News
FORESEE launches first PCIe Gen 4×4 SSD XP2100
At the end of 2021, with the support of more CPUs and controllers in the PC OEM market, mainstream SSDs will gradually transition from PCIe Gen 3×4 to PCIe Gen 4×4, especially in high-performance PC applications.
The performance of the update iteration is particularly obvious. This year, some first-tier PC OEM manufacturers have started to stop the introduction of PCIe 3.0 platforms and gradually turned to the “main battlefield” of PCIe Gen4×4.
Join tip3x on Telegram
Recently, the FORESEE SSD team conformed to the mainstream trend of the industry, actively improved the “market compatibility” of the SSD product line, and successfully developed its first PCIe Gen 4×4 SSD, the FORESEE XP2100 PCIe SSD.
Moreover, the FORESEE XP2100 PCIe SSD adopts a DRAM-less architecture, with a capacity ranging from 256GB to 1TB, complies with the mainstream NVMe 1.4 protocol in the market, and has passed the UNH-IOL certification.
Looking back at the PCIe standard, since PCIe 3.0 was introduced in 2010, it has gradually become popular in PCs and servers to meet the performance requirements of high-speed storage devices.
The maximum transmission speed of PCIe 3.0 can reach 8GT / SX 4=4GB / s, and the PCIe 4.0 formulated by the PCI-SIG organization in 2017, compared with PCIe 3.0, the data transmission speed has increased from 1GB / s per channel to 2GB / per channel providing users with a total data transfer speed of 32GB/s in a 16-lane configuration.
Furthermore, PCIe 4.0 can deliver 16GT/s per lane compared to the previous generation’s 8GT/s. In addition to different grades and speeds, different capacities, hosts, and other designs provide PC OEMs with more options when purchasing storage components.
One of the main points of current SSD products is often overlooked by users – the NVMe protocol standard. The latest NVMe 1.4 version was released in June 2019. One of the most important new features is I/O Determinism, including NVMe Set these two parts of PLM provide technical support for professional needs such as commercial PCs, servers, and data centers.
NVMe Set improves the division of labor during writing, which is equivalent to “dividing a large hard disk into multiple small hard disks”, drawing clear boundaries and not disturbing each other, preventing IO delays from causing performance fluctuations.
Taking a 4TB SSD as an example, the NVMe 1.3 protocol is used. Even in the case of 4 channels, it will only be regarded as a single 4TB space. In any writing space, congestion between channels is prone to occur.
Naturally will affect performance. On the basis of NVMe 1.4, 4TB will be divided into four 1TB spaces, and each channel will be independently controlled, which can effectively reduce latency.
The advantage of PLM (Predictable Latency Mode) is that it provides memory-level speed and low latency. Even if the power is turned off, the data on it can be retained, thereby improving the QoS of the system.
Besides, PLM divides the system into DTWIN (Deterministic Window) and NDWIN (Non-Deterministic Window). In DTWIN, SSD provides Deterministic latency for reading and writing instructions to improve system QoS. Under NDWIN, SSD does not need to provide Deterministic latency. In order to complete operations such as GC or Trim inside the SSD.