Marvell Technology Group
Marvell Technology Group, Limited, is a producer of storage and consumer semiconductor products. The company has over 3,700 employees. Marvell's U. S. operating headquarters is located in Santa Clara and the company operates design centers in Europe, India and China. Marvell is a "fabless" manufacturer of semiconductors that ships more than one billion integrated circuits per year, its market segments include data center, enterprise / campus, automotive and home / consumer. Marvell was founded in 1995 by Sehat Sutardja, his wife Weili Dai, brother Pantas Sutardja; the initial public offering on June 27, 2000 raised $90 million, with the stock listed on NASDAQ with the symbol MRVL. After raising from $19 to over $63 per share, three days it was $55.25. At the time, the five largest customers, Samsung Electronics, Seagate Technology and Toshiba, accounted for 97% of sales; the shares dropped in December when insiders were allowed to sell. In July 2018, Marvell completed its acquisition of Cavium, Inc. strengthening its storage, networking, wireless connectivity and security product portfolios for the infrastructure market.
On the same day, Marvell announced the appointment of Syed Ali, Brad Buss and Dr. Edward Frank to the Marvell Board of Directors. In the summer of 2018, Marvell became the first silicon vendor in North America to open a CISPR 25 qualified automotive electromagnetic compatibility lab with the in-house capability to perform a wide range of emission, immunity and ESD tests to further drive the development of industry-leading automotive connectivity solutions; the company is headquartered in Hamilton, Bermuda. The US operations known as Marvell Semiconductor, are located in Silicon Valley, California. Through the years, Marvell acquired smaller companies to enter new markets. Marvell's first products were sold for computer data storage devices. In March 2000, computer networking products for the Ethernet family were first shipped. In October 2002, the Yukon brand Gigabit Ethernet controller was announced. On June 27, 2006, the sale of Intel's XScale assets was announced. Intel agreed to sell the XScale business to Marvell for an estimated USD 600 million in cash and the assumption of unspecified liabilities.
The acquisition was completed on November 9, 2006. In 2009, Marvell announced that the SheevaPlug, a small, low-power, SoC-based ARM architecture computer, would be released with full schematics. Marvell supplied the Wi-Fi chip for the original Apple iPhone. Marvell Mobile Hotspot is an in-car Wi-Fi connectivity; the 2010 Audi A8 was the first automobile in the market to feature a factory-installed MMH. Google's Chromecast products are powered by Marvell SoCs. Namely the Marvell ARMADA 1500 Mini SoC for the Chromecast 1st gen and Marvell ARMADA 1500 Mini Plus SoC for the Chromecast 2nd gen & Chromecast audio. Synaptics acquired Marvell Multimedia Solutions on 2017-06-12 ARMADA 1500 SoC's are now produced under different names In 2012, Marvell was named one of Thomson Reuters top 100 global innovators. In 2006, the US Securities and Exchange Commission started an inquiry on the company's stock option grant practices. An investigation determined "grant dates were chosen with the benefit of hindsight" to make the options more valuable.
The press estimated that the founders and other executives had made $760 million in gains from the options, which were awarded by the founding couple, Sehat Sutardja and Weili Dai. The SEC asked to interview the company general counsel Matthew Gloss, but Marvell claimed attorney-client privilege. Gloss was fired just before the investigation results were announced in May 2007. Abraham David Sofaer was hired to investigate the investigation after Gloss alleged it was not independent. In announcing the results of its own inquiry, the SEC did not give Marvell the credit granted other companies in the options scandal for cooperating with the SEC’s investigation or for cleaning up. At the time of the announcement, the co-acting regional director of the SEC’s San Francisco office stated, among other things, that the SEC did not believe that the lack of cooperation and remediation shown by Marvell merited a whole lot of credit in terms of giving Marvell a break. In announcing its results, the SEC found that Gloss was not a participant in Dai and Sutardja’s backdating scheme.
