CXMT refutes Micron's arguments regarding 4F² DRAM density and announces first samples by year-end
Read more
Pandaily
pandaily.com

CXMT refutes Micron's arguments regarding 4F² DRAM density and announces first samples by year-end

ChangXin Memory Technologies (CXMT) has responded to Micron's assertion that the 4F² DRAM cell layout does not provide a real increase in density. According to a report published by CXMT for investors on October 31, the company's next technological platform, the sixth generation, is based on the 4F² architecture and has already passed engineering sample verification. The first products based on 4F² will reach customers by the end of this year, followed by mass production contingent upon customer validation and market demand.

The main dispute concerned the factors limiting bit density in DRAM. Scott DeBoer, President and Chief Technology and Product Officer at Micron, stated during the company's webinar on October 21 that Micron will continue to develop the 6F² path and focus on 3D DRAM. He noted that the company had studied 4F² for over twenty years but had not found a viable scaling path, calling some aspects of the 4F² discussion a 'smokescreen.' In DeBoer's view, the theoretical limit of density is determined by the size and pitch of the capacitor, not the transistor arrangement method.

CXMT did not dispute the physical principles but described increasing bit density as a complex systemic problem where word lines, bit lines, and capacitor structures compete for space. The company emphasized that for each architecture at a given node, there is a cost curve determined by process integration complexity, the number of lithography steps, and the thermal budget. CXMT stated that the choice between different approaches depends on the memory manufacturer's technological history, existing production line availability, patent portfolio, and cost structure, and such decisions deserve understanding and respect.

According to an analysis from November 1, Samsung and SK hynix are continuing to develop 4F². Samsung presented a 16 Gbit 4F² prototype at ISSCC 2026, which integrates vertical channel transistors with hybrid copper wafer-to-wafer interconnects. The company plans to use 4F² on its 10a, 10b, and 10c nodes before transitioning to 3D DRAM on the 10d node. Samsung Semiconductor CTO, Jong-Hyuk Son, reported on October 12 that work on 4F² is ongoing. SK hynix uses a vertical gate route, which, they claim, increases bit density by more than 20%, and tested read and write operations on test arrays in July, planning to present work on capacitive 4F² at IEDM in December.

CXMT's version utilizes a Vertical Surrounded Gate (VSG) type transistor, according to the analysis. The company claims to have been working on 4F² since its inception and possesses over 300 published patent families related to this technology. Since 4F² increases density by changing the cell layout rather than reducing line width, it reduces the area factor from 6 to 4 within the limits achievable by DUV lithography. CXMT announced that DDR5 RDIMM, combining 4F² with hybrid interconnect, will appear around the end of the year. The analysis warns that server modules have strict reliability requirements, so the first version will primarily serve for technology verification.

Similar stories

Access to pandaily.com website is restricted using Cloudflare
Read more
pandaily.com

Access to pandaily.com website is restricted using Cloudflare

The owner of the pandaily.com website has temporarily blocked user access to the resource. This restriction is implemented using the Cloudflare service.

For more detailed information about this situation, users can refer to the official Cloudflare documentation on HTTP status codes, specifically at the link https://developers.cloudflare.com/support/troubleshooting/http-status-codes/cloudflare-1xxx-errors/error-1015/.

Stars Micro System Unveils PCIe 6.0 Switch, 800G AI Network Card, and RAID Chip
Read more
pandaily.com

Stars Micro System Unveils PCIe 6.0 Switch, 800G AI Network Card, and RAID Chip

WuXi Stars Micro System Technologies has introduced three interconnect chips designed for artificial intelligence clusters and servers. The company, founded in 2020, held its first full-stack and brand product launch in Uxi on October 15, demonstrating 12 interconnect chips that are already in mass production alongside new components.

The centerpiece of the presentation was the C82-4G, which the company named the first PCIe 6.0 switch developed in China. It is capable of supporting up to 144 lanes at 64 GT/s per lane, doubling the per-lane speed of PCIe 5.0. The switch utilizes PAM4 signaling and FLIT mode, ensuring non-blocking switching with a latency in the range of 100 nanoseconds, as well as support for fabric networking, CXL memory expansion, and memory pooling. It is designed for internal connections in next-generation AI servers, where switches extend processor PCIe lanes to GPUs, network cards, and solid-state drives.

The second product is an 800G multi-protocol AI network card, which the company positions as the first such development from a Chinese supplier. This card incorporates a PCIe Gen6 switch with 48 lanes, supports 800G line-rate transmission with microsecond-level latency, offloads RoCE v2, and offers programmable congestion control. The company notes that the card is optimized for Hygon processors and next-generation supernode architectures focused on inter-node communication in clusters comprising tens of thousands of accelerators.

The third component presented is a multi-mode RAID chip, created using domestic intelligent capabilities. It supports SAS, SATA, and PCIe 5.0 connections, is compatible with both hard drives and SSDs, features a hardware encryption block, and, according to the company, increases throughput and IOPS by approximately eight times compared to the previous generation.

CEO Guo Erhui emphasized that computing power is a system of tightly coupled components: computation, networking, data storage, power, cooling, and planning. He noted that as clusters grow to tens of thousands of cards, interconnects become the bottleneck. The company reported that its products are already used in the internet, finance, telecommunications, and energy sectors, with stated failure rates for the chips being less than 100 DPPM, and for the boards being less than 500 DPPM.

During the launch, cooperation agreements were signed with Kylin Software, UnionTech, Tencent Cloud, Inspur, H3C, ZTE, and Lenovo. These agreements cover the adaptation of the company's interconnect chips to their operating systems and server platforms. The company stated that future steps include the mass production of the next-generation network card and PCIe 6.0 switch, as well as work on optical interconnects and storage pooling. It should be noted that all stated specifications are company claims and have not undergone independent verification.

Popular