Generative AI is introducing a new era of computing for productivity, content creation, gaming and more.
NVIDIA NeMo and DLSS 3 Frame Generation, Meta LLaMa, ChatGPT, Adobe Firefly, Stable Diffusion and other generative AI models and apps use neural networks to discover structures and patterns in existing data to cultivate fresh and original material, according to a May 28 blog post.
“AI will be the single largest driver of innovation for Windows customers in the coming years,” Pavan Davuluri, corporate vice president of Windows silicon and system integration at Microsoft, said in the release. “By working in concert with NVIDIA on hardware and software optimizations, we’re equipping developers with a transformative, high-performance, easy-to-deploy experience.”
These generative AI models can operate up to five times faster on GeForce RTX and NVIDIA RTX GPUs than on rival hardware when they are optimized for these GPUs, which provide up to 1,400 Tensor TFLOPS for AI inferencing, the release reported. Tensor Cores, specialized hardware in RTX GPUs designed to expedite AI calculations, as well as ongoing software updates, are responsible for this.
Performance for generative AI models, such Stable Diffusion, that benefit from new DirectML optimizations has doubled thanks to improvements unveiled last week at the Microsoft Build conference, according to the release.
As more AI inference occurs on local devices, PCs will require strong, efficient hardware to handle these challenging jobs, the release reported. RTX GPUs will include Max-Q low-power inferencing for AI workloads to address this need.
For simpler inferencing jobs, the GPU will use a tiny fraction of the power, while scaling up to unparalleled levels of performance for complex generative AI applications, according to the release. Developers now have access to a full RTX-accelerated AI development stack running on Windows 11, making it simpler to build, train and use sophisticated AI models, starting with the creation and improvement of models using enhanced deep learning frameworks made available by the Windows Subsystem for Linux.
