
Pinterest has engineered a way to serve its photo-sharing community more of the images they love.
The social-image service, with more than 400 million monthly active users, has trained bigger recommender models for improved accuracy at predicting people's interests.
Pinterest handles hundreds of millions of user requests an hour on any given day. And it must also narrow down relevant images from roughly 300 billion images on the site to roughly 50 for each person.
The last step - ranking the most relevant and engaging content for everyone using Pinterest - required a leap in acceleration to run heftier models, with minimal latency, for better predictions.
Pinterest has improved the accuracy of its recommender models powering people's home feeds and other areas, increasing engagement by as much as 16%.
The leap was enabled by switching from CPUs to NVIDIA GPUs, which could easily be applied next to other areas, including advertising images, according to Pinterest.
Normally we would be happy with a 2% increase, and 16% is just a beginning for home feeds. We see additional gains - it opens a lot of doors for opportunities, said Pong Eksombatchai, a software engineer at Pinterest.
Transformer models capable of better predictions are shaking up industries from retail to entertainment and advertising. But their leaps in performance gains of the past few years have come with a need to serve models that are some 100x bigger as their number of model parameters and computations skyrockets.
Huge Inference Gains, Same Infrastructure Cost Like many, Pinterest engineers wanted to tap into state-of-the-art recommender models to increase engagement. But serving these massive models on CPUs presented a 100x increase in cost and latency. That wasn't going to maintain its magical user experience - fresh and more appealing images - occurring within a fraction of a second.
If that latency happened, then obviously our users wouldn't like that very much because they would have to wait forever, said Eksombatchai. We are pretty close to the limit of what we can do on CPU basically.
The challenge was to serve these hundredfold larger recommender models within the same cost and latency constraints.
Working with NVIDIA, Pinterest engineers began architectural changes to optimize their inference pipeline and recommender models to enable the transition from CPU to GPU cloud instances. The technology transition began late last year and required major changes to how the company manages workloads. The result is a 100x gain in inference efficiency on the same IT budget, meeting their goals.
We are starting to use really, really big models now. And that is where the GPU comes in - to help make these models possible, Eksombatchai said.
Tapping Into cuCollections Switching from CPUs to GPUs required rethinking its inference systems architecture. Among other issues, engineers had to change how they send workloads to their inference servers. Fortunately, there are tools to assist in making the transition easier.
The Pinterest inference server built for CPUs had to be altered because it was set up to send smaller batch sizes to its servers. GPUs can handle much larger workloads, so it's necessary to set up larger batch requests to increase efficiency.
One area where this comes into play is with its embedding table lookup module. Embedding tables are used to track interactions between various context-specific features and interests of user profiles. They can track where you navigate, and what people Pin on Pinterest, share or numerous other actions, helping refine predictions on what users might like to click on next.
They are used to incrementally learn user preference based on context in order to make better content recommendations to those using Pinterest. Its embedding table lookup module required two computation steps repeated hundreds of times because of the number of features tracked.
Pinterest engineers greatly reduced this number of operations using a GPU-accelerated concurrent hash table from NVIDIA cuCollections. And they set up a custom consolidated embedding lookup module so they could merge requests into a single lookup. Better results were seen immediately.
Using cuCollections helped us to remove bottlenecks, said Eksombatchai.
Enlisting CUDA Graphs Pinterest relied on CUDA Graphs to eliminate what was remaining of the small batch operations, further optimizing its inference models.
CUDA Graphs helps reduce the CPU interactions when launching on GPUs. They're designed to enable workloads to be defined as graphs rather than single operations. They provide a mechanism to launch multiple GPU operations through a single CPU operation, reducing CPU overheads.
Pinterest enlisted CUDA Graphs to represent the model inference process as a static graph of operation instead of as those individually scheduled. This enabled the computation to be handled as a single unit without any kernel launching overhead.
The company now supports CUDA Graph as a new backend of its model server. When a model is first loaded, the model server runs the model inference once to build the graph instance. This graph can then be run repeatedly in inference to show content on its app or site.
Implementing CUDA Graphs helped Pinterest to significantly reduce inference latency of its recommender models, according to its engineers.
GPUs have enabled Pinterest to do something that was impossible with CPUs on the same budget, and by doing this they can make changes that have a direct impact on various business metrics.
Learn about Pinterest's GPU-driven inference and optimizations at its GTC session, Serving 100x Bigger Recommender Models, and in the Pinterest Engineering blog.
Register for GTC, running Sept. 19-22, for free to attend sessions with NVIDIA and dozens of industry leaders.
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...
17/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
17/07/2026
Foundry Releases SmartRoto for Nuke
Brie Clayton July 17, 2026
0 Comments
Spline-based AI powered plugin accelerates time-consuming rotoscoping, helping...
17/07/2026
A Short Documentary About a Giant Pencil Edited with DaVinci Resolve Studio
Brie Clayton July 17, 2026
0 Comments
SBIFF jury award winner finished fro...
