In industrial scenarios such as machine vision, high-speed data acquisition, and edge AI inference, computing performance, scalability, and environmental adaptability are indispensable. The ADLINK MXC-6400 series, as a new member of the Matrix family, is equipped with the 6th generation Intel ® Core ™ The i7/i5/i3 processor (Skylake) and QM170 chipset support DDR4 memory, PCIe Gen2 expansion (1 x PCI+2 x PCIe x8 or 1 x PCI+1 x PCIe x16), three-way independent 4K display output, 6 x USB 3.0, 3 x GbE (including iAMT 11.0 and teaming), as well as dual front hot swappable SATA III hard drive bays and dual built-in SATA (supporting RAID 0/1/5/10), combined with a fanless wide temperature design of -20~70 ° C, making it a versatile player in the fields of industrial automation, railway transportation, and monitoring. However, high performance also brings new challenges such as heat dissipation planning, storage array configuration, and PCIe bandwidth allocation. This article is based on MXC-6400 technical data, and systematically summarizes the most concerned hardcore issues for engineers from hardware selection, deployment points, RAID tuning to typical troubleshooting.
MXC-6400 Architecture Highlights and Positioning
MXC-6400 offers three CPU options:
I7-6820EQ (quad core eight thread, 2.8GHz, TDP 45W)
I5-6440EQ (quad core quad threading, 2.7GHz, TDP 45W)
I3-6100E (dual core four thread, 2.7GHz, TDP 35W)
Compared to the previous generation MXC-6300 (3rd generation Core+DDR3+USB3.0), the main improvements are:
Memory: DDR4 SO-DIMM dual slot, maximum 32GB, frequency 2133 MHz, doubled bandwidth.
Storage flexibility: Two front hot swappable SATA III (6Gb/s) disk slots, allowing for non-stop storage replacement (suitable for data logging applications); There are also two built-in SATA ports that support RAID 0/1/5/10 (requiring four disks but only four ports, and can be configured with four disks but only four ports for RAID 10).
Expansion slot: Optional configurations include 1 x PCI+2 x PCIe x8, or 1 x PCI+1 x PCIe x16. PCIe x16 can be used for high-performance GPUs or four port network cards, while x8 can be used for acquisition cards. Note that PCIe is Gen2 (5GT/s).
Display: 2 x DisplayPort+1 x DVI-I, supports 4K@60Hz (DP), Three independent appearances.
USB: 6 x USB 3.0 front/rear, with internal USB 2.0 plug (for encryption dongle).
Network port: 3 × GbE (Intel i219LM+i210AT? Specific details not specified, but supports iAMT 11.0 and teaming).
Expansion: 2 x internal mPCIe (including USIM for 3G/4G).
Optional isolation DIO: 16CH isolation DI/DO, with digital filtering.
Remote switch: The front panel provides a remote power switch connector for easy cabinet wiring.
Management: Built in SEMA 3.0 (ADLINK system monitoring API).
This model is positioned as a high-performance fanless expansion type, suitable for scenarios that require multi card collaboration (such as motion control+image acquisition) and high reliability (hot swappable disks, RAID).
Expansion slot configuration and PCIe bandwidth budget
MXC-6400 offers two expansion backplane options (factory fixed):
Configuration A: 1 x PCI (32-bit/33MHz)+2 x PCIe x8 (physical x8, electrical x8)
Configuration B: 1 x PCI+1 x PCIe x16 (electrical x16)
Bandwidth considerations:
PCIe Gen2 x8 has a theoretical bandwidth of 4GB/s (unidirectional), while x16 has a bandwidth of 8GB/s. If inserting a GPU (such as NVIDIA Quadro P1000), it is recommended to choose an x16 configuration for optimal performance.
If two high-speed acquisition cards (such as Camera Link or CoaXpress) need to be inserted simultaneously, an x8+x8 configuration can be selected, with each card having an independent x8 link.
Attention: The number of PCIe channels directly connected to the CPU is limited (QM170 provides 16 PCIe 3.0 channels, but the chipset goes down to PCIe 2.0), and actual allocation needs to refer to the specific design. It is recommended to prioritize inserting high bandwidth cards into slots near the CPU (usually x16 slots).
PCI slot precautions: PCI is a traditional 5V/3.3V universal card that can be used to insert old-fashioned motion control cards, but be aware of its IRQ sharing to avoid conflicts with PCIe cards (which can be adjusted through BIOS).
Practical Application of Storage Array (RAID)
Four SATA interfaces (2 hot swappable front ports+2 built-in ports) can be used to build RAID through Intel RST (Fast Storage Technology):
RAID 0: Striped, improves read and write speed (suitable for real-time caching), but without redundancy.
RAID 1: Mirroring, protecting data (suitable for system disks).
RAID 5: At least 3 disks with parity check, balancing performance and redundancy.
RAID 10: Requires 4 disks, mirror+stripe, achieving both performance and redundancy.
Hot swappable disk slots: The front two trays support 2.5-inch SATA SSD/HDD, which can be unplugged during system operation (requiring software uninstallation first). This is very useful for replacing faulty disks or exporting data without stopping the machine. However, please note that hot swapping in RAID arrays must follow the controller rules (such as marking as Failed before unplugging).