of Nvidia's AI-powered upscaling technology, which generates entirely new frames rather than just pixels, promising significant performance improvements in games.
Ada Lovelace-based GPUs are used in a variety of products, including high-end consumer graphics cards for gaming, professional workstation GPUs for content creation and scientific applications, and data center GPUs for AI and machine learning workloads. The microarchitecture aims to provide significant performance increases and improved efficiency compared to previous generations, particularly in ray tracing and AI-related tasks.