© ROOT-NATION.com - Use of content is permitted with a backlink.
Modern mobile phones have encountered serious challenges in the area of memory architecture, and this obstacle cannot be overcome by a simple operating system or software update. As many users know, developers of portable devices are currently actively trying to integrate large AI models directly into the devices’ hardware.
Fortunately, modern mobile chipsets are powerful enough to support the operation of such large language models locally without relying on external servers. However, current LPDDR-standard RAM technologies have significant limitations regarding the speed at which they can deliver the necessary data streams to these computing platforms. For this very reason, large data centers specializing in cloud AI rely on HBM architecture, which is known as high-bandwidth memory.
See also: AERONAUT – everything that flies above the ground: aviation, UAVs and drones, rockets, and space
According to recent analytical reports, to address this pressing issue, active development is currently underway on a completely new type of memory called LLW (Low Latency Wide DRAM). It is claimed that when developing LLW, the designers drew inspiration from HBM architecture, which delivers tremendous speed due to the close proximity of multiple memory layers to the central processing unit and their connection via an extremely wide interface. The result of this approach is a dramatic increase in bandwidth, which is 10 to 15 times higher than that of the best modern LPDDR modules.
At the same time, the HBM architecture requires highly complex engineering solutions for component layout and cooling system design, making its use inside the compact enclosures of smartphones completely impractical. The new LLW standard is expected to address this challenge by providing developers with significantly higher bandwidth and substantially lower data transfer latencies compared to traditional LPDDR memory. At the same time, the new technology will not impose critical limitations in terms of heat dissipation or space constraints, which typically accompany multi-layer vertical chip designs.
Despite all the optimistic forecasts, such claims should currently be viewed with a healthy dose of skepticism. All information currently available in the public domain comes from various industry bloggers and leak sources, rather than from official press releases or statements by representatives of manufacturing companies. In this context, a well-known insider going by the pseudonym Fixed Focus Digital reports that the implementation of LLW will allow for approximately a 50% reduction in power consumption while increasing overall performance by about 1.5 times.
As for when this technology will hit the commercial market, the source indicates that it will take several more years before LLW is widely deployed and integrated into consumer electronics. More specifically, the large-scale appearance of new chips in finished devices is not expected until at least the second half of 2027. Among the potential pioneers who will be the first to adopt this technology, insiders single out Xiaomi and Huawei. Despite these rumors, neither of these companies has yet officially participated in public discussions regarding the aforementioned memory standard.
See also:
- Commodore’s Spectacular Comeback: Introducing the Callback 8020 Flip Phone, Free of Browsers and Ads
- Everything About Project Helix: Xbox’s Last Big Bet or a Way Out of the Memory Crisis