Marvell restated its financial results, stated that Dai will no longer be executive vice president, chief operating officer, a director but continue with the company in a non-management position. The company agreed to pay a $10 million fine in 2008, but did not fire Dai nor replace Sutardja as chairman as stated by the investigating committee. In December 2012, a Pittsburgh jury ruled that Marvell had infringed two patents by incorporating hard disk technology developed and owned by Carnegie Mellon University without a license; the technology, relating to improving hard disk data read accuracy at high speeds, was reported to have been used in 2.3 billion chips sold by Marvell between 2003 and 2012. The jury awarded damages of $1.17 billion, the third largest in a patent case at the time. The jury found that the breach had been "willful", giving the judge discretion to award up to three times the original damage amount. In December 2
Token Ring local area network technology is a communications protocol for local area networks. It uses a special three-byte frame called a "token" that travels around a logical "ring" of workstations or servers; this token passing is a channel access method providing fair access for all stations, eliminating the collisions of contention-based access methods. Introduced by IBM in 1984, it was standardized with protocol IEEE 802.5 and was successful in corporate environments, but eclipsed by the versions of Ethernet. A wide range of different local area network technologies were developed in the early 1970s, of which one, the Cambridge Ring had demonstrated the potential of a token passing ring topology, many teams worldwide began working on their own implementations. At the IBM Zurich Research Laboratory Werner Bux and Hans Müller in particular worked on the design and development of IBM's Token Ring technology, while early work at MIT led to the Proteon 10 Mbit/s ProNet-10 Token Ring network in 1981 – the same year that workstation vendor Apollo Computer introduced their proprietary 12 Mbit/s Apollo Token Ring network running over 75-ohm RG-6U coaxial cabling.
Proteon evolved a 16 Mbit/s version that ran on unshielded twisted pair cable. IBM launched their own proprietary Token Ring product on October 15, 1985, it ran at 4 Mbit/s, attachment was possible from IBM PCs, midrange computers and mainframes. It used a convenient star-wired physical topology, ran over shielded twisted-pair cabling, shortly thereafter became the basis for the /IEEE standard 802.5. During this time, IBM argued that Token Ring LANs were superior to Ethernet under load, but these claims were fiercely debated. In 1988 the faster 16 Mbit/s Token Ring was standardized by the 802.5 working group, an increase to 100 Mbit/s was standardized and marketed during the wane of Token Ring's existence. However it was never used, while a 1000 Mbit/s standard was approved in 2001, no products were brought to market and standards activity came to a standstill as Fast Ethernet and Gigabit Ethernet dominated the local area networking market. Ethernet and Token Ring have some notable differences: Token Ring access is more deterministic, compared to Ethernet's contention-based CSMA/CD Ethernet supports a direct cable connection between two network interface cards by the use of a crossover cable or through auto-sensing if supported.
Token Ring does not inherently support this feature and requires additional software and hardware to operate on a direct cable connection setup. Token Ring eliminates collision by the use of a single-use token and early token release to alleviate the down time. Ethernet alleviates collision by carrier sense multiple access and by the use of an intelligent switch. Token Ring network interface cards contain all of the intelligence required for speed autodetection and can drive themselves on many Multistation Access Units that operate without power. Ethernet network interface cards can theoretically operate on a passive hub to a degree, but not as a large LAN and the issue of collisions is still present. Token Ring employs ` access priority'. Unswitched Ethernet does not have provisioning for an access priority system as all nodes have equal contest for traffic. Multiple identical MAC addresses are supported on Token Ring. Switched Ethernet cannot support duplicate MAC addresses without reprimand.
