
Large language models and the applications they power enable unprecedented opportunities for organizations to get deeper insights from their data reservoirs and to build entirely new classes of applications.
But with opportunities often come challenges.
Both on premises and in the cloud, applications that are expected to run in real time place significant demands on data center infrastructure to simultaneously deliver high throughput and low latency with one platform investment.
To drive continuous performance improvements and improve the return on infrastructure investments, NVIDIA regularly optimizes the state-of-the-art community models, including Meta's Llama, Google's Gemma, Microsoft's Phi and our own NVLM-D-72B, released just a few weeks ago.
Relentless Improvements Performance improvements let our customers and partners serve more complex models and reduce the needed infrastructure to host them. NVIDIA optimizes performance at every layer of the technology stack, including TensorRT-LLM, a purpose-built library to deliver state-of-the-art performance on the latest LLMs. With improvements to the open-source Llama 70B model, which delivers very high accuracy, we've already improved minimum latency performance by 3.5x in less than a year.
We're constantly improving our platform performance and regularly publish performance updates. Each week, improvements to NVIDIA software libraries are published, allowing customers to get more from the very same GPUs. For example, in just a few months' time, we've improved our low-latency Llama 70B performance by 3.5x.
NVIDIA has increased performance on the Llama 70B model by 3.5x. In the most recent round of MLPerf Inference 4.1, we made our first-ever submission with the Blackwell platform. It delivered 4x more performance than the previous generation.
This submission was also the first-ever MLPerf submission to use FP4 precision. Narrower precision formats, like FP4, reduces memory footprint and memory traffic, and also boost computational throughput. The process takes advantage of Blackwell's second-generation Transformer Engine, and with advanced quantization techniques that are part of TensorRT Model Optimizer, the Blackwell submission met the strict accuracy targets of the MLPerf benchmark.
Blackwell B200 delivers up to 4x more performance versus previous generation on MLPerf Inference v4.1's Llama 2 70B workload. Improvements in Blackwell haven't stopped the continued acceleration of Hopper. In the last year, Hopper performance has increased 3.4x in MLPerf on H100 thanks to regular software advancements. This means that NVIDIA's peak performance today, on Blackwell, is 10x faster than it was just one year ago on Hopper.
These results track progress on the MLPerf Inference Llama 2 70B Offline scenario over the past year. Our ongoing work is incorporated into TensorRT-LLM, a purpose-built library to accelerate LLMs that contain state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. TensorRT-LLM is built on top of the TensorRT Deep Learning Inference library and leverages much of TensorRT's deep learning optimizations with additional LLM-specific improvements.
Improving Llama in Leaps and Bounds More recently, we've continued optimizing variants of Meta's Llama models, including versions 3.1 and 3.2 as well as model sizes 70B and the biggest model, 405B. These optimizations include custom quantization recipes, as well as efficient use of parallelization techniques to more efficiently split the model across multiple GPUs, leveraging NVIDIA NVLink and NVSwitch interconnect technologies. Cutting-edge LLMs like Llama 3.1 405B are very demanding and require the combined performance of multiple state-of-the-art GPUs for fast responses.
Parallelism techniques require a hardware platform with a robust GPU-to-GPU interconnect fabric to get maximum performance and avoid communication bottlenecks. Each NVIDIA H200 Tensor Core GPU features fourth-generation NVLink, which provides a whopping 900GB/s of GPU-to-GPU bandwidth. Every eight-GPU HGX H200 platform also ships with four NVLink Switches, enabling every H200 GPU to communicate with any other H200 GPU at 900GB/s, simultaneously.
Many LLM deployments use parallelism over choosing to keep the workload on a single GPU, which can have compute bottlenecks. LLMs seek to balance low latency and high throughput, with the optimal parallelization technique depending on application requirements.
For instance, if lowest latency is the priority, tensor parallelism is critical, as the combined compute performance of multiple GPUs can be used to serve tokens to users more quickly. However, for use cases where peak throughput across all users is prioritized, pipeline parallelism can efficiently boost overall server throughput.
The table below shows that tensor parallelism can deliver over 5x more throughput in minimum latency scenarios, whereas pipeline parallelism brings 50% more performance for maximum throughput use cases.
For production deployments that seek to maximize throughput within a given latency budget, a platform needs to provide the ability to effectively combine both techniques like in TensorRT-LLM.
Read the technical blog on boosting Llama 3.1 405B throughput to learn more about these techniques.
Different scenarios have different requirements, and parallelism techniques bring optimal performance for each of these scenarios. The Virtuous Cycle Over the lifecycle of our architectures, we deliver significant performance gains from ongoing software tuning and optimization. These improvements translate into additional value for customers who train and deploy on our platforms. They're able to create more capable models and applications and deploy their existing models using less infrastructure, enhancing th
Most recent headlines
05/01/2027
Worlds first 802.15.4ab-UWB chip verified by Calterah and Rohde & Schwarz to be ...
07/10/2026
Dalet, a leading technology and service provider for media-rich organizations, today announced the latest Long-Term Supported (LTS) release of Dalet Flex. Build...
06/09/2026
June 9 2026, 23:00 (PDT) Dolby and MagentaTV Bring Fans Closer to the FIFA Worl...
04/08/2026
Dalet, a leading technology and service provider for media-rich organizations, t...
03/08/2026
SVG's annual TranSPORT conference returns to New York City for a full day of sessions focused on the current state of live-sports-transmission technology. S...
03/08/2026
A five-time Emmy winner, this esports vet has built a production career on live storytelling, audience connection, and meeting viewers wherever they are
The sp...
