IBM researchers have demonstrated a way to bring optics' speed and capacity inside data centers.
In a technical paper, IBM introduced a new co-packaged optics (CPO) prototype module that can enable high-speed optical connectivity. Researchers have pioneered a new process for CPO to enable connectivity within data centers at the speed of light to complement existing short reach electrical wires. This technology could significantly increase the bandwidth of data center communications, minimizing GPU downtime while drastically accelerating AI processing. This research innovation, as described, could dramatically improve how data centers train and run generative AI models, IBM said.
By designing and assembling the first publicly announced successful polymer optical waveguide (PWG) to power this technology, IBM researchers demonstrated how CPO will redefine the way the computing industry transmits high-bandwidth data between chips, circuit boards, and servers.
Today, fiber optic technology carries data at high speeds across long distances for commerce and communications. Data centers use fiber optics for external communications networks, but racks inside those data centers still predominantly run communications on copper-based electrical wires. These wires connect GPU accelerators that may spend more than half of their time idle, waiting for data from other devices in a large, distributed training process which can incur significant expense and energy, IBM noted.
This new approach, IBM said, could result in lower costs for scaling generative AI through a more than 5x power reduction in energy consumption compared to mid-range electrical interconnects, while extending the length of data center interconnect cables from one to hundreds of meters.
In addition, it could speed up AI model training, enabling developers to train a large language model (LLM) up to five times faster -- reducing the time it takes to train a standard LLM from three months to three weeks, with performance gains increasing by using larger models and more GPUs. And it could dramatically increase energy efficiency for data centers, saving the energy equivalent of 5,000 U.S. homes' annual power consumption per AI model trained, IBM researchers said.
As always, drop me a line at [email protected] if you have news to share or want to say hi!
Grab your coffee. Here's what you need to know today.
1. Cirata, Databricks partnership: Hadoop data transfer and integration firm Cirata this week is expanding its partnership with data and AI company Databricks to speed up data migration and modernization in the cloud. Cirata's new Data Migration as a Service (DMaaS) offering will enable easier migration and modernization of data, the companies said. Via the Databricks Data Intelligence Platform, customers can leverage Cirata's professional services to transition legacy data infrastructures, such as Hadoop, to Databricks.
2. New HPE storage for SMBs: HPE last week introduced the HPE MSA Gen7 Storage System, providing higher storage capacity, speed, and data protection for SMBs at affordable prices, the company said. HPE said the new solutions offer up to 2X faster random read speeds (up to 783,000 IOPS) with throughput reaching 14.1 GB/s, and scalability up to 7.37 PB. There's also streamlined management features, including Smart Assist, online firmware updates, and the HPE MSA Health Check tool to improve system reliability. HPE MSA Gen7's also includes intuitive sizing tools so HPE partners can design customized solutions for customers.
3. Mission AI Gateway: Mission, a CDW company and U.S.-based Amazon Web Services (AWS) Premier Tier Services Partner, announced the launch of Mission AI Gateway, a new software platform that helps enterprises manage and optimize their AI solutions on AWS. Mission AI Gateway combines advanced FinOps capabilities, infrastructure operations expertise, and strategic architectural guidance to ensure AI solutions are aligned with AWS best practices. The platform integrates with Mission Control, Mission's cloud services platform, to provide visibility and management across AI workloads.
4. Actively exploited Cleo flaw: Cleo's Harmony, LexiCom, and VLTrader managed file transfer products are impacted by an actively exploited zero-day vulnerability, tracked as CVE-2024-50623, which has been inadequately resolved by a previous patch, according to The Record, a news site by cybersecurity firm Recorded Future. Users have been urged to promptly apply a new fix, and Cleo also recommended immediate blocking of IP addresses leveraging the issue.
5. Sherweb acquires MicroWarehouse: Sherweb this week officially announced its acquisition of MicroWarehouse (MWH), an Ireland-based cloud and IT distributor. Terms of the deal were not disclosed, but the companies said in a statement that MicroWarehouse will continue to operate as a local business unit in Ireland, maintaining its own brand identity.