Token Ring was more complex than Ethernet, requiring a specialized processor and licensed MAC/LLC firmware for each interface. By contrast, Ethernet included both the lower licensing cost in the MAC chip; the cost of a token Ring interface using the Texas Instruments TMS380C16 MAC and PHY was three times that of an Ethernet interface using the Intel 82586 MAC and PHY. Both networks used expensive cable, but once Ethernet was standardized for unshielded twisted pair with 10BASE-T and 100BASE-TX, it had a distinct advantage and sales of it increased markedly. More significant when comparing overall system costs was the much-higher cost of router ports and network cards for Token Ring vs Ethernet; the emergence of Ethernet switches may have been the final straw. Stations on a Token Ring LAN are logically organized in a ring topology with data being transmitted sequentially from one ring station to the next with a control token circulating around the ring controlling access. Similar token passing mechanisms are used by ARCNET, token bus, 100VG-AnyLAN and FDDI, they have theoretical advantages over the CSMA/CD of early Ethernet.
A Token Ring network can be modeled as a polling system where a single server provides service to queues in a cyclic order. The data transmission process goes as follows: Empty information frames are continuously circulated on the ring; when a computer has a message to send, it seizes the token. The computer will be able to send the frame; the frame is examined by each successive workstation. The workstation that identifies itself to be the destination for the message copies it from the frame and changes the token back to 0; when the frame gets back to the originat
Realtek Semiconductor Corp. is a fabless semiconductor company situated in the Hsinchu Science Park, Taiwan. It was founded in October 1987 and subsequently listed on the Taiwan Stock Exchange in 1998. Realtek manufactures and sells a variety of microchips globally and its product lines broadly fall into three categories: communications network ICs, computer peripheral ICs, multimedia ICs; as of 2017, Realtek employs 4,000 people, of whom 78 % work in development. Communication network IC products manufactured and marketed by Realtek include: network interface controllers, physical layer controllers, network switch controllers, gateway controllers, wireless LAN ICs, as well as ADSL router controllers. In particular, the RTL8139 series 10/100M Fast Ethernet controllers reached their height during the late 1990s, continued to take up a significant, predominant share in the worldwide market in the following years; those devices categorized as Realtek’s computer peripheral IC products consist of the traditional AC'97 audio codecs, the High Definition Audio codecs, card reader controllers, clock generators and IEEE 1394 ICs.
Multimedia IC products include LCD Monitor Controllers, LCD TV Controllers and Digital Media Processors. Notable Realtek products include 10/100M Ethernet controllers and audio codecs, where Realtek had a 50% market share in 2003 and a 60% market share in 2004 concentrated in the integrated OEM on-board audio market-segment; as of 2013 the ALC892 HD Audio codec and RTL8111 Gigabit Ethernet chip have become particular OEM favorites, offering low prices and basic feature-sets. RTL8139-based NICs are dubbed "crab cards" in Taiwan, alluding to the crab-like appearance of the Realtek logo; the increasing popularity of HD media players in 2009 led to the entry of Realtek into that market. The first series, the 1xx3 models sold at a lower price than similar quality chipsets of Realtek's competitors. Realtek produced three major versions of several minor variations; the three major 1xx3 chipset versions all featured the same chip in terms of format support and performance, the only difference being the added ability to record AV sources in the 1283.
HD Audio support in the 1xx3 improved through the chipset's life with several revisions. The DD and CC versions of the chipset both added full 7.1 HD-audio support to the chipset. The 1073 players all built on a common SDK provided by Realtek; this meant that they were all similar in performance and interface. It meant that producing these players was easy for manufacturers, all they had to do was create the hardware and Realtek provided the software. Key players from the Realtek 1073 era were the original Xtreamer, the Asus O! PlayHD, ACRyan PlayOn and the Mede8er MED500X. Manufacturers released hundreds of Realtek 1073 players. In early 2011 Realtek released series 1xx5, including the 1055, 1185; these are the successors to the 1073 series. All three chips ran at 500Mhz. Otherwise, the chips offered the same comprehensive format support as the previous generation. All chips ran the same Realtek SDK4 Casablanca, which offered improved user-experience from the stock SDK; as with the version of the 1xx8 chipset, full 7.1 HD-audio downmix and passthrough are supported in the 1xx5.
