PCIe Gen4 x4 U.2 and U.3, single (1×4)
and dual-port (2×2) interposer with SI-Fi technology
Coupled with the recently announced Kodiak PCIe/NVMe Gen4 analysis system, the SI-Fi interposers solve complex issues relating to signal quality and system transparency and eliminate the need for cumbersome, time-intensive calibration and tuning routines. Interposers are designed test adapters, or probes, that are electrically and physically placed between components of a communications system to intercept and relay high-speed signaling and discrete data lines for analysis purposes.
“Driven by the need for ever-faster data transfers, PCIe signaling has become exceptionally complex in design and difficult to monitor unobtrusively. Signal conditioning methods used for PCIe Gen1 and Gen2 now seem primitive compared to the complex approaches used for PCIe Gen3 and Gen4. Further challenges are presented by NVMe, which adds critical requirements like hot-plug and NVM Subsystem Reset (NSSR), where the PCIe signals are renegotiated,” said Paul Mutschler, CEO, SerialTek. “Our SI-Fi technology directly meets and overcomes these challenges and gives our customers the features and capabilities they need to do their work efficiently.“
Transparency in probe’s design is key
With the company’s proprietary SI-Fi interposer technology, the transmitter threshold and pre-emphasis from one link partner reach the receiver of the other link partner so the link can be properly trained to optimum conditions, making the interposer as transparent as possible. At the core of this technology is a specialized linea amplifier design where PCIe analog signaling is received at a differential input and distributed to two separate phase-matched differential outputs with a nominal, idealized gain of 0dB. This approach results in setup of the analyzer and product under test and avoids a variety of limitations inherent to other probing approaches where link training sequences don’t go through the interposer.
No need for calibration
Competing PCIe Gen4 analyzers and interposers require tuning, or calibration, which means they cannot transparently and reliably support modern PCIe link training, since it can occur, not just at boot-up. With SI-Fi technology, users can save hours over these legacy approaches. This technology offers high-quality signal integrity, even over changing conditions. This expands and enables coverage in critical test areas, including link training (LTSSM), power management, hot plug, reset and other situations where the physical link/lane characteristics may change.
SI-Fi PCIe Gen4 Interposers
Multiple form factor support
SI-Fi interposer form factors include AIC (x4), M.2 (x4), U.2 (x4) and U.3 (x4). Additionally, U.2, U.3, single-port (1×4) and dual-port (2×2) analysis can be combined into one interposer unit, providing cost savings in enterprise environments where all form factors are required. SI-Fi interposers also support all relevant sidebands, including SMBus (e.g., NVMe-MI) from the host or from external/third-party injection or generation tools.
Kodiak PCIe/NVMe Gen4 analysis system
Kodiak platform complements SI-Fi technology with processing architecture
The company’s SI-Fi interposers work in conjunction with its announced Kodiak PCIe/NVMe Gen4 analysis system. This platform introduces performance-oriented innovations made possible by an embedded architecture that breaks free from legacy data upload practices in favor of advanced, optimized embedded data processing, fully reimagined by the firm’s engineering to handle latest- and future-generation PCIe and NVMe technologies. Interface responsiveness is markedly advanced, searches involving massive amounts of data are fast, and hardware filtering is flexible and powerful.
SI-Fi interposers are available for purchase, with shipments six weeks from order placement. Various configurations are provided to meet a variety of customer price and feature requirements.
SerialTek: Kodiak PCIe/NVMe Analysis System
Includes several performance-oriented innovations, made possible by embedded architecture that breaks free from status quo legacy data upload practices in favor of advanced, optimized embedded data processing.
March 13, 2019 | Press Release