
The full-stack NVIDIA accelerated computing platform has once again demonstrated exceptional performance in the latest MLPerf Training v4.0 benchmarks.
NVIDIA more than tripled the performance on the large language model (LLM) benchmark, based on GPT-3 175B, compared to the record-setting NVIDIA submission made last year. Using an AI supercomputer featuring 11,616 NVIDIA H100 Tensor Core GPUs connected with NVIDIA Quantum-2 InfiniBand networking, NVIDIA achieved this remarkable feat through larger scale - more than triple that of the 3,584 H100 GPU submission a year ago - and extensive full-stack engineering.
Thanks to the scalability of the NVIDIA AI platform, Eos can now train massive AI models like GPT-3 175B even faster, and this great AI performance translates into significant business opportunities. For example, in NVIDIA's recent earnings call, we described how LLM service providers can turn a single dollar invested into seven dollars in just four years running the Llama 3 70B model on NVIDIA HGX H200 servers. This return assumes an LLM service provider serving Llama 3 70B at $0.60/M tokens, with an HGX H200 server throughput of 24,000 tokens/second.
NVIDIA H200 GPU Supercharges Generative AI and HPC The NVIDIA H200 Tensor GPU builds upon the strength of the Hopper architecture, with 141GB of HBM3 memory and over 40% more memory bandwidth compared to the H100 GPU. Pushing the boundaries of what's possible in AI training, the NVIDIA H200 Tensor Core GPU extended the H100's performance by up to 47% in its MLPerf Training debut.
NVIDIA Software Drives Unmatched Performance Gains Additionally, our submissions using a 512 H100 GPU configuration are now up to 27% faster compared to just one year ago due to numerous optimizations to the NVIDIA software stack. This improvement highlights how continuous software enhancements can significantly boost performance, even with the same hardware.
This work also delivered nearly perfect scaling. As the number of GPUs increased by 3.2x - going from 3,584 H100 GPUs last year to 11,616 H100 GPUs with this submission - so did the delivered performance.
Learn more about these optimizations on the NVIDIA Technical Blog.
Excelling at LLM Fine-Tuning As enterprises seek to customize pretrained large language models, LLM fine-tuning is becoming a key industry workload. MLPerf introduced a new LLM fine-tuning benchmark this round, based on the popular low-rank adaptation (LoRA) technique applied to Meta Llama 2 70B.
The NVIDIA platform excelled at this task, scaling from eight to 1,024 GPUs, with the largest-scale NVIDIA submission completing the benchmark in a record 1.5 minutes.
Accelerating Stable Diffusion and GNN Training NVIDIA also accelerated Stable Diffusion v2 training performance by up to 80% at the same system scales submitted last round. These advances reflect numerous enhancements to the NVIDIA software stack, showcasing how software and hardware improvements go hand-in-hand to deliver top-tier performance.
On the new graph neural network (GNN) test based on R-GAT, the NVIDIA platform with H100 GPUs excelled at both small and large scales. The H200 delivered a 47% boost on single-node GNN training compared to the H100. This showcases the powerful performance and high efficiency of NVIDIA GPUs, which make them ideal for a wide range of AI applications.
Broad Ecosystem Support Reflecting the breadth of the NVIDIA AI ecosystem, 10 NVIDIA partners submitted results, including ASUS, Dell Technologies, Fujitsu, GIGABYTE, Hewlett Packard Enterprise, Lenovo, Oracle, Quanta Cloud Technology, Supermicro and Sustainable Metal Cloud. This broad participation, and their own impressive benchmark results, underscores the widespread adoption and trust in NVIDIA's AI platform across the industry.
MLCommons' ongoing work to bring benchmarking best practices to AI computing is vital. By enabling peer-reviewed comparisons of AI and HPC platforms, and keeping pace with the rapid changes that characterize AI computing, MLCommons provides companies everywhere with crucial data that can help guide important purchasing decisions.
And with the NVIDIA Blackwell platform, next-level AI performance on trillion-parameter generative AI models for both training and inference is coming soon.
North America Stories
27/05/2026
Telestream has announced that its Board of Directors has appointed Benjamin Desbois as Chief Executive Officer, effective July 1, 2026. Desbois, currently Teles...
27/05/2026
ESPN garnered 10 awards; NBC's Sunday Night Football received the Outstandin...
27/05/2026
Matrox Video is celebrating its 50th anniversary, marking five decades of operations from its headquarters in Montreal, Canada. Founded in 1976, the company has...
27/05/2026
Major League Baseball has announced a series of initiatives tied to America's Semiquincentennial, including a national marketing campaign, Fourth of July br...