Realtek released the next generation of its chipsets, the 1xx6 series 1186, in early October 2011. These ran at 750Mhz, supported HDMI 1.4, were capable of 3D including 3D ISO, were able to dual-boot into Android. Key 1186 players include the Mede8er X3D Series, Xtreamer Prodigy 3D and HiMedia 900B. According to the comprehensive analysis released by Symantec in 2011 regarding the Stuxnet virus, Realtek's digital certificate for Windows was compromised, allowing attackers to digitally sign malicious drivers without users being notified; the certificate was revoked by Verisign: "The attackers would have needed to obtain the digital certificates from someone who may have physically entered the premises of the two companies and stole them, as the two companies are in close physical proximity." States the report. List of Taiwanese companies Network interface card Sound card RTL8139 Official website List of products using Realtek chipsets
IEEE 1394 is an interface standard for a serial bus for high-speed communications and isochronous real-time data transfer. It was developed in early 1990s by Apple, which called it FireWire; the 1394 interface is known by the brands i. LINK, Lynx; the copper cable it uses in its most common implementation can be up to 4.5 metres long. Power is carried over this cable, allowing devices with moderate power requirements to operate without a separate power supply. FireWire is available in Cat 5 and optical fiber versions; the 1394 interface is comparable to USB. USB gained much greater market share. USB requires a master controller whereas IEEE 1394 is cooperatively managed by the connected devices. FireWire is Apple's name for the IEEE 1394 High Speed Serial Bus, it was initiated by Apple and developed by the IEEE P1394 Working Group driven by contributions from Apple, although major contributions were made by engineers from Texas Instruments, Digital Equipment Corporation, IBM, INMOS/SGS Thomson. IEEE 1394 is a serial bus architecture for high-speed data transfer.
FireWire is a serial bus. Parallel buses utilize a number of different physical connections, as such are more costly and heavier. IEEE 1394 supports both isochronous and asynchronous applications. Apple intended FireWire to be a serial replacement for the parallel SCSI bus, while providing connectivity for digital audio and video equipment. Apple's development began in the late 1980s presented to the IEEE, was completed in January 1995. In 2007, IEEE 1394 was a composite of four documents: the original IEEE Std. 1394-1995, the IEEE Std. 1394a-2000 amendment, the IEEE Std. 1394b-2002 amendment, the IEEE Std. 1394c-2006 amendment. On June 12, 2008, all these amendments as well as errata and some technical updates were incorporated into a superseding standard, IEEE Std. 1394-2008. Apple first included on-board FireWire in some of its 1999 Macintosh models, most Apple Macintosh computers manufactured in the years 2000 through 2011 included FireWire ports. However, in February 2011 Apple introduced the first commercially available computer with Thunderbolt.
Apple released its last computers featuring FireWire late 2012. By 2014, Thunderbolt had become a standard feature across Apple's entire line of computers becoming the spiritual successor to FireWire in the Apple ecosystem. Sony's implementation of i. LINK, used a smaller connector with only four signal conductors, omitting the two conductors that provide power for devices in favor of a separate power connector; this style was added into the 1394a amendment. This port is sometimes labeled S400 to indicate speed in Mbit/s; the system was used to connect data storage devices and DV cameras, but was popular in industrial systems for machine vision and professional audio systems. Many users preferred it over the more common USB 2.0 for its greater effective speed and power distribution capabilities. Benchmarks show that the sustained data transfer rates are higher for FireWire than for USB 2.0, but lower than USB 3.0. Results are marked on Apple Mac OS X but more varied on Microsoft Windows. Implementation of IEEE 1394 is said to require use of 261 issued international patents held by 10 corporations.
