Nvidia boosts Next-Gen CPU and GPU design with new Vera CPUs

Nvidia’s silicon is speeding up the company’s future CPU/GPU designs

Nvidia has confirmed that it is now using its Vera CPUs in its engineering workflows. This means that Nvidia Vera silicon is being used to design future Nvidia CPUs and GPUs, with Nvidia noting 1.5x performance gains in key verifications and simulation workloads.

Silicon design is incredibly complex, requiring detailed simulations, formal verification, and lengthy retesting when changes are made. Chips can go through thousands of iterations before they are manufactured, making any speed boost for critical electronic design automation (EDA) applications incredibly valuable.

Nvidia tested its Vera CPUs using Cadence Jasper and Synopsys VCS. Both applications showed up to 1.5x higher performance on selected workloads. Nvidia is also working with both companies on application profiling, software optimisation, and system-level tuning to further boost their productivity.

Designing future CPU/GPU designs faster

Nvidia is rolling out its Vera CPUs for its EDA workflows. This will enable faster verification times. This allows Nvidia’s engineers to test more design alternatives, find problems faster, and design chips at an accelerated pace.

Nvidia’s Vera CPUs feature 88 custom ARM-based Olympus CPU cores and LDDR5X memory. Nvidia claims that its high-bandwidth, low-latency design and strong fabric make its Vera CPUs well suited to EDA applications.

Nvidia plans to use its own hardware to design and validate its future products. This gives Nvidia a strong incentive to produce increasingly powerful CPUs. After all, Nvidia’s new chips will directly impact their engineers and their daily workflows. It also means that Nvidia is not relying on external CPU providers for these vital workloads. Each Nvidia CPU generation will help build what comes next. That’s an exciting prospect.

You can join the discussion on Nvidia using its own chips to design its future products on the OC3D Forums.