03/08/2026
Live production in the cloud is a growing trend across the industry and Major Le...
03/08/2026
(L-R) Nyah B., Nala B., Shannon Elizabeth S., Celia K., Olive G., and Nikki B. attend the Q&A for Cookie Queens by Alysa Nahmias, an official selection of the...
03/08/2026
Recreates a 1977 Hohner Clavinet D6
After launching a software recreation of their own electric piano, Rhodes set about capturing another classic instrument...
03/08/2026
A modern solution for an old technique
Peak Tamer is a new standalone application that automatically applies gain automation to keep rogue peaks under cont...
03/08/2026
Plug-in gains object metadata automation export
Sound Particles have just released an update that adds a new Metadata Automation function to their their rec...
03/08/2026
SGL Carbon and X-energy today announced a binding agreement to double SGL's production capacity for medium-grain isotropic graphite ( NBG-18 ), a specialize...
03/08/2026
Glensound will launch the Greater Divine Supernal at PLASA Show 2026, extending its intelligent loudspeaker range for professional studio monitoring and live au...
03/08/2026
How Agua Tinta's Class Project Launched Her Songwriting Career Two years after graduating from Berklee, the songwriter has a Latin Grammy nomination, a So...
03/08/2026
Leiden University's new Campus Spui, located in the heart of The Hague, the diplomatic and administrative centre of the Netherlands, opened in February 2026...
03/08/2026
Bitfocus, the specialist in media control and monitoring, has received a further boost to its plans for continuing growth, particularly in Europe and North Amer...
03/08/2026
Jason Flemyng, Tessa Peake-Jones, Lee Ingleby and Georgia Tennant join the cast ...
03/08/2026
Building on the momentum of a landmark year for soccer in North America, Leagues...
02/08/2026
Classic EQ design goes virtual
Following on from the launch of the 500-series version of their BAX EQ unit, Dangerous Music have now teamed up with modellin...
02/08/2026
The Government's Missing Dimension
This article was originally published on the National Preparedness Commission website on 25 June 2026.
----------------...
01/08/2026
Latest free Vaults instrument released
The Crow Hill Company's Vaults collection offers a continually rotating selection of virtual instruments and samp...
01/08/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
01/08/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
01/08/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
01/08/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
01/08/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
01/08/2026
COW Jobs: Lead Generation Representative - Animation Services
Brie Clayton July 31, 2026
0 Comments
Lead Generation Representative - Animation Service...
01/08/2026
Emilie Lowe Shoots Award Winning Short Film Where Dead Things Grow with Blackmag...
01/08/2026
Early Registration Opens for the 2026 SMPTE Media Technology Summit
Brie Clayton July 31, 2026
0 Comments
Initial Program Announced Alongside New Medi...
31/07/2026
With an emphasis on local, the league is looking beyond the initial four teams t...
31/07/2026
The 2026 Sundance Institute Directors Lab fellows in Estes Park, Colorado (Photo by Gabe Rovick)
Dear Friends,
This has been an extraordinary year for our ar...
31/07/2026
Offers 360-degree sound placement in any DAW
Audiocube have just announced the launch of a new plug-in that embeds a complete 3D acoustic environment inside...
31/07/2026
10 new Chinese C-Pop voices introduced
While Sonarworks are best known to many for their room-correction software, their product range also includes an inno...
31/07/2026
Reimagines the vintage BBD sound character
PSPaudioware have now officially launched the new delay plug-in that they were previewing at GearExpo UK. Said to...
31/07/2026
The Savannah Bananas has built its reputation on refusing to play baseball by traditional rules, and the team's production arm Banana Ball TV (BTV) has draf...
31/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
31/07/2026
SMPTE, the home for media professionals, technologists and engineers, today opened early registration for the 2026 Media Technology Summit (MTS) and announced i...
31/07/2026
Clear-Com announced that Central Christian Church has transformed communication across its live production operations by implementing Clear-Com's LQ Seri...
31/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
31/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
31/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
31/07/2026
IBC has unveiled the finalists for the IBC2026 Innovation Awards, recognising collaborative projects from around the world that solve real-world challenges and ...
31/07/2026
From focus puller to colourist
Caroline Shawley July 30, 2026
0 Comments
Ana Mar a Ormaza shares her story of instinct, storytelling and finding a cre...
31/07/2026
Amazon Prime Documentary Andata e Ritorno Relies on Blackmagic Design
Brie Clayton July 30, 2026
0 Comments
DaVinci Resolve Studio and Blackmagic Clou...
31/07/2026
Krotos Launches Video to Sound Plugin for DaVinci Resolve
Brie Clayton July 30, 2026
0 Comments
AI-assisted workflow helps editors add synchronized, p...
31/07/2026
Athletics, Camogie, the Dublin Horse Show, the FAI Cup, Golf, Hockey, the UEFA C...
31/07/2026
Luxembourg, July 31, 2026 - Pursuant to the liquidity contract entered into by SES with BNP Paribas as of 7 April 2026, please see the below update on the progr...
31/07/2026
Luxembourg, July 31, 2026 - SES today published a restatement of its FY2025 EU Taxonomy (Article 8) disclosure. The restatement updates the FY2025 EU Taxonomy d...
31/07/2026
RT Commercial has today announced Mercedes-Benz Ireland will return as the offi...
31/07/2026
RT Supporting the Arts is delighted to spotlight a diverse range of arts, culture and heritage events taking place across Ireland this August. From major festi...