Use of these patents requires licensing. Companies holding IEEE 1394 IP formed a patent pool with MPEG LA, LLC as the license administrator, to whom they licensed patents. MPEG LA sublicenses these patents to providers of equipment implementing IEEE 1394. Under the typical patent pool license, a royalty of US$0.25 per unit is payable by the manufacturer upon the manufacture of each 1394 finished product. A person or company may review the actual 1394 Patent Portfolio License upon request to MPEG LA. Implementors would thereby ordinarily reveal some interest to MPEG LA early in the design process. MPEG LA does not provide assurance of protection to licensees beyond its own patents. At least one licensed patent is known to be removed from the pool, other hardware patents exist that reference 1394-related hardware and software functions related to use in IEEE 1394. In total, over 1770 patents issued in the 20 years preceding 2011 contain "IEEE 1394" in their titles alone, placing 1500 unavailable from MPEG LA.
The 1394 High Performance Serial Bus Trade Association was formed to aid marketing of IEEE 1394. Its bylaws prohibit dealing with intellectual property issues; the 1394 Trade Association operates on an individual no cost membership basis to further enhancements to 1394 standards. The Trade Association is the library source for all 1394 documentation and standards available. FireWire can connect up to 63 peripherals in a daisy-chain topology, it allows peer-to-peer device communication — such as communication between a scanner and a printer — to take place without using system memory or the CPU. FireWire supports multiple hosts per bus, it is designed to support hot swapping. The copper cable it uses in its most common implementation can be up to 4.5 metres long and is more flexible than most parallel SCSI cables. In its six-conductor or nine-conductor variations, it can supply up to 45 watts of power per port at up to 30 volts, allowing moderate-consumption devices to operate without a separate power supply.
FireWire devices implement
In computer networking, Fast Ethernet physical layers carry traffic at the nominal rate of 100 Mbit/s. The prior Ethernet speed was 10 Mbit/s. Of the Fast Ethernet physical layers, 100BASE-TX is by far the most common. Fast Ethernet was introduced in 1995 as the IEEE 802.3u standard and remained the fastest version of Ethernet for three years before the introduction of Gigabit Ethernet. The acronym GE/FE is sometimes used for devices supporting both standards; the "100" in the media type designation refers to the transmission speed of 100 Mbit/s, while the "BASE" refers to baseband signalling. The letter following the dash refers to the physical medium that carries the signal, while the last character refers to the line code method used. Fast Ethernet is sometimes referred to as 100BASE-X, where "X" is a placeholder for the FX and TX variants. Fast Ethernet is an extension of the 10 megabit Ethernet standard, it runs on twisted pair or optical fiber cable in a star wired bus topology, similar to the IEEE standard 802.3i called 10BASE-T, itself an evolution of 10BASE5 and 10BASE2.
Fast Ethernet devices are backward compatible with existing 10BASE-T systems, enabling plug-and-play upgrades from 10BASE-T. Most switches and other networking devices with ports capable of Fast Ethernet can perform autonegotiation, sensing a piece of 10BASE-T equipment and setting the port to 10BASE-T half duplex if the 10BASE-T equipment cannot perform auto negotiation itself; the standard specifies the use of CSMA/CD for media access control. A full-duplex mode is specified and in practice all modern networks use Ethernet switches and operate in full-duplex mode as legacy devices that use half duplex still exist. A Fast Ethernet adapter can be logically divided into a media access controller, which deals with the higher-level issues of medium availability, a physical layer interface; the MAC is linked to the PHY by a four-bit 25 MHz synchronous parallel interface known as a media-independent interface, or by a two-bit 50 MHz variant called reduced media independent interface. In rare cases the MII may be an external connection but is a connection between ICs in a network adapter or two sections within a single IC.
The specs are written based on the assumption that the interface between MAC and PHY will be an MII but they do not require it. Fast Ethernet or Ethernet hubs may use the MII to connect to multiple PHYs for their different interfaces; the MII fixes the theoretical maximum data bit rate for all versions of Fast Ethernet to 100 Mbit/s. The information rate observed on real networks is less than the theoretical maximum, due to the necessary header and trailer on every Ethernet frame, the required interpacket gap between transmissions. 100BASE-T is any of several Fast Ethernet standards for twisted pair cables, including: 100BASE-TX, 100BASE-T4, 100BASE-T2. The segment length for a 100BASE-T cable is limited to 100 metres. All are or were standards under IEEE 802.3. All 100BASE-T installations are 100BASE-TX. 100BASE-TX is the predominant form of Fast Ethernet, runs over two wire-pairs inside a category 5 or above cable. Each network segment can have a maximum cabling distance of 100 metres. One pair is used for each direction, providing full-duplex operation with 100 Mbit/s of throughput in each direction.
