JEDEC recently announced the ratification of JESD79-5 DDR5 SDRAM to support the standardization of next-generation memory devices, catering to demand from rapid expansion in high performance computing and data center applications. This new standard promises to deliver 2X memory bandwidth, 4X larger density dies, and much improved power efficiency (1.1V Vdd). The DDR5 DIMM will operate in dual-channel mode all on its own, with two 40-bit fully independent sub-channels on the same module.
The Joint Electron Device Engineering Council (JEDEC) has been developing and maintaining DRAM standards for years, defining emerging Memory standards like the DRAM standard. The most recent announcement declares the fifth generation of the DRAM, DDR5, is finally ready for release. The work to define DDR5 began in 2017 with the objective of delivering a standard that could move beyond the DDR4 speed limitations of 16 Gb and 3200 MT/s. The intention was to address new applications around data centers high-end servers for handling AI/ML workloads.
Modern computer applications rely heavily on graphics processing and rendering which involve a lot of simultaneous mathematical calculations. A typical CPU is not suitable for jobs that require simultaneous processing, which is why the concept of a dedicated Graphics Processing Unit (GPU) was introduced. The GPU has found its scope not only in graphics processing but also several emerging applications like AI, machine learning, VR, autonomous driving, and network routing.
We recently published the VIP Newsletter for Apr 2019, containing trending topics, leading solutions, in depth technical articles, videos, webinars, and updates on next generation protocols. In case you missed the latest buzz on Verification IP, you can read it here.
The key features driving future memories are memory density, speed, lower operating voltage, and faster access. DDR5 supports memory density from 8Gb to 64Gb combined with a wide range of data rate from 3200 MT/s to 6400 MT/s. The operating voltage of DDR5 is further reduced from 1.2V of DDR4 to 1.1V.
Synopsys recently announced the fastest, and most power efficient DDR5 and LPDDR5 IP solutions. Industry’s first LPDDR5 controller, PHY, and verification IP solution supports data rates up to 6400 Mbps with up to 40% less area than previous generations. The LPDDR5 IP provides significant area and power savings for mobile and automotive SoCs with its dual-channel memory interface option that shares common circuitry between independent channels. The DesignWare DDR5 IP, operating at up to 4800 Mbps data rates, can interface with multiple DIMMs per channel up to 80 bits wide, delivering the fastest DDR memory interface solution for artificial intelligence (AI) and data center system-on-chips (SoCs). The DDR5 and LPDDR5 controller and PHY seamlessly interoperate via the latest DFI 5.0 interface.
We recently published the VIP Newsletter for Q3 2018, containing trending topics, leading solutions, in depth technical articles, videos, webinars, and updates on next generation protocols. The newsletter covers content on DFI 5.0 for DDR5/LPDDR5, NVMe 1.3, USB 3.2, PCIe 5.0, next generation gaming displays, MIPI CSI-2 v2.1 for Automotive and IoT, and Verdi performance analyzer and protocol debug. In case you missed the latest buzz on Verification IP, you can read it here.
The growth of datacenter, storage, automotive and other emerging market applications is driving the development of next-generation memory technologies – DDR5, LPDDR5. Like their predecessors, the latest memory technologies also use DFI, a standard interface between memory controller and PHY, to reduce the integration cost and increase performance and data throughput efficiency. DFI also has evolved along with the memory technologies, and next generation DFI 5.0 is here to ensure higher performance in the systems using DDR5/LPDDR5. In this blog, we will discuss the new features of DFI 5.0 specification.
New applications like Cloud Computing, Artificial Intelligence, Autonomous cars, Augmented reality, Embedded vision are driving stricter requirements around memory performance and power efficiency. Memory is central to these systems, that require high bandwidth and speed along with lower power and lower cost. With these emerging market needs, the memory industry started to move from planar (2D) DRAMs to wide I/O or a 3D technology TSVs (Through Silicon Vertical interconnect access) such as HBM (high bandwidth memory). For more insight on HBM, read our blog “Next Generation Memory Technology for Graphics, Networking and HPC.” Low Power DRAM technology, evolved to the fifth-generation(LPDDR5) to deliver significant reduction in power and extremely high bandwidth as compared to LPDDR4. In this blog, we discuss LPDDR5 new features based on our understanding from collaboration with memory vendors and early adopters of Synopsys VIP over last 2 years.
SoC performance is a key competitive advantage in the marketplace, and the choice and configuration of protocol IP and interconnects is geared towards maximizing said performance. A case in point is the use of HBM (High Bandwidth Memory) technology and memory controllers. Currently in its third generation, HBM boasts of high-performance while using lesser power in a substantially smaller form factor than DDR. That said, how do teams ensure that the performance is delivered in the context of their SoC design?