The "memory wall" is the growing disparity of speed between CPU and memory outside the CPU chip. An important reason for this disparity is the limited communication bandwidth beyond chip boundaries. From 1986 to 2000, CPU speed improved at an annual rate of 55% while memory speed only improved at 10%. Given these trends, it was expected that memory latency would become an overwhelming bottleneck in computer performance. [3]
Currently, CPU speed improvements have slowed significantly partly due to major physical barriers and partly because current CPU designs have already hit the memory wall in some sense. Intel summarized these causes in their Platform 2015 documentation (PDF)
“First of all, as chip geometries shrink and clock frequencies rise, the transistor leakage current increases, leading to excess power consumption and heat (more on power consumption below). Secondly, the advantages of higher clock speeds are in part negated by memory latency, since memory access times have not been able to keep pace with increasing clock frequencies. Third, for certain applications, traditional serial architectures are becoming less efficient as processors get faster (due to the so-called Von Neumann bottleneck), further undercutting any gains that frequency increases might otherwise buy. In addition, partly due to limitations in the means of producing inductance within solid state devices, resistance-capacitance (RC) delays in signal transmission are growing as feature sizes shrink, imposing an additional bottleneck that frequency increases don't address.”
The RC delays in signal transmission were also noted in Clock Rate versus IPC: The End of the Road for Conventional Microarchitectures which projects a maximum of 12.5% average annual CPU performance improvement between 2000 and 2014. The data on Intel Processors clearly shows a slowdown in performance improvements in recent processors. However, Intel's new processors, Core 2 Duo (codenamed Conroe) show a significant improvement over previous Pentium 4 processors; due to a more efficient architecture, performance increased while clock rate actually decreased.
Saturday, April 18, 2009
DDR2 RAM
Modern types of writable RAM generally store a bit of data in either the state of a flip-flop, as in SRAM (static RAM), or as a charge in a capacitor (or transistor gate), as in DRAM (dynamic RAM), EPROM, EEPROM and Flash. Some types have circuitry to detect and/or correct random faults called memory errors in the stored data, using parity bits or error correction codes. RAM of the read-only type, ROM, instead uses a metal mask to permanently enable/disable selected transistors, instead of storing a charge in them.
As both SRAM and DRAM are volatile, other forms of computer storage, such as disks and magnetic tapes, have been used as persistent storage in traditional computers. Many newer products instead rely on flash memory to maintain data when not in use, such as PDAs or small music players. Certain personal computers, such as many rugged computers and netbooks, have also replaced magnetic disks with flash drives. With flash memory, only the NOR type is capable of true random access, allowing direct code execution, and is therefore often used instead of ROM; the lower cost NAND type is commonly used for bulk storage in memory cards and solid-state drives.
History of RAM
An early type of widespread writable random access memory was the magnetic core memory, developed from 1949 to 1952, and subsequently used in most computers up until the development of the static and dynamic integrated RAM circuits in the late 1960s and early 1970s. Before this, computers used relays, delay line memory or various kinds of vacuum tube arrangements to implement "main" memory functions (i.e., hundreds or thousands of bits), some of which were random access, some not. Latches built out of vacuum tube triodes, and later, out of discrete transistors, were used for smaller and faster memories such as registers and (random access) register banks. Prior to the development of integrated ROM circuits, permanent (or read-only) random access memory was often constructed using semiconductor diode matrices driven by address decoders.
History of hard disk drives
The commercial usage of hard disk drives began in 1956 with the shipment of an IBM 305 RAMAC system including IBM Model 350 disk storage[1].
For many years, hard disk drives were large, cumbersome devices, more suited to use in the protected environment of a data center or large office than in a harsh industrial environment (due to their delicacy), or small office or home (due to their size and power consumption). Before the early 1980s, most hard disk drives had 8-inch (actually, 210 - 195 mm) or 14-inch platters, required an equipment rack or a large amount of floor space (especially the large removable-media drives, which were frequently comparable in size to washing machines), and in many cases needed high-current and/or three-phase power hookups due to the large motors they used. Because of this, hard disk drives were not commonly used with microcomputers until after 1980, when Seagate Technology introduced the ST-506, the first 5.25-inch hard drives, with a formatted capacity of 5 megabytes.