Like 10BASE-T, the active pairs in a standard connection are terminated on pins 1, 2, 3 and 6. Since a typical category 5 cable contains 4 pairs, it can support two 100BASE-TX links with a wiring adaptor. Cabling is conventional wired to TIA/EIA-568-B's termination standards, T568A or T568B; this places the active pairs on the green pairs. The configuration of 100BASE-TX networks is similar to 10BASE-T; when used to build a local area network, the devices on the network are connected to a hub or switch, creating a star network. Alternatively it is possible to connect two devices directly using a crossover cable. With today's equipment, crossover cables are not needed as most equipment support auto-negotiation along with auto MDI-X to select and match speed and pairing. With 100BASE-TX hardware, the raw bits, presented 4 bits wide clocked at 25 MHz at the MII, go through 4B5B binary encoding to generate a series of 0 and 1 symbols clocked at a 125 MHz symbol rate; the 4B5B encoding provides DC spectrum shaping.
Just as in the 100BASE-FX case, the bits are transferred to the physical medium attachment layer using NRZI encoding. However, 100BASE-TX introduces an additional, medium dependent sublayer, which employs MLT-3 as a final encoding of the data stream before transmission, resulting in a maximum fundamental frequency of 31.25 MHz. The procedure is borrowed with minor changes. 100BASE-T4 was an early implementation of Fast Ethernet. It requires four twisted copper pairs of voice grade twisted pair, a lower performing cable compared to category 5 cable used by 1000BASE-TX. Maximum distance is limited to 100 meters. One pair is reserved for transmit, one for receive, the remaining two switch direction; the fact that 3 pairs are used to transmit in each direction makes 100BASE-T4 inherently half-duplex. A unusual 8B6T code is used to convert 8 data bits into 6 base-3 digits (the signal shaping is possible as there are nearly three times as many 6-digit base-3 n
Thunderbolt is the brand name of a hardware interface developed by Intel that allows the connection of external peripherals to a computer. Thunderbolt 1 and 2 use the same connector as Mini DisplayPort, whereas Thunderbolt 3 re-uses the Type-C connector from USB, it was developed and marketed under the name Light Peak, first sold as part of a consumer product on 24 February 2011. Thunderbolt combines PCI Express and DisplayPort into two serial signals, additionally provides DC power, all in one cable. Up to six peripherals may be supported by one connector through various topologies. Thunderbolt controllers multiplex one or more individual data lanes from connected PCIe and DisplayPort devices for transmission via two duplex Thunderbolt lanes de-multiplex them for use by PCIe and DisplayPort devices on the other end. A single Thunderbolt port supports up to six Thunderbolt devices via hubs or daisy chains. A single Mini DisplayPort monitor or other device of any kind may be connected directly or at the end of the chain.
Thunderbolt is interoperable with DP-1.1a compatible devices. When connected to a DP-compatible device, the Thunderbolt port can provide a native DisplayPort signal with four lanes of output data at no more than 5.4 Gbit/s per Thunderbolt lane. When connected to a Thunderbolt device, the per-lane data rate becomes 10 Gbit/s and the four Thunderbolt lanes are configured as two duplex lanes, each 10 Gbit/s comprising one lane of input and one lane of output. Thunderbolt can be implemented on PCIe graphics cards, which have access to DisplayPort data and PCIe connectivity, or on the motherboard of new computers with onboard video, such as the MacBook Air; the interface was intended to run on an optical physical layer using components and flexible optical fiber cabling developed by Intel partners and at Intel's Silicon Photonics lab. It was marketed under the name Light Peak, after 2011 as Silicon Photonics Link. However, it was discovered that conventional copper wiring could furnish the desired 10 Gbit/s per channel at lower cost.
