Q-Logic IB6054601-00 D Manuel D’Utilisation
E – Glossary
E-2
IB6054601-00 D
Q
GID
For Global Identifier. Used for routing between different
InfiniBand subnets.
InfiniBand subnets.
GUID
For Globally Unique Identifier for the InfiniPath chip.
Equivalent to Ethernet MAC address.
Equivalent to Ethernet MAC address.
head node
Same as front end node.
HCA
For Host Channel Adapter. HCAs are I/O engines
located within processing nodes, connecting them to
the InfiniBand fabric.
located within processing nodes, connecting them to
the InfiniBand fabric.
hosts file
Same as mpihosts file. Not the same as the /etc/hosts
file.
file.
HTX
A specification that defines a connector and form factor
for HyperTransport-enabled daughtercards and EATX
motherboards.
for HyperTransport-enabled daughtercards and EATX
motherboards.
InfiniBand
Also referred to as IB. An input/output architecture used
in high-end servers. It is also a specification for the
serial transmission of data between processors and I/O
devices. InfiniBand typically uses switched,
point-to-point channels. These channels are usually
created by attaching host channel adapters (HCAs) and
target channel adapters (TCAs) through InfiniBand
switches.
in high-end servers. It is also a specification for the
serial transmission of data between processors and I/O
devices. InfiniBand typically uses switched,
point-to-point channels. These channels are usually
created by attaching host channel adapters (HCAs) and
target channel adapters (TCAs) through InfiniBand
switches.
IPoIB
For Internet Protocol over InfiniBand, as per the
OpenFabrics standards effort. This protocol layer
allows the traditional Internet protocol (IP) to run over
an InfiniBand fabric.
OpenFabrics standards effort. This protocol layer
allows the traditional Internet protocol (IP) to run over
an InfiniBand fabric.
iSER
For iSCSI Extensions for RDMA. An upper layer
protocol.
protocol.
kDAPL
For kernel Direct Access Provider Library.kDAPL is the
kernel mode version of the DAPL protocol.
kernel mode version of the DAPL protocol.
latency
The delay inherent in processing network data. In terms
of MPI, it is the time required to send a message from
one node to another, independent of message size.
Latency can be further split into sender and receiver
processing overheads, as well as wire and switch
overhead.
of MPI, it is the time required to send a message from
one node to another, independent of message size.
Latency can be further split into sender and receiver
processing overheads, as well as wire and switch
overhead.
launch node
Same as front end node.
layered driver
A driver that does not directly manage any target
devices. The layered driver calls another driver’s
routines, which in turn manages the target devices.
devices. The layered driver calls another driver’s
routines, which in turn manages the target devices.