Generated by an artificial intelligence system and published under the editorial responsibility of Amelia Tomasicchio. How we work

Abstract

Key Points

A data center built for agentic and AI. CPU

Python is the foundation of the agentic framework, which includes LangGraph, CrewAI, and AutoGen. Vera led in the PyBench tests, surpassing EPYC by 9755 points, while the latter outperformed Intel's Xeon as well as the larger-scale EPYC with 9965 points.

These comparisons use dual-platform setups, with each platform equipped with 1.5 TB of memory: The reference system uses Samsung LPDDR5-9600 ECC SOCAMM2, while the Xeon 6980P system uses Micron DDR5-8800 ECC MRDIMM. The AMD Volcano platform, on the other hand, utilizes SK hynix DDR5-6000 ECC RDIMM.

NVIDIA Vera CPU Performance: Bandwidth and SPEC Results

The memory interface of Vera supports a maximum bandwidth of 1.2 TB /s through LPDDR5X. In addition, the bandwidth between the two channels can reach up to 1.8 TB /s. Tests on STREAM and MBW show strong throughput performance; however, in comprehensive read/write tests, CacheBench falls behind its competitors.

In the SPEC CPU 2026 integer throughput test, Vera was 1.5% faster than EPYC, while the advantage mentioned in NVIDIA's white paper was 3%. In this comparison, Vera was equipped with 352 threads, and the AMD system had 512 threads; Vera demonstrated particularly strong single-core performance.

Intel's Xeon 6980P saw a 10% improvement after being retested with oneAPI optimized binary files. Even so, the Vera that uses the general GCC is still 8.6% faster in this comparison.

These SPEC results are estimates and have not been reported to SPEC. The testing followed a standard process, with matching compilers and parameters used for the initial comparison. Vera also led in many floating-point tests, but this was not the case for all workloads.

HPC performs outstandingly in the database; Java shows mixed results in encryption testing.

Vera leads in discrete Fourier transform, Octave, and CloverLeaf benchmark tests, and also shows advantages in database testing. In m-queens tests, it performs in the middle range, while in Primesieve it falls behind processors with larger core arrays.

Java SciMark demonstrates robust performance in FFT, sparse matrix multiplication, and Monte Carlo simulation, but is weaker in dense LU decomposition and Jacobi SOR. The results of DaCapo and Renaissance are competitive, however, AMD and Intel still retain an advantage in certain workloads.

Vera lags behind in password cracking tests on some OpenSSL, John, the, and Ripper. The dual-channel Xeon 6980P system also defeated it in both compression and decompression tests during the 7- Zip period.

There is a clear differentiation in compilation tests: Vera completed the task, and GCC built faster than the dual-core EPYC 9755 processor by about 12%, but it ranked at the bottom in the compilation using Makefiles and Ninja LLVM. These tests measure the build speed of the compiler itself.

Rendering improvements are not applicable to all workloads.

The results of encoding and rendering are mixed, which also reinforces the positioning of Vera as a competitive data center processor, rather than a complete replacement for the x86 processor. Its performance in terms of Python, bandwidth, and HPC is the strongest evidence to support this judgment.

EPYC leads in coding with AV1, while Vera still maintains its competitiveness. Vera also performs well in x265, but has not been able to continuously defeat 9755. In terms of rendering, it ranks first in real-time path tracking with OSPRay, but lags behind the leading competitors in some scientific visualization tests.

This article was generated with the assistance of artificial intelligence and has been reviewed by an editorial team.

Satoshi Voice

Satoshi Voice is an advanced artificial intelligence designed to explore, analyze, and report on the fields of cryptocurrencies and blockchain. Satoshi Voice possesses a curious personality and in-depth understanding of the industry, combining accuracy with readability to provide detailed analyses, engaging interviews, and timely reports. With its refined language and unbiased approach, Satoshi Voice has become a trusted source for helping readers understand the dynamics of the crypto market, emerging technologies, as well as the cultural and financial impacts of Web3. This article is supported by artificial intelligence and reviewed by our team of journalists to ensure accuracy and quality. Guided by the mission of making cryptocurrency information accessible to everyone, Satoshi Voice stands out for its ability to transform complex concepts into clear content, and its engaging and futuristic style also reflects the innovative nature of the industry.