This copper-based version of the Light Peak concept was co-developed by Intel. Apple registered Thunderbolt as a trademark, but transferred the mark to Intel, which held overriding intellectual-property rights. Thunderbolt was commercially introduced on Apple's 2011 MacBook Pro, using the same Apple-developed connector as Mini DisplayPort, electrically identical to DisplayPort, but uses a smaller, non-locking connector. Sumitomo Electric Industries started selling up to 30-metre-long optical Thunderbolt cables in Japan in January, 2013, Corning, Inc. began selling up to 60-metre-long optical cables in the U. S. in late September, 2013. Intel introduced Light Peak at the 2009 Intel Developer Forum, using a prototype Mac Pro logic board to run two 1080p video streams plus LAN and storage devices over a single 30-meter optical cable with modified USB ends; the system was driven with two optical buses powering four ports. Jason Ziller, head of Intel's Optical I/O Program Office showed the internal components of the technology under a microscope and the sending of data through an oscilloscope.
The technology was described as having an initial speed of 10 Gbit/s over plastic optical cables, promising a final speed of 100 Gbit/s. At the show, Intel said Light Peak-equipped systems would begin to appear in 2010, posted to YouTube a video showing Light Peak-connected HD cameras, docking stations, HD monitors. On 4 May 2010, in Brussels, Intel demonstrated a laptop with a Light Peak connector, indicating that the technology had shrunk enough to fit inside such a device, had the laptop send two simultaneous HD video streams down the connection, indicating that at least some fraction of the software/firmware stacks and protocols were functional. At the same demonstration, Intel officials said they expected hardware manufacturing to begin around the end of 2010. In September 2010, some early commercial prototypes from manufacturers were demonstrated at Intel Developer Forum 2010. Though Thunderbolt was conceived as an optical technology, Intel switched to electrical connections to reduce costs and to supply up to 10 watts of power to connected devices.
In 2009, Intel officials said the company was "working on bundling the optical fiber with copper wire so Light Peak can be used to power devices plugged into the PC". In 2010, Intel said the original intent was "to have one single connector technology" that would allow "electrical USB 3.0... and piggyback on USB 3.0 or 4.0 DC power". Light Peak aimed to make great strides in consumer-ready optical technology, by having achieved " for 7,000 insertions, which matches or exceeds other PC connections... cables in multiple knots to make sure it didn't break and the loss is acceptable" and "you can get two people pulling on it at once and it won't break the fibre". They predicted that "Light Peak cables will be no more expensive than HDMI". In January 2011, Intel's David Perlmutter told Computerworld that initial Thunderbolt implementations would be based on copper wires. "The copper came out good better than what we thought", he said. A major advantage of copper is the ability to carry power; the final Thunderbolt standard specifies 10 W DC on every port.
See comparison section below. Intel and industry partners are still developing optical Thunderbolt hardware and cables; the optical fiber cables are to run "tens of meters" but will not supply power, at least not initially. The version from Corning contains four 80/125 µm VSDN fibers to transport an i
Conventional PCI shortened to PCI, is a local computer bus for attaching hardware devices in a computer. PCI is part of the PCI Local Bus standard; the PCI bus supports the functions found on a processor bus but in a standardized format, independent of any particular processor's native bus. Devices connected to the PCI bus appear to a bus master to be connected directly to its own bus and are assigned addresses in the processor's address space, it is a parallel bus, synchronous to a single bus clock. Attached devices can take either the form of an integrated circuit fitted onto the motherboard itself or an expansion card that fits into a slot; the PCI Local Bus was first implemented in IBM PC compatibles, where it displaced the combination of several slow ISA slots and one fast VESA Local Bus slot as the bus configuration. It has subsequently been adopted for other computer types. Typical PCI cards used in PCs include: network cards, sound cards, extra ports such as USB or serial, TV tuner cards and disk controllers.
