Clockwork.io Secures $31 Million to Enhance AI Workload Resilience with Innovations
In a significant leap for AI infrastructure, Clockwork.io has raised $31 million in its latest funding round, aimed at advancing its fault-tolerance solutions for GPU workloads. This funding comes at a time when large-scale AI operations face myriad challenges involving downtime and inefficiencies, significantly impacting performance and resource utilization. LinkedIn, Together AI, and WhiteFiber are among the prominent players adopting Clockwork.io's innovative software solutions, designed specifically to maintain the integrity of AI training, reinforcement learning, and inference workloads despite the occurrence of infrastructure failures.
The challenges in managing AI workloads are substantial, as companies rely on thousands of GPUs that must remain synchronized. Even a minor issue, such as a malfunctioning GPU or a dropped network link, can derail an entire job, leading to significant delays and wasted computational hours. Recent benchmarks indicate that during the training of Meta's Llama 3 AI model, unexpected interruptions led to an average downtime every three hours, underscoring the critical need for effective fault-tolerance solutions.
Clockwork.io's innovative , named TorchPass, introduces cutting-edge capabilities such as multi-node platform snapshots. This feature allows teams to capture the state of a running distributed AI job without requiring code modifications. Similar advancements include asynchronous application checkpoints that run in the background while the job is ongoing, enhancing the speed of reinforcement learning by updating model weights more swiftly. As a result, inference replicas can be trained using the most up-to-date information and can operate with minimal interruptions.
The system is built on the premise that failures are inevitable in large-scale AI operations. Clockwork.io addresses this by preventing hours of productive work from being lost during unforeseen disruptions. Their solutions allow infrastructure teams to divert traffic around damaged links or faulty GPUs, ensuring continuity in operations. Suresh Vasudevan, the CEO of Clockwork.io, emphasized that this fault-tolerance capability is a multiplier for workforce efficiency, as it reduces the idle time of GPUs and enables them to focus on performing productive tasks.
As organizations increasingly rely on their GPU fleets, the adoption of Clockwork.io's solutions is rapidly expanding across various enterprises, hyperscalers, and neo-cloud platforms. LinkedIn, for example, has successfully implemented LinkPass technology within its AI infrastructure, effectively eliminating tens of thousands of GPU-hours of downtime each month. Raghu Hiremagalur, SVP and CTO of Infrastructure at LinkedIn, highlighted the valuable transformation brought about by Clockwork.io’s technology, stating that their solutions have redefined the operational model by automating traffic rerouting during infrastructure faults.
In addition, Clockwork.io's partnership with Together AI marks a significant step in bringing TorchPass to market on a service basis, showcasing its capabilities in live demonstrations. Together AI aims to emphasize the reliability of their GPU Clusters while demonstrating how Clockwork.io’s solutions can keep multi-node training jobs operational, even in the event of network or GPU failures.
Furthermore, Clockwork.io is not only improving existing infrastructure but also expanding its capabilities to new clients. WhiteFiber, known for its GPU-as-a-service model, is broadening its deployment of Clockwork.io's solutions to enhance their operational efficiency and customer satisfaction. Chief Technology Officer Tom Sanfilippo stressed the importance of validating cluster reliability before deployment to mitigate hidden faults that could impact customers negatively.
The latest funding round was co-led by several venture capital firms, including Premji Invest, Wing Venture Capital, and Seligman Ventures, alongside existing investors such as NEA and e Capital. It propels Clockwork.io's total funding to $73 million, which will facilitate an expedited rollout of their fault-tolerance suite across AI training and inference frameworks, asserting the company’s position as a leader in AI infrastructure reliability.
To summarize, with the contemporary demands and scale of AI operations, the increasing adoption of Clockwork.io’s fault-tolerance solutions is crucial. By seamlessly integrating these innovations into their operations, organizations can ensure the continuous function of AI workloads, ultimately leading to greater efficiency and productivity across various sectors.
The challenges in managing AI workloads are substantial, as companies rely on thousands of GPUs that must remain synchronized. Even a minor issue, such as a malfunctioning GPU or a dropped network link, can derail an entire job, leading to significant delays and wasted computational hours. Recent benchmarks indicate that during the training of Meta's Llama 3 AI model, unexpected interruptions led to an average downtime every three hours, underscoring the critical need for effective fault-tolerance solutions.
Clockwork.io's innovative , named TorchPass, introduces cutting-edge capabilities such as multi-node platform snapshots. This feature allows teams to capture the state of a running distributed AI job without requiring code modifications. Similar advancements include asynchronous application checkpoints that run in the background while the job is ongoing, enhancing the speed of reinforcement learning by updating model weights more swiftly. As a result, inference replicas can be trained using the most up-to-date information and can operate with minimal interruptions.
The system is built on the premise that failures are inevitable in large-scale AI operations. Clockwork.io addresses this by preventing hours of productive work from being lost during unforeseen disruptions. Their solutions allow infrastructure teams to divert traffic around damaged links or faulty GPUs, ensuring continuity in operations. Suresh Vasudevan, the CEO of Clockwork.io, emphasized that this fault-tolerance capability is a multiplier for workforce efficiency, as it reduces the idle time of GPUs and enables them to focus on performing productive tasks.
As organizations increasingly rely on their GPU fleets, the adoption of Clockwork.io's solutions is rapidly expanding across various enterprises, hyperscalers, and neo-cloud platforms. LinkedIn, for example, has successfully implemented LinkPass technology within its AI infrastructure, effectively eliminating tens of thousands of GPU-hours of downtime each month. Raghu Hiremagalur, SVP and CTO of Infrastructure at LinkedIn, highlighted the valuable transformation brought about by Clockwork.io’s technology, stating that their solutions have redefined the operational model by automating traffic rerouting during infrastructure faults.
In addition, Clockwork.io's partnership with Together AI marks a significant step in bringing TorchPass to market on a service basis, showcasing its capabilities in live demonstrations. Together AI aims to emphasize the reliability of their GPU Clusters while demonstrating how Clockwork.io’s solutions can keep multi-node training jobs operational, even in the event of network or GPU failures.
Furthermore, Clockwork.io is not only improving existing infrastructure but also expanding its capabilities to new clients. WhiteFiber, known for its GPU-as-a-service model, is broadening its deployment of Clockwork.io's solutions to enhance their operational efficiency and customer satisfaction. Chief Technology Officer Tom Sanfilippo stressed the importance of validating cluster reliability before deployment to mitigate hidden faults that could impact customers negatively.
The latest funding round was co-led by several venture capital firms, including Premji Invest, Wing Venture Capital, and Seligman Ventures, alongside existing investors such as NEA and e Capital. It propels Clockwork.io's total funding to $73 million, which will facilitate an expedited rollout of their fault-tolerance suite across AI training and inference frameworks, asserting the company’s position as a leader in AI infrastructure reliability.
To summarize, with the contemporary demands and scale of AI operations, the increasing adoption of Clockwork.io’s fault-tolerance solutions is crucial. By seamlessly integrating these innovations into their operations, organizations can ensure the continuous function of AI workloads, ultimately leading to greater efficiency and productivity across various sectors.
Topics Consumer Technology）
【About Using Articles】
You can freely use the title and article content by linking to the page where the article is posted.
※ Images cannot be used.
【About Links】
Links are free to use.