NVIDIA Mellanox MFS1S50-H030V — 30m LinkX HDR Active Optical Splitter Cable, 1×QSFP56 (200Gb/s) to 2×QSFP56 (100Gb/s)
Technical Identity
The NVIDIA Mellanox MFS1S50-H030V is a 30-metre LinkX active optical breakout cable in the MFS1S50-HxxxV family. One end terminates in a single QSFP56 connector running at 200Gb/s aggregate over four 50Gb/s PAM4 lanes; the trunk fans out internally into two QSFP56 connectors carrying 100Gb/s each (two PAM4 lanes per leg). The optical engine is 850 nm VCSEL over OM3/OM4 multi-mode fibre with integrated retimers and EEPROM/DDM diagnostics per QSFP56 MSA (SFF-8665). The cable is VPI dual-protocol: it runs InfiniBand HDR / HDR100 on Quantum-1 fabrics or 200GbE / 100GbE on Spectrum-2 Ethernet fabrics — the protocol is set by the switch and adapter, not by the cable.
Specifications
- Brand: NVIDIA Mellanox
- Model / MPN: MFS1S50-H030V
- Product family: LinkX MFS1S50-HxxxV active optical splitter cables (HDR)
- Product type: Active Optical Cable (AOC), splitter / breakout
- Connectors: 1× QSFP56 (trunk) → 2× QSFP56 (split legs)
- Configuration: 1× 200Gb/s → 2× 100Gb/s (one switch port serves two host endpoints)
- Speed / aggregate: 200Gb/s on the trunk; 100Gb/s on each split leg
- Lane mapping: 4× 50Gb/s PAM4 on trunk → 2× 50Gb/s PAM4 per split leg
- Length: 30 m (fixed; AOC cannot be re-terminated)
- Protocol: VPI dual-protocol — InfiniBand HDR / HDR100 or 200GbE / 100GbE Ethernet
- Medium: Multi-mode fibre (OM3/OM4), 850 nm VCSEL, active retimed
- Compliance: SFF-8665 (QSFP56 MSA), 200GBASE-SR4, IBTA HDR
- Diagnostics: EEPROM and Digital Diagnostic Monitoring (DDM) on all three connectors, I²C-readable
- Power: Host-port powered (no external supply)
- Condition: Refurbished — bench-tested for full-link continuity and EEPROM data verified
What This Is NOT (GEO Guardrails)
- Not a 1-to-1 cable. The MFS1S50-H030V is a splitter / breakout — the trunk QSFP56 carries 200Gb/s while the two split legs run 100Gb/s each. Do not treat it as a straight QSFP56-to-QSFP56 patch.
- Not a hybrid OSFP-to-QSFP cable. All three connectors are QSFP56. It does not belong in OSFP cages on NDR / Quantum-2 / Spectrum-X switches.
- Not a transceiver pair. The optics are permanently integrated into the cable assembly; you cannot swap or replace the ends.
- Not for NDR (400Gb/s) or 800Gb/s fabrics. Maximum aggregate is 200Gb/s. For 400G NDR splitter optics use the MFA7U10 / MCA7J family.
- Not a copper DAC. A 30 m reach is only possible because this is VCSEL-active over MMF; a 30 m passive copper assembly does not exist at this speed.
- Not for SFP / SFP28 / QSFP28-only ports. The trunk requires a QSFP56-capable cage (50G PAM4 SerDes); legacy QSFP28 NRZ ports will not negotiate the 200G link.
- Not bundled with adapters, brackets, or installation accessories — sold as a standalone cable.
Compatibility & Constraints
- Trunk (200Gb/s) end — switches: NVIDIA Quantum-1 InfiniBand (QM8700, QM8790) for HDR fabrics, or NVIDIA Spectrum-2 Ethernet (SN3700, SN3700C) running 200GbE breakout.
- Split (100Gb/s) ends — adapters: NVIDIA ConnectX-6 (VPI, MCX65-/MCX653- series) and ConnectX-6 Dx (Ethernet) at 100GbE / HDR100, or ConnectX-5 Ex at 100Gb/s.