PCI video cards replaced ISA and VESA cards until growing bandwidth requirements outgrew the capabilities of PCI. The preferred interface for video cards became AGP, itself a superset of conventional PCI, before giving way to PCI Express; the first version of conventional PCI found in consumer desktop computers was a 32-bit bus using a 33 MHz bus clock and 5 V signalling, although the PCI 1.0 standard provided for a 64-bit variant as well. These have one locating notch in the card. Version 2.0 of the PCI standard introduced 3.3 V slots, physically distinguished by a flipped physical connector to prevent accidental insertion of 5 V cards. Universal cards, which can operate on either voltage, have two notches. Version 2.1 of the PCI standard introduced optional 66 MHz operation. A server-oriented variant of conventional PCI, called PCI-X operated at frequencies up to 133 MHz for PCI-X 1.0 and up to 533 MHz for PCI-X 2.0. An internal connector for laptop cards, called Mini PCI, was introduced in version 2.2 of the PCI specification.
The PCI bus was adopted for an external laptop connector standard – the CardBus. The first PCI specification was developed by Intel, but subsequent development of the standard became the responsibility of the PCI Special Interest Group. Conventional PCI and PCI-X are sometimes called Parallel PCI in order to distinguish them technologically from their more recent successor PCI Express, which adopted a serial, lane-based architecture. Conventional PCI's heyday in the desktop computer market was 1995–2005. PCI and PCI-X have become obsolete for most purposes. Many kinds of devices available on PCI expansion cards are now integrated onto motherboards or available in USB and PCI Express versions. Work on PCI began at Intel's Architecture Development Lab c. 1990. A team of Intel engineers defined the architecture and developed a proof of concept chipset and platform partnering with teams in the company's desktop PC systems and core logic product organizations. PCI was put to use in servers, replacing MCA and EISA as the server expansion bus of choice.
In mainstream PCs, PCI was slower to replace VESA Local Bus, did not gain significant market penetration until late 1994 in second-generation Pentium PCs. By 1996, VLB was all but extinct, manufacturers had adopted PCI for 486 computers. EISA continued to be used alongside PCI through 2000. Apple Computer adopted PCI for professional Power Macintosh computers in mid-1995, the consumer Performa product line in mid-1996; the 64-bit version of plain PCI remained rare in practice though, although it was used for example by all G3 and G4 Power Macintosh computers. Revisions of PCI added new features and performance improvements, including a 66 MHz 3.3 V standard and 133 MHz PCI-X, the adaptation of PCI signaling to other form factors. Both PCI-X 1.0b and PCI-X 2.0 are backward compatible with some PCI standards. These revisions were used on server hardware but consumer PC hardware remained nearly all 32 bit, 33 MHz and 5 volt; the PCI-SIG introduced the serial PCI Express in c. 2004. At the same time, they renamed PCI as Conventional PCI.
Since motherboard manufacturers have included progressively fewer Conventional PCI slots in favor of the new standard. Many new motherboards do not provide conventional PCI slots at all, as of late 2013. PCI provides separate memory and I/O port address spaces for the x86 processor family, 64 and 32 bits, respectively. Addresses in these address spaces are assigned by software. A third address space, called the PCI Configuration Space, which uses a fixed addressing scheme, allows software to determine the amount of memory and I/O address space needed by each device; each device can request up to six areas of memory space or I/O port space via its configuration space registers. In a typical system, the firmware queries all PCI buses at startup time to find out what devices are present and what system resources each needs, it allocates the resources and tells each device what its allocation is. The PCI configuration space contains a small amount of device type information, which helps an operating system choose device drivers for it, or at least to have a dialogue with a user about the system configuration.
Devices may have an on-board ROM containing executable code for x86 or PA-RISC processors, an Op