Categories
News

New Marvell OCTEON 10 networking processors

New Marvell OCTEON 10 processors

In the rapidly evolving world of technology, Marvell Technology has once again made a significant mark with the introduction of two new OCTEON 10 data processing units (DPUs), the OCTEON 10 CN102 and OCTEON 10 CN103 networking processors. These innovative units are designed to enhance the performance and efficiency of networking equipment and firewall manufacturers, bringing server-class performance to networking devices.

The OCTEON 10 CN102 and CN103 are optimized for data and control plane applications in various devices such as routers, firewalls, 5G small cells, SD-WAN appliances, and control plane applications in top-of-rack switches and line card controllers. These units are manufactured using 5 nm process technology, showcasing the company’s commitment to staying at the forefront of technology innovation.

OCTEON 10 DPU platform

These new DPUs from Marvell Technology deliver three times the performance of the company’s current DPU solutions, a significant leap in processing power. But it’s not just about increased performance. Marvell has also managed to reduce power consumption by 50% to 25 W, a critical factor in today’s energy-conscious world. These units can serve as an offload DPU for host processors or as the primary processor in devices, offering flexibility in application.

The OCTEON 10 DPU platform, first announced in 2021, was designed to meet the growing requirements for network, storage, and security processing among various end users. It shares a common foundation of 5 nm process technology, Arm Neoverse N2 processors, hardware accelerators, industry-leading I/O, and DDR5 support, among other technologies.

Marvell’s approach to the OCTEON 10 DPU platform is based on the idea that every market and customer segment deserves optimized processors. To this end, the company varies the number of CPU cores, the optimal data paths, the type and number of hardware accelerators, and other technologies to create different models optimized for specific applications, customer segments, and performance requirements. This approach allows Marvell to cater to a wide range of customer needs and applications. Here are some other articles you may find of interest on the subject of networking :

Networking processors

The OCTEON 10 CN102 and CN103 are part of Marvell’s broader compute strategy to increase overall performance while holding down power and chip cost. This strategy is key to the company’s commitment to delivering high-performance, energy-efficient solutions to its customers.

Marvell’s OCTEON technology has been incorporated into platforms by carriers, equipment vendors, and other semiconductor designers. The new OCTEON 10 CN102 and CN103 continue this trend, delivering technology usually associated with hyperscale cloud data centers for accelerating network, storage, and security workloads to OEMs and enterprises.

Other OCTEON 10 DPUs include OCTEON 10 CN106, designed for cloud, enterprise, and baseband for 5G wireless networks, and OCTEON 10 Fusion CN105, optimized for the intense radio processing demands of massive MIMO and other 5G RAN tasks. These units further highlight the versatility and range of the OCTEON 10 DPU platform.

In terms of availability, the OCTEON 10 CN102 and CN103 DPUs are set to be available in production quantities in Q4 of this year and Q1 of 2024, respectively. This timeline highlights Marvell’s commitment to delivering its innovative solutions to customers in a timely manner.

The introduction of the new OCTEON 10 CN102 and CN103 DPUs by Marvell Technology marks a significant step forward in the realm of data processing units. With improved performance, reduced power consumption, and application in a wide range of devices, these units are set to bring server-class performance to networking devices, meeting and exceeding the demands of today’s technology landscape.

Source: Marvell

Filed Under: Technology News, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.

Categories
News

New NVIDIA Ethernet AI networking platform soon available

New NVIDIA Ethernet networking platform for AI soon available

NVIDIA has announce the imminent availability of its new Spectrum-X Ethernet networking solution, a technology poised to significantly boost AI system communication. This leap forward in networking performance is not just a standalone achievement; it’s part of a strategic alliance with industry giants Dell Technologies, Hewlett Packard Enterprise (HPE), and Lenovo. This collaboration ensures that Spectrum-X will be a core component of their server offerings, empowering enterprises to tackle the increasing demands of generative AI workloads with newfound efficiency.

AI applications are becoming more complex and data-intensive, requiring robust communication systems that can keep up. NVIDIA’s Spectrum-X is engineered to meet these challenges head-on, delivering a 1.6 times increase in networking performance over traditional Ethernet solutions. This enhancement is more than just a number—it’s the key to unlocking faster and more reliable AI communication, which is essential for the success of AI-driven projects.

The partnership between NVIDIA and the server manufacturers is a strategic move that goes beyond just improving performance. It’s about creating a unified system that leverages the full potential of each component. The forthcoming server systems will integrate Spectrum-X with NVIDIA’s advanced Tensor Core GPUs, AI Enterprise software, and AI Workbench software, creating an ecosystem that’s perfectly attuned to the needs of AI research and development.

Spectrum-X Ethernet AI networking

Leaders from NVIDIA, Dell, HPE, and Lenovo have all emphasized the significance of this joint effort. They recognize that the future of generative AI hinges on the seamless integration of accelerated networking and computing power. By combining these technologies, they are setting new standards for AI applications, from natural language processing to complex data analysis.

At the heart of Spectrum-X lies the Spectrum-4 Ethernet switch, a groundbreaking piece of technology that’s the first of its kind—a 51Tb/sec Ethernet switch designed with AI in mind. Its role is to reduce network congestion and streamline data flow, allowing AI systems to function with exceptional efficiency.

The Spectrum-4’s capabilities are further enhanced by the BlueField-3 SuperNIC network accelerator. These accelerators are not just about speed; they also provide secure, multi-tenant environments that are essential for modern data centers. In today’s world, where cloud computing and as-a-service models dominate, the need for security and isolation is paramount. The BlueField-3 SuperNICs are engineered to meet these critical requirements.

Other articles we have written that you may find of interest on the subject of generative AI :

Networks  designed for artificial intelligence

NVIDIA’s acceleration software suite, which includes development kits and the DOCA software framework, complements the hardware. These tools are designed to maximize the platform’s performance, giving developers the ability to harness the full power of the hardware. For those involved in AI model development, NVIDIA’s AI Enterprise and AI Workbench offer a set of indispensable resources. These tools are crucial for creating and scaling AI models, providing a foundation for innovation and progress.

The NVIDIA Israel-1 supercomputer, which runs on Spectrum-X, exemplifies the potential of this technology. It serves as a reference architecture for future AI systems, showcasing what can be achieved when these advanced technologies are integrated in a complementary fashion. The new systems that incorporate NVIDIA’s complete AI stack are expected to hit the market in the first quarter of the next year. This gives you ample time to plan and prepare for the integration of these sophisticated systems into your business operations.

As you consider the next steps for your organization’s AI capabilities, NVIDIA’s Spectrum-X Ethernet networking, combined with the server solutions from Dell, HPE, and Lenovo, represents a pivotal moment. With these cutting-edge tools at your disposal, you are well-positioned to tackle the most demanding AI workloads with confidence and skill. This strategic move by NVIDIA and its partners is not just about keeping pace with the evolution of AI—it’s about setting the pace, ensuring that enterprises like yours are always one step ahead.

Source: NVIDIA

Filed Under: Gadgets News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.