Introduction
Product Overview
Product Overview
Gigabit Ethernet is a high-performance Ethernet networking solution for HP servers and workstations.
The LAN adapters are data link adapters that support the IEEE 802.3z standard for 1000Base-SX operation over multimode fiber, and the IEEE 802.3ab standard for 1000Base-T operation over 4-pair Cat-5 or Cat-5E UTP copper wiring.
The Gigabit Ethernet intelligent adapters are designed to maximize host CPU efficiency by performing functions such as TCP/IP checksum, interrupt coalescing, and byte swapping. Bulk data transfers can be optimized with the use of Jumbo Frames; the large 9000 byte maximum transfer unit (MTU) improves system efficiency.
1000Base-T brings high bandwidth with 10, 100, and 1000 Mbps speeds, whch makes more processing power available for applications. The tri-speed adapter supports autonegotiation and autosensing. It operates in full-duplex mode at 10, 100, and 1000 Mbps or in half-duplex mode at 10 and 100 Mbps.
1000Base-SX supports full-duplex point-to-point or back-to-back (via switch to adapter or adapter to adapter) operations only. The adapters do not support half-duplex and do not speed negotiate (1000 Mbps operation only), but do perform autonegotiation for other link parameters.
A6847A supports PCI only. Older 1000Base-SX adapters support HSC/PCI. 1000Base-T supports PCI only.
The PCI adapters require a single slot in the host system and support all PCI bus configurations: 32-bit, 33/66 MHz and 64-bit, 33/66 MHz. (Note: A6794A, A6847A, and A6825A support 64-bit only.)
Features
Features of Gigabit Ethernet include:
•1000Base-T features:
—Triple speed 10/100/1000 Mbps
—Full duplex operation at 10/100/1000 Mbps and half-duplex operation at 10/100 Mbps (no half-duplex support for 1000 Mbps)
—Autonegotiation and autosensing to the highest available link speed
•1000Base-SX features:
—1000 Mbps
—Full-duplex operation
•Supports IEEE 802.1Q VLAN tagging and IEEE 802.1p Priority Queuing (HP-UX 11i)
•Minimized host CPU utilization
—Protocol offloading through on-board TCP, UDP, and IP checksum calculations
—Adaptive interrupt coalescence based on traffic load
