Intel Arrow Lake Processors Limit PCIe 5.0 NVMe SSD Speeds to 12GB/s Due to Latency Issues

Intel's Arrow Lake processors, part of the Core 200S series, have been identified as having a significant bottleneck on PCIe 5.0 NVMe SSDs. This issue manifests as a roughly 16% reduction in peak speed, capping the transfer rate at 12GB/s instead of the expected 14GB/s. The performance limitation is particularly pronounced on LGA 1851 motherboards, specifically affecting the M.2 storage ports. The discovery was made by a storage-focused review outlet during tests where a Samsung 9100 Pro and a Micron 4600 SSD, both capable of 14GB/s speeds, were only able to reach 12.3GB/s on Arrow Lake-based Z890 motherboards. In contrast, when the same SSDs were tested on previous-generation Z790 motherboards, they achieved their full potential of 14.3GB/s. Further confirmation came from testing the SSDs with an Asus Hyper M.2 add-in PCIe card, which bypassed the M.2 slots and allowed both drives to hit 14.3GB/s, proving the issue is specific to the physical M.2 slots on Z890 motherboards. The root cause of this bottleneck is attributed to the design changes in Arrow Lake processors. Intel's move from a monolithic die design to a chiplet-style architecture introduces additional latency through the I/O Extender component. The Arrow Lake-S CPUs feature 24 PCIe lanes: 16 PCIe 5.0 lanes for add-in graphics cards, 4 PCIe 5.0 lanes for SSDs, and 4 PCIe 4.0 lanes for other SSDs. Importantly, PCIe lanes 21 through 24, which are routed through the I/O Extender tile, experience increased latency and lower peak bandwidth. Intel acknowledged the issue in a statement, explaining that "the PCIe lanes 21 to 24 Gen5 root ports on Intel Core Ultra 200S series processors may exhibit increased latencies due to a longer die-to-die data path" and that performance variations depend on the workload and the PCIe endpoint device's capabilities. However, this latency is not confined to specific motherboard manufacturers; both Asus and ASRock have reported consistent results, further validating the problem. This bottleneck adds to the list of performance-related concerns with Arrow Lake processors. Intel's transition to a chiplet design has previously been noted to adversely affect gaming performance. For example, the Core Ultra 9 285K has been found to be slower than the previous generation Core i9-13900K and i9-14900K in gaming benchmarks. Despite firmware updates aimed at addressing these issues, the new processors still lag behind their Raptor Lake counterparts in gaming scenarios. Now, with the emergence of NVMe storage latency problems, Intel faces an increased burden to resolve these performance regressions. Industry insiders suggest that the issues stem from the inherent complexity and challenges of managing multi-chip designs, particularly with communication between different tiles. While firmware tweaks can mitigate some performance drops, the latency and bandwidth limitations associated with the I/O Extender are deeply rooted in the hardware design itself. Therefore, a complete fix might only come with Intel's next-generation processors. Evaluation and Company Profiles Experts in the tech industry highlight that Intel's move to a chiplet design is a strategic shift aimed at improving manufacturing efficiency and potentially enhancing performance in certain areas. However, the current issues with gaming and NVMe storage performance underscore the teething problems that often accompany such radical architectural changes. The company is known for its robust research and development capabilities and is expected to address these issues in future updates or subsequent processor generations. As for Intel, it remains one of the leading semiconductor companies in the world, renowned for its innovation in CPU technology. The Core 200S series is part of Intel's ambitious plan to reclaim leadership in the high-performance desktop market. Despite the current setbacks, Intel's history of overcoming technical hurdles suggests that they will continue to refine their chiplet approach, potentially yielding significant improvements in the future.