Throughout 2021 and early 2022, Rambus has continued to make data faster and safer with the launch of key products, industry initiatives, and strategic partnerships. To address the insatiable demand for more bandwidth in the data center, we announced our 8.4 Gbps HBM3-Ready Memory Subsystem, confirmed the sampling of our DDR5 5600 MT/s 2nd-generation RCD chip, demonstrated our PCI Express® (PCIe) 5.0 digital controller IP on leading FPGA platforms, and unveiled our CXL Memory Interconnect Initiative. Looking ahead to 2022 and beyond, these products, initiatives, and partnerships will help power the next generation of bandwidth-hungry AI/ML applications and support the new accelerators and servers arriving in data centers over the coming months.
We also continued to meet increased demand for a hardware-based security paradigm across multiple verticals, including the IoT and automotive markets. To help protect IoT devices, Kyocera selected the FIPS 140-2 CMVP-certified Rambus RT-130 Root of Trust and AES-IP-38 AES Accelerator, while NextChip chose the Rambus RT-640 Root of Trust and MACsec-IP-160 Protocol Engine to secure its Apache6 automotive processor. As high-profile security exploits, breaches, and counterfeit silicon multiply in 2022, we will see an increasing emphasis placed on a hardware-based security paradigm in both the IoT and automotive spaces. To be sure, we expect a proliferation of dedicated silicon that is specifically designed to protect sensitive cryptographic functions and data. This model is the most effective way to secure data when at rest (processed or stored in a device) and when in motion (communicated between connected devices).
Let’s take a more in-depth look at how Rambus continues to make data faster and safer in 2022 and beyond.
Faster Speeds for Higher Bandwidth
In the summer of 2021, we announced our HBM3-ready memory interface subsystem comprising a fully integrated PHY and digital controller. Supporting breakthrough data rates of up to 8.4 Gbps, the solution delivers over a terabyte per second of bandwidth—more than double that of high-end HBM2E memory subsystems. According to Soo Kyoum Kim, associate VP, Memory Semiconductors at IDC, the memory bandwidth requirements of AI/ML training are “insatiable,” with leading-edge training models now surpassing billions of parameters. As Kim emphasizes, the Rambus HBM3-ready memory subsystem “raises the bar” for performance enabling state-of-the-art AI/ML and HPC applications.
According to Joel Hruska of ExtremeTech, early HBM3 hardware should be capable of ~1.4x more bandwidth than current HBM2E. However, as the standard evolves, that figure will rise to ~1.075TB/s of memory bandwidth per stack, with maximum I/O transfer rates of up to 8.4Gbps.
“These figures are per stack and many GPUs use HBM with 2-4 stacks, so total bandwidth provided by a four-stack HBM3 solution at 665GB/s is ~2.7TB/s,” he adds. It should be noted that Both AMD (Genoa) and Intel (Sapphire Rapids) are expected to begin shipping their respective HBM-equipped server processors in 2022.
In late 2021, we confirmed the sampling of our 5600 MT/s 2nd-generation RCD chip with major DDR5 memory module (RDIMM) suppliers. The new level of performance represents a 17% increase in data rate over the first-generation 4800 MT/s Rambus DDR5 RCD. With DDR5 memory, more intelligence is built into the DIMMs, enabling up to double the data rate and four times the capacity of DDR4 DIMMs, while at the same time reducing power and increasing memory efficiency.
According to Shane Rau, research vice president, Computing Semiconductors at IDC, advanced workloads are driving the increased demand for greater memory bandwidth.
“It [is therefore] essential that DDR5 ecosystem players like Rambus continue to raise the bar on performance to meet the rapidly rising needs of data center applications,” says Rau.
As we noted in our introduction, Rambus memory interface chips will enable next-generation DDR5-based servers to achieve new levels of performance. These new servers are slated to hit data centers in 2022 and beyond, with RDIMMs running at 4800 MT/s. This number represents a 33% increase in data rate over top-end 3200 MT/s DDR4 RDIMMs in current high-performance servers.
CXL™ Memory Interconnect Initiative
In the closing months of 2021, we announced our CXL Memory Interconnect Initiative to develop semiconductor solutions for advanced data center architectures that maximize performance, improve efficiency, and reduce system cost. Compute Express Link™ (CXL) is an open industry standard interconnect delivering high-bandwidth, low-latency connectivity between dedicated compute, memory, I/O and storage elements within the data center to allow the provision of the optimal mix of each for a given workload.