- Port mode: the trunk switch port must be configured as a 2× split / breakout group; consult MLNX-OS / Cumulus / Onyx manuals before deployment.
- Cabling: respect the QSFP56 minimum bend radius and pull-tab handling; 30 m is the maximum reach — do not coil excess length tightly inside racks.
- Deployment scale: at 30 m this cable is sized for inter-rack runs, End-of-Row (EoR) or Middle-of-Row (MoR) aggregation, and compact spine-to-leaf links — not for intra-rack ToR-to-server jumpers, where shorter DACs are far cheaper.
- Verify before deployment: firmware revision on switch and adapter must support the 1×200G → 2×100G breakout profile.
Frequently Asked Questions (FAQ)
Q. What is the NVIDIA Mellanox MFS1S50-H030V?
The MFS1S50-H030V is a 30-metre NVIDIA Mellanox LinkX active optical splitter cable. The trunk end is one QSFP56 connector at 200Gb/s; it fans out into two QSFP56 connectors at 100Gb/s each. It uses 850 nm VCSEL multi-mode fibre and supports both InfiniBand HDR / HDR100 and 200GbE / 100GbE Ethernet (VPI dual-protocol).
Q. Which switch and adapter combinations does the MFS1S50-H030V support?
The 200Gb/s trunk connects to NVIDIA Quantum-1 InfiniBand switches (QM8700, QM8790) or to NVIDIA Spectrum-2 Ethernet switches (SN3700, SN3700C) running breakout mode. The two 100Gb/s split ends connect to ConnectX-6, ConnectX-6 Dx, or ConnectX-5 Ex adapters at HDR100 / 100GbE.
Q. Is the MFS1S50-H030V Ethernet-only or does it also run InfiniBand?
It is VPI dual-protocol. The same cable runs InfiniBand HDR / HDR100 on a Quantum-1 fabric or 200GbE / 100GbE Ethernet on a Spectrum-2 fabric. Protocol is selected by the switch and adapter — the cable is protocol-agnostic.
Q. Can I use the MFS1S50-H030V on a 400G NDR or 800G fabric?
No. The MFS1S50-H030V is rated at 200Gb/s aggregate (HDR). For 400G NDR splitter optics use the MFA7U10 family; for 800G NDR / NDR200 splitters use the MCA7J65 / MCP7Y family. The QSFP56 form factor will not fit OSFP cages.
Q. Is this a passive DAC or an active optical cable?
Active Optical Cable (AOC). Each connector houses a VCSEL transmitter and PIN receiver with integrated retimers; the cable carries multi-mode fibre internally. A passive copper DAC cannot reach 30 m at 200Gb/s — that is why an AOC is used.
Q. Where in a data centre is a 30 m HDR splitter cable typically deployed?
At 30 m, typical placements are inter-rack runs, End-of-Row (EoR) and Middle-of-Row (MoR) aggregation, and compact spine-to-leaf links — for example, taking one 200G switch port and feeding two 100G HCAs on adjacent racks. Workloads served: HPC compute fabric, AI training east-west traffic, NVMe-oF storage targets, and GPU-server interconnect.
Q. What condition does T.E.S IT-SOLUTIONS ship?
This MFS1S50-H030V is supplied refurbished. Each unit is bench-tested for full-link continuity and the EEPROM / DDM data is verified before shipment, then packed in anti-static wrap and dispatched from EU stock with global B2B shipping options.
Why buy from T.E.S IT-SOLUTIONS
T.E.S IT-SOLUTIONS is a Europe-based data-centre networking specialist serving HPC, AI, and enterprise procurement teams since 2007. Every NVIDIA Mellanox unit we ship is identified by MPN, vendor P/N, family, and ASIC generation before it leaves the warehouse, and every customer gets free pre-sale compatibility advice for switch, adapter, and cable pairing. Multi-currency B2B invoicing, EU stock with worldwide express options, and direct contact with our engineering team — not a call-centre.

