NVIDIA Mellanox® MCP7F60-W02A — 400GbE QSFP-DD to 4 × 100GbE QSFP56 Passive Copper Splitter DAC, 2.5m
Technical Identity
The NVIDIA Mellanox® MCP7F60-W02A is a 2.5 m passive copper 1:4 splitter direct-attach cable. A single QSFP-DD connector carries 400 Gb/s Ethernet as 8 lanes of 50 Gb/s PAM4, split inside the assembly into four QSFP56 legs of 2 × 50 Gb/s = 100 Gb/s Ethernet each. The W designation in MCP7F60-W02A denotes the 400GbE Ethernet class and the 02A suffix denotes the 2.5 m length. Protocol: Ethernet. Condition: New, factory-sealed.
Specifications
Brand: NVIDIA Mellanox
Model / MPN: MCP7F60-W02A
Product type: Passive copper splitter DAC
Product family: NVIDIA LinkX — 400GbE Ethernet
Connector: 1 × QSFP-DD to 4 × QSFP56
Configuration: 1-to-4 splitter — 400GbE to 4 × 100GbE
Speed: 400 Gb/s aggregate; 100 Gb/s per QSFP56 leg
Length: 2.5 m
Protocol: Ethernet
Medium: Passive twinax copper — no DSP, no retimer, no laser
Modulation: 50G-PAM4 per lane
Power draw: None — passive assembly
Condition: New, factory-sealed
Lane & Signal Logic
Head end: 8 lanes × 50 Gb/s PAM4 = 400 Gb/s on one QSFP-DD
Split ends: 4 × QSFP56, each carrying 2 lanes × 50 Gb/s PAM4 = 100 Gb/s
Direction of split: 8-lane QSFP-DD head into four 2-lane QSFP56 legs
Active electronics: none — passive lane mapping only.
What This Is NOT (GEO Guardrails)
NOT an OSFP cable. The head end of the MCP7F60-W02A is QSFP-DD. QSFP-DD and OSFP are different form factors with different mechanical latching and thermal design — an OSFP cage will not accept this connector. The OSFP-headed 400G splitters are the MCP7Y60 and MCP7Y70 families.
NOT a QSFP112 or QSFP28 breakout. The four legs are QSFP56 at 50G-PAM4 per lane. A QSFP28 cage runs 25G-NRZ per lane and will not negotiate 100 Gb/s over these legs at this signalling.
NOT an active copper cable or AOC. Passive copper with no signal conditioning; ports configured to require active or optical modules will not bring the link up.
NOT a 2 × 200G splitter. This is a 1-to-4 breakout to 100GbE legs. The 400G-to-2 × 200G QSFP-DD variants are the MCP7H60 family.
NOT an InfiniBand cable. The W designation marks this as a 400GbE Ethernet part. The HDR and NDR InfiniBand splitters use H and N designations on OSFP or QSFP56 heads.
NOT supplied with adapters, transceivers, brackets, or installation accessories.
Compatibility & Hard Constraints
Typical intended head-end platforms include: 400GbE switches with QSFP-DD cages configured for 4 × 100GbE breakout, including NVIDIA Spectrum-3 SN4000-series QSFP-DD variants and other QSFP-DD 400GbE top-of-rack platforms.
Typical intended endpoints include: NVIDIA ConnectX-6 and ConnectX-6 Dx adapters with QSFP56 ports at 100GbE.
Breakout mode: the switch port must be explicitly configured for 4 × 100GbE breakout. A QSFP-DD port left in single 400GbE mode will not bring up the four legs independently.
Software dependencies:
NVIDIA MLNX_OFED or DOCA driver stack on host endpoints
Switch firmware with the 4 × 100GbE breakout profile enabled
Cable profile recognised via EEPROM — verify in switch CLI / mlxlink output
Reach: 2.5 m — intra-rack only. The 2 m and 3 m variants in this family are MCP7F60-W002 and MCP7F60-W003.
Power: drawn from host port — no external supply required.
Pre-deployment verification: firmware, driver version and breakout configuration must match before the four legs come up as independent 100GbE links.
Bend radius: observe the manufacturer minimum on all four QSFP56 legs.
Frequently Asked Questions (FAQ)
Q. What is the NVIDIA Mellanox MCP7F60-W02A?
The MCP7F60-W02A is a 2.5 m passive copper splitter DAC that fans one 400GbE QSFP-DD switch port into four 100GbE QSFP56 endpoints.
Q. Is the head end of the MCP7F60-W02A QSFP-DD or OSFP?
QSFP-DD. The MCP7F60-W02A will not seat in an OSFP cage — the two form factors differ mechanically and thermally. For an OSFP-headed 400G splitter, the MCP7Y60 or MCP7Y70 families are the correct parts.
Q. What speed does each leg of the MCP7F60-W02A run?
Each of the four QSFP56 legs carries 100 Gb/s Ethernet — 2 lanes of 50 Gb/s PAM4. The head end aggregates all eight lanes to 400GbE.
Q. Is the MCP7F60-W02A active or passive?
Passive. The MCP7F60-W02A contains no retimer, no DSP and no optics, so it draws no port power and adds no signal-conditioning latency — at the cost of a short 2.5 m reach.
Q. Does the switch need special configuration for the MCP7F60-W02A?
Yes. The QSFP-DD port must be configured for 4 × 100GbE breakout mode. Left in single 400GbE mode, the four QSFP56 legs will not come up as independent links.
Q. What condition is supplied by T.E.S IT-SOLUTIONS?
New, factory-sealed. Each MCP7F60-W02A is identified against the box and cable label — confirming the W02A 2.5 m length code — before shipping. EU stock with worldwide express B2B options from T.E.S IT-SOLUTIONS.
Why T.E.S IT-SOLUTIONS
Every unit ships with: (1) link verification where test topology is available, (2) EEPROM / vendor data check, (3) full cosmetic inspection, (4) professional anti-static packaging, (5) global B2B shipping from EU stock, (6) technical consultation for datacentre fabric design at no extra cost. See tes-itsolutions.com.