27/05/2026
Advanced Systems Group (ASG) has announced that Brian Gross has joined the company as an Account Manager on its Audio team, based in the Burbank office. He will...
27/05/2026
Nielsen has released new research on soccer fandom ahead of the FIFA World Cup 2...
27/05/2026
ESL FACEIT Group (EFG) has unveiled a new partnership with TikTok to bring broad...
27/05/2026
FIFA's Oscar Sanchez gives a deeper look to how this tournament will be cove...
27/05/2026
The soon-to-be senior from Charlottesville is building her skills in replay, TD, and even creative content for HokieVision and its ACC Network productions
In t...
27/05/2026
FOX Sports' Mike Davies breaks down the vision for this summer's showcas...
27/05/2026
HBS's Paul King, FIFA's Oscar Sanchez preview how the masses at home wil...
27/05/2026
FOX's MLB coverage dominated the night at the 47th Annual Sports Emmy Awards...
27/05/2026
One of the most memorable Postseasons in baseball history would have had no memo...
27/05/2026
NBC's Sunday Night Football is among the most decorated and most watched programs in the history of television. It added to its jam-packed trophy case on Tu...
27/05/2026
The 2026 Sports Emmys marked a watershed moment for Prime Video Sports. After bu...
27/05/2026
With the Opening Match just over two weeks away, the entire sports-production-te...
27/05/2026
The XL Converge 300P radio system emerges with a groundbreaking feature set enhancing the mission-critical communications of public safety, federal and critica...
27/05/2026
Pairing Two47 MCX software with existing LTE networks means tailored system upgrades that can save time, money and lives....
27/05/2026
PAC-3 MSE offers improved range, speed, and maneuverability, making it an effect...
27/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
27/05/2026
Star Trek VFX: Recreating John Knoll's Iconic Warp Stars without a Slitscan ...
27/05/2026
Adventure World Uses Blackmagic Replay for Marine Live
Brie Clayton May 27, 2026
0 Comments
Large screen displays and slow motion replays dynamically ...
27/05/2026
Berklee Alumna and Assistant Professor Olivia P rez-Collellmir to Premiere Origi...
27/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
27/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
27/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
27/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
27/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
27/05/2026
Co-founder Dan Castles to transition to Executive Chair; internal promotion reinforces continuity and long-term growth
Telestream, a global leader in media wor...
27/05/2026
Big Blue Marble today announced that its Nakolos platform is the first end-to-end 5G Broadcast solution worldwide to implement the complete feature set introduc...
27/05/2026
Lightware recently hosted the Girls' Day event in April at its headquarters in Budapest, welcoming students for an interactive introduction to engineering a...
27/05/2026
May 27th, 2026 May 27, 2026
Press Materials Available Here
TRIBECA X Announc...
26/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
26/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
26/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
26/05/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
26/05/2026
Cobalt Digital to Showcase End-to-End IPMX Ecosystem at InfoComm 2026, Making ST 2110 Easy for Pro AV
blueCORE standalone processors headline solutions designe...
26/05/2026
Matrox Video today celebrates its 50th anniversary, marking five decades of innovation, engineering excellence, and customer-focused evolution from its headquar...
26/05/2026
ITV Studios, the production arm of the UKs largest commercial broadcaster, has deployed the Cuez live production platform to unify the management of three back-...
26/05/2026
CETA Software releases Morpheus, an AI tool for real-time post-production projec...
26/05/2026
In a move set to redefine motorsports coverage across the Asia Pacific region, Ikegami Electronics announces that Two Wheels Motor Racing Sdn Bhd (TWMR), a lead...
26/05/2026
Back to All News
Netflix Releases Official Trailer and Poster for The Root Of T...
26/05/2026
Back to All News
MED, Production Begins on Netflixs First Medical Drama From Br...
26/05/2026
Back to All News
Netflix Announces Five New Brazilian Productions and Expands I...
26/05/2026
Back to All News
Netflix Celebrates the Release of the New Animated Series Due ...
26/05/2026
Back to All News
Made In New Mexico: Building The Boroughs' From the Ground Up
A photo from The Boroughs.' (Courtesy of Netflix 2026)
Entertainme...
26/05/2026
Smart Production Control. Total Confidence.
Tyngsborough, MA, May 27, 2026 - Broadcast Pix today announced the ONix Pro Control Panel, its most advanced hard...
26/05/2026
The shift to agentic AI creates a new CPU requirement for the AI factory: fast cores, massive memory bandwidth and the ability to sustain high performance when ...
25/05/2026
The former University of Wyoming wrestler is essential in helping the rapidly growing streamer delivers more than 50,000 live events per year
The sports-produc...
25/05/2026
Image courtesy of GA-ASI...