CXL memory expansion and pooling chips are key components for both traditional and disaggregated architectures. To support the continuing growth and specialization in server workloads, data centers are moving to disaggregated architectures composed from shared and scalable pools of computing and memory resources. CXL is a critical enabler of these next-generation disaggregated server architectures.
According to Matt Jones, general manager of IP cores at Rambus, CXL interconnects are quite versatile due to their high-bandwidth, low-latency characteristics—and can therefore be used to interconnect various hosts and resources in the system.
“We see the device evolving here into one that supports multiple hosts on the upstream side and being able to share efficiently a pool of memory on the downstream side so that you can assign multiple hosts to efficiently share that memory,” Jones tells The Next Platform. “The key building blocks are here from an IP standpoint that tie back to the acquisitions we made on the PHY and the controller on both sides.”
Securing Silicon to Protect Data
Kyocera Selects Rambus Root of Trust for IoT Security
In early 2022, Rambus announced that Kyocera Evolution Series MFPs will offer data security meeting Federal Information Processing Standards (FIPS) 140-2 Cryptographic Module Validation Program (CMVP) standards using the Rambus RT-130 Root of Trust and AES-IP-38 AES Accelerator IP. Specifically, the FIPS-certified Kyocera Evolution Series MFPs utilize the Rambus RT-130 Root of Trust and AES-IP-38 AES Accelerator as part of a system security architecture that provides the most robust and up-to-date protection for customers.
According to Neeraj Paliwal, general manager of security IP at Rambus, secure by design is a fundamental property of solutions from industry leaders like Kyocera. By building on Rambus FIPS CMVP-certified IP solutions, chip and system providers can better navigate the certification process and accelerate the development of secure solutions.
This is increasingly important in 2022 and beyond. Because data centers have transformed into virtual fortresses (both in the physical and digital domains), adversaries have turned their focus to more vulnerable edge and end points.
NextChip Selects Rambus Security IP to Secure Apache6 Automotive Processor
In January 2022, NextChip selected the Rambus RT-640 Root of Trust and MACsec-IP-160 Protocol Engine to provide hardware-level security for its next-generation Apache6 automotive processor. The Apache6 ADAS SoC combines CPU, GPU, ISP, and NPU processors to enable advanced automotive vision and domain/zone controller applications such as AVP.
The Rambus RT-640 Root of Trust provides security services and protection of data processed by the Apache6 SoC. The RT-640 is a powerful security co-processor featuring automotive grade embedded security software, high-performance cryptographic accelerators for AES, HMAC, SHA-2, and more. In addition, dedicated safety integrity mechanisms ensure correct operations and extensive error handling and the advanced anti-tamper features of the RT-640 protect chips from side-channel and fault injection (FI) attacks. Meanwhile, the Rambus MACsec-IP-160 encrypts and protects data at speeds up to 100 Gbps over Ethernet in-car networks.
According to CTO NextChip Hweihn Chung, the company is raising the bar for reliable, compact, and affordable ADAS solutions with the Apache6.
“With Rambus security IP solutions, Apache6 offers state-of-the-art protection of mission-critical data while meeting full ASIL-B compliance,” he adds.
Rambus continues to make data faster and safer with the launch of key products, industry initiatives, and strategic partnerships. In 2022 and beyond, we are addressing the insatiable demand for more bandwidth in the data center to support new AI/ML applications with our 8.4 Gbps HBM3-Ready Memory Subsystem, DDR5 5600 MT/s 2nd-generation RCD chip, and CXL Memory Interconnect Initiative.
On the security side, we continue to meet increased demand for a hardware-based security paradigm across multiple verticals, including the IoT and automotive markets. Recent examples include Kyocera selecting the FIPS 140-2 CMVP-certified Rambus RT-130 Root of Trust and AES-IP-38 AES Accelerator for its IoT silicon, and NextChip choosing the Rambus RT-640 Root of Trust and MACsec-IP-160 Protocol Engine to secure its Apache6 automotive processors. As high-profile security exploits, breaches, and counterfeit silicon multiply in 2022, we will see an increasing emphasis placed on a hardware-based security paradigm in both the IoT and automotive spaces. To be sure, we expect a proliferation of dedicated silicon that is specifically designed to protect sensitive cryptographic functions and data.