Nvidia’s journey to forge the one CPU SKU to rule them all
A deeper look at Vera
Nvidia's Ian Finder, who oversees the development of the company's datacenter CPUs, believes that most use cases can be served by a single, well-balanced CPU SKU. Unlike other vendors, Nvidia has avoided offering a wide range of CPUs with varying core counts, frequencies, power efficiencies, and other specifications. Instead, the company aims to create a CPU that maintains low latencies and high throughput even when all 88 cores are loaded.
This approach involves making compromises, such as a slightly less powerful vector unit, but it allows Nvidia to offer a CPU that provides the same performance per core as some frequency-optimized x86 parts while also delivering the same throughput per watt as some density-optimized parts. Nvidia's Vera CPU, which is being offered in standalone CPU racks, storage, and context memory systems, is designed to be a versatile chip for various workloads, including high-performance computing (HPC) tasks.
The 88-core CPU runs at 3.3 GHz, and in some cases, Nvidia's processor offers functionality equivalent to 176 narrower cores when SMT is enabled, providing flexibility for different performance and workload requirements.
Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
- Nvidia’s journey to forge the one CPU SKU to rule them all theregister.com