Nvidia speeds design; launches Ising
Nvidia said AI reduced a multi‑month GPU design task to an overnight process, while stressing that fully autonomous chip design remains distant. The company also introduced Ising, an open family of quantum AI models intended to help researchers build and control quantum processors. (tomshardware.com) (nvidianews.nvidia.com) (cio.com)
Nvidia says it is already using artificial intelligence to shrink one chip-design job from 10 months to a single overnight run, while adding new quantum-computing models to its software stack. (tomshardware.com) (nvidianews.nvidia.com) The chip-design example came from standard cell library porting, a step that adapts basic logic building blocks to a manufacturing process. Nvidia chief scientist Bill Dally said work that once took eight engineers 10 months can now run overnight on one graphics processing unit. (tomshardware.com) Dally said Nvidia is applying artificial intelligence across design exploration, verification and bug handling, but he also said the company is still “a long way” from software designing chips without human input. The current role is to speed engineering work and let teams test more options, not replace chip architects. (tomshardware.com) Quantum computing uses qubits, which are fragile data units that can drift out of tune or pick up errors from heat, noise and tiny hardware shifts. Two of the hardest jobs are calibration, which keeps the machine aligned, and error correction, which spots and fixes mistakes before they spread. (nvidianews.nvidia.com) (cio.com) Nvidia’s new Ising family targets those two jobs. The company said Ising Calibration is a vision-language model that can cut calibration work from days to hours, and Ising Decoding includes two 3D convolutional neural network models for real-time quantum error correction. (nvidianews.nvidia.com) (cio.com) Nvidia said the decoding models are up to 2.5 times faster and 3 times more accurate than traditional approaches on the company’s benchmarks. The models are open source, and Nvidia said they are meant for researchers and companies building quantum processors and hybrid quantum-classical systems. (nvidianews.nvidia.com) (cio.com) The two announcements fit the same strategy: Nvidia is pushing artificial intelligence deeper into the tools that design, run and stabilize advanced computing hardware. That extends the company’s reach beyond selling chips into the software layers that shape how future chips and quantum machines get built. (tomshardware.com) (nvidianews.nvidia.com) (cio.com) For now, Nvidia is describing both efforts as assistive systems. In chips, engineers still make the design calls; in quantum computing, researchers still face the larger problem of building machines stable enough to run useful applications at scale. (tomshardware.com) (nvidianews.nvidia.com)
