
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 ...
09/10/2026
September 10 2026, 06:00 (PDT) Dolby Expands Dolby OptiView Platform with New Capabilities at IBC 2026
New Sports Intelligence helps providers better unders...
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...
22/09/2026
From Malaysia to Mexico and from DR Congo to Ukraine, this year's 12 finalists for the 2026 Thomson Foundation Young Journalist of the Year award tell extra...
22/09/2026
The annual preseason content-gathering event brought 36 NHL players to Las Vegas, where ESPN used 2.5 kilowatts of laser power and a motion-control camera syste...
22/09/2026
Neumann has introduced the KK 104 A (cardioid) and KK 105 A (supercardioid) wireless capsule heads, succeeding the KK 204 and KK 205. The new capsules are compa...
22/09/2026
The NFL and TMRW Sports have announced a new annual Pro Flag Showcase to be held during Super Bowl LXI week in Los Angeles. The event will feature top flag foot...
22/09/2026
Washington State University has unveiled a new end zone video display and audio ...
22/09/2026
Disney's SVP of Live Operations and Engineering details the technology behin...
22/09/2026
Cosm has announced two senior promotions. Mazen Alawar has been promoted to SVP,...
22/09/2026
Mississippi State University's University Television Center has purchased four KOKUSAI DENKI SK-UHD7000-S2 cameras since 2024, with plans to add two more, a...
22/09/2026
Net Insight has announced that Christine Kjellvard will assume the role of interim Chief Financial Officer on November 1, 2026. Kjellvard has previously held in...
22/09/2026
Net Insight has announced Zyntai TimeNode Access, a compact PTP Grandmaster designed for distributed edge sites. The product carries secure, precise time across...
22/09/2026
Sam Hazeldine, Austin Amelio, Padraic McKinley, Julia Jones, and Ethan Hawke attend the premiere of The Weight at The Ray Theatre on January 26, 2026, in Park...
22/09/2026
Second high-gain tone plug-in released
Following on from their debut release, Chainsaw Suite, Avalanche Tones have announced the launch of their second plug...
22/09/2026
Uncompressed audio with below 6ms of latency
Lewitt's have just introduced their first pair of headphones which, despite being aimed primarily at studio...
22/09/2026
Two new capsule heads for Sennheiser wireless systems
Neumann have announced the upcoming launch of the KK 104 A and KK 105 A, a pair of new premium condens...
22/09/2026
Apogee Plugins Are Moving to Plugin Alliance We're excited to share an important update about the future of Apogee plugins.
Apogee plugins are moving to Pl...
22/09/2026
Parkland, Florida, US, September 22, 2026: SipRadius, widely recognized for making content processing and connectivity secure and seamless, has been honored wit...
22/09/2026
VanityFilter 3.1.5 adds project files and adjustable protection around lips, bro...
22/09/2026
Chile's TV Modernizes Broadcast Operations with Blackmagic Design
Brie Clayton September 22, 2026
0 Comments
URSA Broadcast G2 brings versatility...
22/09/2026
OWC Shares Sneak Peek at Amazon Prime Big Deal Days Pricing
Brie Clayton September 22, 2026
0 Comments
Image courtesy of Deposit Photos
Business and ...
22/09/2026
Berklee's First Homecoming Block Party Brings Alumni Home for Global Music F...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
22/09/2026
From Academy Award-winning Passion Pictures and BAFTA-winning director Marian Mo...
22/09/2026
Tuesday 22 September 2026
Portrait Artist of the Year returns to Sky Arts with ...
22/09/2026
Whether it be from the sun or from our phones and other devices, we are all exposed to electromagnetic radiation in some shape or form. With the dramatic growth...
22/09/2026
Comscore's latest AI Intelligence Report Shows Competition Intensifying as C...
22/09/2026
The Factory-X Project Sets Standards for Data Sovereignty in Industry
Arvato Systems Brings Digital Sovereignty to Industrial Practice
Factory-X Successfully...
22/09/2026
LONDON Apple today announced the opening of Apple Music Hall, a brand-new state-...
22/09/2026
RT , Creative Ireland and Shared Island Initiative ask young artists
to respon...
22/09/2026
To build and deploy sophisticated robotics applications that can perceive, reason and act in dynamic environments, developers need new physical AI models and to...
21/09/2026
Manifold Technologies, developer of software-defined, FPGA-accelerated infrastructure for live production, has named Ryhaan Williams as Chief Commercial Officer...
21/09/2026
Studio Network Solutions (SNS) has won Best Journal Article at the 2026 IAMT Impact Awards for What Video Editors Need When Everything Is Changing Around Them,...
21/09/2026
Major League Wrestling (MLW) has announced a streaming distribution agreement with Tubi for its weekly series MLW FUSION. Beginning September 17, new episodes w...
21/09/2026
Disney Entertainment has announced that Adam Smith has been named Chairman, Direct-to-Consumer, with responsibility for Disney and Hulu including product, engi...
21/09/2026
United Airlines and DISH have announced an agreement to bring live professional ...
21/09/2026
SES and JioStar have announced a multi-year agreement under which SES will provide satellite distribution services for JioStar's television channel portfoli...
21/09/2026
The Detroit Red Wings have announced details for their inaugural season on Detro...
21/09/2026
Manifold Technologies has appointed Ryhaan Williams as Chief Commercial Officer for North America, effective October 1, 2026. Based in Jacksonville, Florida, Wi...
21/09/2026
Angels Broadcast Television (ABTV), the local television home of the Los Angeles...