|
This is Hardware page for
ClusterGate.RU
|
- GPU General-Purpose Computation Using Graphics
Hardware.
GPGPU stands for General-Purpose computation on GPUs. With the increasing
programmability of commodity graphics processing units (GPUs), these chips are
capable of performing more than the specific graphics computations for which
they were designed. They are now capable coprocessors, and their high speed
makes them useful for a variety of applications. The goal of this page is to
catalog the current and historical use of GPUs for general-purpose
computation.
- Myricom - home for Myrinet Myrinet is a
cost-effective, high-performance, packet-communication and switching
technology that is widely used to interconnect clusters of workstations, PCs,
servers, blade servers, or single-board computers. Clusters provide an
economical way of achieving:
- high performance, by distributing demanding computations across an
array of cost-effective hosts. For "tightly coupled" distributed
omputations, the interconnect must provide high-data-rate and
low-latency communication between host processes.
- high availability, by allowing a computation to proceed with a subset of
the hosts. The interconnect should be capable of detecting and isolating
aults, and of using alternative communication paths.
Conventional networks such as Ethernet can be used to build clusters, but
do not provide the performance or features required for high-performance or
high-availability clustering. Characteristics that distinguish Myrinet from
other networks include:
- Flow control, error control, and "heartbeat" continuity monitoring on
every link.
- Low-latency, cut-through switches, with monitoring for high-availability
applications.
- Switch networks that can scale to tens of thousands of hosts, and that
can also provide alternative communication paths between hosts.
- Network Interface Cards (NICs) that execute firmware to offload protocol
processing from the host computer. The firmware interacts directly with
host processes ("OS bypass") for low-latency communication, and directly
with the network to send, receive, and buffer packets.
- Silver Storm Technologies - Infiniband switches
InfiniBand utilizes high-speed serial technology to interconnect data center
elements. Each link consists of a signal pair that transfers information in
both directions simultaneously at 2.5Gb/s. To increase throughput even more,
four or twelve pairs can be implemented, yielding transfer rates of 312.5MB/s,
12.5GB/s, or 37.5GB/s.
Unlike other networking technologies, InfiniBand permits local and remote
application software entities to perform memory-to-memory transfers without
the intervention of the processor. This frees up most of the processor.s
bandwidth to deal with computational problems. The architecture provides
increased reliability, better sharing of data, and built-in security.
- The InfiniBand -- Linux InfiniBand
Project. The InfiniBand. Architecture (IBA) is an industry standard that
defines a new high-speed switched fabric subsystem designed to connect
processor nodes and I/O nodes to form a system area network. This new
interconnect method moves away from the local transaction-based I/O model
across busses to a remote message-passing model across channels. The
architecture is independent of the host operating system (OS) and the
processor platform.
IBA provides both reliable and unreliable transport mechanisms in which
messages are enqueued for delivery between end systems. Hardware transport
protocols are defined that support reliable and unreliable messaging
(send/receive), and memory manipulation semantics (e.g., RDMA read/write)
without software intervention in the data transfer path.
The InfiniBand specification primarily defines the hardware electrical,
mechanical, link-level, and management aspects of an InfiniBand fabric, but
does not define the lowest layers of the operating system stack needed to
communicate over an InfiniBand fabric. The remainder of the operating system
stack to support storage, networking, IPC, and systems management is left to
the operating system vendor for definition. More on the InfiniBand
architecture can be found here.
- InfiniBand Trade association INTERCONNECT OF
CHOICE FOR HPC AND DATA CENTER
InfiniBand is a high performance, switched fabric interconnect standard for
servers. The technology is deployed worldwide in server clusters ranging from
two to thousands of nodes. From Prudential Financial to Sandia National
Laboratories, InfiniBand has become the standard interconnect of choice for
HPC environments and is quickly becoming the preferred standard in high
performance, enterprise data centers.
Founded in 1999, the InfiniBand Trade Association (IBTA) is comprised of
leading enterprise IT vendors including Agilent, Dell, Hewlett-Packard, IBM,
SilverStorm, Intel, Mellanox, Network Appliance, Oracle, Sun, Topspin and
Voltaire. The organization completed its first specification in October 2000.
In the past 12 months all major system vendors have announced InfiniBand
products and hundreds of products have completed interoperability testing and
are commercially available.
More recently the IBTA announced an effort to extend the technology.s
signaling rate beyond its current 30Gbps limitation to 120Gbps, maintaining
InfiniBand.s leadership as the high performance standard interconnect.
- Introduction to InfiniBand for IBM eServer pSeries Servers (PDF) IBM RedPaper
- iSCSI: SCSI is Internet SCSI (Small Computer System Interface), an Internet
Protocol (IP)-based storage networking standard for linking data storage
facilities, developed by the Internet Engineering Task Force (IETF). By
carrying SCSI commands over IP networks, iSCSI is used to facilitate data
transfers over intranets and to manage storage over long distances. The iSCSI
protocol is among the key technologies expected to help bring about rapid
development of the storage area network (SAN) market, by increasing the
capabilities and performance of storage data transmission. Because of the
ubiquity of IP networks, iSCSI can be used to transmit data over local area
networks (LANs), wide area networks (WANs), or the Internet and can enable
location-independent data storage and retrieval.
- Linux Ethernet-Howto
Frequently Asked Questions, Performance Tips, and many other
topics
- Hardware portal in Russia of course many topics
in Russian
- ASUS Corporation Motherboards
- SUPERMICRO Motherboards
- TYAN Motherboards
- Tom's PC Hardware Guide - useful info on hardware
- System Optimization and PC Performance (Hardware + software)
- Microprocessor resources - mainly about Intel. On-line
Intel documentation, news about Intel microprocessors, etc.
- Foundry networks network products and technologies
- KVM This is one of the KVM implementations. A KVM switch (with KVM being an
acronym (acronym: A word formed from the initial letters of a multi-word name)
for Keyboard, Video, Mouse) is a hardware device that allows a user to control
multiple computer (computer: A machine for performing calculations
automatically) s from a single keyboard, video monitor
and mouse (mouse: Any of numerous small
rodents typically resembling diminutive rats having pointed snouts and small
ears on elongated bodies with slender usually hairless tails) . Although
multiple computers are connected to the KVM, only one can be controlled at a
time. Modern devices have also added the ability to share USB
devices and speakers with multiple computers.
A user connects their monitor, keyboard and mouse to the KVM device instead of
directly to the computers themselves and then uses special cables to connect
the KVM device to the computers. Control is switched from one computer to
another by the use of buttons on the KVM device, with the KVM passing the
signals between the computers and the keyboard, mouse and monitor depending on
which computer is currently selected. Most devices also allow control to be
switched through keyboard commands (such as hitting a certain key, often
Scroll Lock, rapidly two or three times). Some KVM devices also send signals
to the computers that are not currently selected to ensure that they do not
think that the keyboard, mouse and monitor are dis-connected.
-
- KVM and KVM Over IP KVM over IP
devices use a dedicated microcontroller and potentially specialized video
capture hardware to capture the video signals keyboard and mouse signals,
compress and packetize them, and send them over an Ethernet link to a remote
console application that unpacks and reconstitutes the dynamic graphical
image. This KVM over IP subsystem is typically connected to a system's standby
power plane so that it's available during the entire BIOS boot process. These
devices allow multiple computers to be controlled remotely across a wide area
network, local area network or telephone-line using the TCP/IP protocols.
There are performance issues related with LAN/WAN hardware, standard protocols
and network latency so user management is commonly referred to as "near real
time". And, remote KVM over IP devices offer much smaller matrix frameworks.
|
|