Intel AI software, powered by oneAPI, delivers high performance across a variety of devices. See a demo of text-to-image generative AI running on a CPU and a GPU with the same TensorFlow code.
AI workloads require high performance from development to deployment, often across a variety of compute environments. Intel AI software, powered by oneAPI, gets the most performance out of your Intel CPUs and GPUs. And it’s supported across frameworks such as PyTorch and TensorFlow.
This short demo shows Stable Diffusion, a compute-intensive text-to-image generate AI model, running on a 4th Gen Intel® Xeon® Scalable processor and an Intel® Flex Series GPU with no code changes required to switch devices.
This example runs in TensorFlow* using Intel® Extension for TensorFlow. oneAPI libraries enable this example to also run in PyTorch* with similar performance.