17/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
17/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
17/07/2026
Lightware has published its second voluntary Sustainability Report, showing how energy efficiency, product longevity and responsible material use are increasing...
17/07/2026
Creative software developer Foundry today announced Griptape Enterprise, a new tier of the Griptape AI workflow orchestration platform, extended to meet the str...
17/07/2026
Creative software developer Foundry today announced the availability of SmartRoto, a new AI-powered plugin upgrade for Nuke, NukeX, Nuke Studio, and Nuke Indie ...
17/07/2026
The Final Release of Blackmagic Fairlight Live is Now Available!
Brie Clayton July 16, 2026
0 Comments
Fairlight Live is a new audio mixer designed fo...
17/07/2026
Leveraging All of Entertainment's IP
Andy Marken July 16, 2026
0 Comments
Nobody ever wins the games. Period. There are survivors. There\s no win...
17/07/2026
New and returning series, live festival broadcasts, powerful documentaries and major sporting moments headline RT Radio 1's summer schedule
RT Radio 1 is...
17/07/2026
Think of a professional athlete. What separates elite performers is what happens...
17/07/2026
Up For The Match - Live From Croke Park this Saturday at 8pm on RT One and RT ...
17/07/2026
RT Commercial has today announced DiscoverIreland.ie, as sponsor of The Traitor...
16/07/2026
Media organisations with tested plans for new products, services or revenue models can now apply for grants of up to 30,000 through the Business Innovation Syn...
16/07/2026
This recent graduate and Georgia native has turned an early fascination with live production into a growing passion for camera work
In the live-sports-video in...
16/07/2026
TNDV is marking the first anniversary of Aspiration 35, a mobile production truck built around ARRI Alexa 35 Live camera systems. Over its first year, the truck...
16/07/2026
ATSC has completed a major revision to its A/85 Recommended Practice: Techniques...
16/07/2026
FloSports has announced an exclusive global media partnership with CrossFit for the 2026 CrossFit Games, presented by Air National Guard, beginning July 21 in S...
16/07/2026
Sennheiser has released firmware version 1.4 for its Spectera wireless system and announced the Spectera Command Button (Cat. No. 701014). The update adds comma...
16/07/2026
ESPN and the Southwestern Athletic Conference (SWAC) have announced a multi-year...
16/07/2026
Audio-Technica has announced the 20 Series Control Application, a Stream Deck plug-in for compatible Audio-Technica USB microphones. The application is compatib...
16/07/2026
The fifteenth season of The Voice - La plus belle voix on TF1 used four Solid St...
16/07/2026
Gray Media and the Atlanta Hawks have announced a broadcast partnership that wil...
16/07/2026
Season 3 of Netflix's docuseries, which dropped this week, follows Jayden Da...
16/07/2026
The Texas Rangers and Rangers Sports Network presented by Progressive (RSN) have announced BZZR as the new direct-to-consumer distributor for Rangers game broad...
16/07/2026
With England's FIFA World Cup 2026 team getting ready to fly home after losi...
16/07/2026
In a return to Royal Birkdale, the team enjoys partnering with ETP and Gravity M...
16/07/2026
Along with a new studio at Ball Arena, the Colorado-based RSN is working with XR...
16/07/2026
Austria was back at the World Cup for the first time in 28 years, and the team d...
16/07/2026
With 42 RF cameras and high-power radio mics deployed, the spectrum environment ...
16/07/2026
A new benchmark in bass enhancement
Described as a new benchmark in bass enhancement , Baby Audio's latest release has been designed to augment the lo...
16/07/2026
Popular library rebuilt in new Orbita Engine
Other Worlds Evolved is the latest of Zero G's sample libraries to be treated to a ground-up rebuild using ...
16/07/2026
Created in collaboration with A. G. Cook
The newest addition to Native Instruments' software collection introduces a new synth created by A. G. Cook, a ...
16/07/2026
NITV and SBS bring Australians to the heart of Garma
16 July, 2026
Media releases
The Garma Festival returns from 31 July-3 August, with NITV and SBS deliv...
16/07/2026
Rohde & Schwarz supplies conformance test systems to Marquistech for India's...
16/07/2026
June brought stabilization to the television market. Poles spent an average of 3 hours and 36 minutes a day in front of their TV screens exactly the same as i...
16/07/2026
Glensound returns to IBC with a major new addition to its intelligent loudspeaker range, the Greater Divine Supernal studio monitor, alongside recent developmen...
16/07/2026
33 leading exhibitors are taking part in the GREAT Britain and Northern Ireland Pavilions and linked locations across IBC2026 (Amsterdam RAI, 11 14 September)...
16/07/2026
Hitomi Broadcast will introduce Spectra, a new HDR colour verification solution measured in picture, not using VPID, at IBC2026 (Hall 10, Stand A40). Expanding ...
16/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
16/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
16/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
16/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
16/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
16/07/2026
Zixi, a leader in live video delivery and workflow orchestration, and Comcast Technology Solutions (CTS) today announced a strategic new integration to deliver ...