The capacity of hard drives has grown exponentially over time. With early personal computers, a drive with a 20 megabyte capacity was considered large. During the mid to late 1990s, when PCs were capable of storing not just text files and documents but pictures, music, and video, internal drives were made with 8 to 20 GB capacities. As of early 2009, desktop hard disk drives typically have a capacity of 320 to 500 gigabytes, while the largest-capacity drives are 2 terabytes.
1950s - 1970s
Main article: early IBM disk storage
The IBM 350 Disk File, invented by Reynold Johnson, was introduced in 1956 with the IBM 305 RAMAC computer. This drive had fifty 24 inch platters, with a total capacity of five million characters. A single head assembly having two heads was used for access to all the platters, making the average access time very slow (just under 1 second).
The IBM 1301 Disk Storage Unit[2], announced in 1961, introduced the usage of a head for each data surface with the heads having self acting air bearings (flying heads).
The first disk drive to use removable media was the IBM 1311 drive, which used the IBM 1316 disk pack to store two million characters.
In 1973, IBM introduced the IBM 3340 "Winchester" disk drive, the first significant commercial use of low mass and low load heads with lubricated media. All modern disk drives now use this technology and/or derivatives thereof. Project head designer/lead designer Kenneth Haughton named it after the Winchester 30-30 rifle after the developers called it the "30-30" because of it was planned to have two 30 MB spindles; however, the actual product shipped with two spindles for data modules of either 35 MB or 70 MB[3].
1980s - PC era
Internal drives became the system of choice on PCs in the 1980s. Most microcomputer hard disk drives in the early 1980s were not sold under their manufacturer's names, but by OEMs as part of larger peripherals (such as the Corvus Disk System and the Apple ProFile). The IBM PC/XT had an internal hard disk drive, however, and this started a trend toward buying "bare" drives (often by mail order) and installing them directly into a system.
External hard drives remained popular for much longer on the Apple Macintosh and other platforms. Every Mac made between 1986 and 1998 has a SCSI port on the back, making external expansion easy; also, "toaster" Compact Macs did not have easily accessible hard drive bays (or, in the case of the Mac Plus, any hard drive bay at all), so on those models, external SCSI disks were the only reasonable option
For many years, hard disk drives were large, cumbersome devices, more suited to use in the protected environment of a data center or large office than in a harsh industrial environment (due to their delicacy), or small office or home (due to their size and power consumption). Before the early 1980s, most hard disk drives had 8-inch (actually, 210 - 195 mm) or 14-inch platters, required an equipment rack or a large amount of floor space (especially the large removable-media drives, which were frequently comparable in size to washing machines), and in many cases needed high-current and/or three-phase power hookups due to the large motors they used. Because of this, hard disk drives were not commonly used with microcomputers until after 1980, when Seagate Technology introduced the ST-506, the first 5.25-inch hard drives, with a formatted capacity of 5 megabytes.
The capacity of hard drives has grown exponentially over time. With early personal computers, a drive with a 20 megabyte capacity was considered large. During the mid to late 1990s, when PCs were capable of storing not just text files and documents but pictures, music, and video, internal drives were made with 8 to 20 GB capacities. As of early 2009, desktop hard disk drives typically have a capacity of 320 to 500 gigabytes, while the largest-capacity drives are 2 terabytes.
1950s - 1970s
Main article: early IBM disk storage
The IBM 350 Disk File, invented by Reynold Johnson, was introduced in 1956 with the IBM 305 RAMAC computer. This drive had fifty 24 inch platters, with a total capacity of five million characters. A single head assembly having two heads was used for access to all the platters, making the average access time very slow (just under 1 second).
The IBM 1301 Disk Storage Unit[2], announced in 1961, introduced the usage of a head for each data surface with the heads having self acting air bearings (flying heads).
The first disk drive to use removable media was the IBM 1311 drive, which used the IBM 1316 disk pack to store two million characters.
In 1973, IBM introduced the IBM 3340 "Winchester" disk drive, the first significant commercial use of low mass and low load heads with lubricated media. All modern disk drives now use this technology and/or derivatives thereof. Project head designer/lead designer Kenneth Haughton named it after the Winchester 30-30 rifle after the developers called it the "30-30" because of it was planned to have two 30 MB spindles; however, the actual product shipped with two spindles for data modules of either 35 MB or 70 MB[3].
1980s - PC era
Internal drives became the system of choice on PCs in the 1980s. Most microcomputer hard disk drives in the early 1980s were not sold under their manufacturer's names, but by OEMs as part of larger peripherals (such as the Corvus Disk System and the Apple ProFile). The IBM PC/XT had an internal hard disk drive, however, and this started a trend toward buying "bare" drives (often by mail order) and installing them directly into a system.
External hard drives remained popular for much longer on the Apple Macintosh and other platforms. Every Mac made between 1986 and 1998 has a SCSI port on the back, making external expansion easy; also, "toaster" Compact Macs did not have easily accessible hard drive bays (or, in the case of the Mac Plus, any hard drive bay at all), so on those models, external SCSI disks were the only reasonable option
Generations
SSI, MSI and LSI
The first integrated circuits contained only a few transistors. Called "Small-Scale Integration" (SSI), they used circuits containing transistors numbering in the tens.
SSI circuits were crucial to early aerospace projects, and vice-versa. Both the Minuteman missile and Apollo program needed lightweight digital computers for their inertial guidance systems; the Apollo guidance computer led and motivated the integrated-circuit technology[citation needed], while the Minuteman missile forced it into mass-production.
These programs purchased almost all of the available integrated circuits from 1960 through 1963, and almost alone provided the demand that funded the production improvements to get the production costs from $1000/circuit (in 1960 dollars) to merely $25/circuit (in 1963 dollars).[citation needed] They began to appear in consumer products at the turn of the decade, a typical application being FM inter-carrier sound processing in television receivers.
The next step in the development of integrated circuits, taken in the late 1960s, introduced devices which contained hundreds of transistors on each chip, called "Medium-Scale Integration" (MSI).
They were attractive economically because while they cost little more to produce than SSI devices, they allowed more complex systems to be produced using smaller circuit boards, less assembly work (because of fewer separate components), and a number of other advantages.
Further development, driven by the same economic factors, led to "Large-Scale Integration" (LSI) in the mid 1970s, with tens of thousands of transistors per chip.
Integrated circuits such as 1K-bit RAMs, calculator chips, and the first microprocessors, that began to be manufactured in moderate quantities in the early 1970s, had under 4000 transistors. True LSI circuits, approaching 10000 transistors, began to be produced around 1974, for computer main memories and second-generation microprocessors.
The first integrated circuits contained only a few transistors. Called "Small-Scale Integration" (SSI), they used circuits containing transistors numbering in the tens.
SSI circuits were crucial to early aerospace projects, and vice-versa. Both the Minuteman missile and Apollo program needed lightweight digital computers for their inertial guidance systems; the Apollo guidance computer led and motivated the integrated-circuit technology[citation needed], while the Minuteman missile forced it into mass-production.
These programs purchased almost all of the available integrated circuits from 1960 through 1963, and almost alone provided the demand that funded the production improvements to get the production costs from $1000/circuit (in 1960 dollars) to merely $25/circuit (in 1963 dollars).[citation needed] They began to appear in consumer products at the turn of the decade, a typical application being FM inter-carrier sound processing in television receivers.
The next step in the development of integrated circuits, taken in the late 1960s, introduced devices which contained hundreds of transistors on each chip, called "Medium-Scale Integration" (MSI).
They were attractive economically because while they cost little more to produce than SSI devices, they allowed more complex systems to be produced using smaller circuit boards, less assembly work (because of fewer separate components), and a number of other advantages.
Further development, driven by the same economic factors, led to "Large-Scale Integration" (LSI) in the mid 1970s, with tens of thousands of transistors per chip.
Integrated circuits such as 1K-bit RAMs, calculator chips, and the first microprocessors, that began to be manufactured in moderate quantities in the early 1970s, had under 4000 transistors. True LSI circuits, approaching 10000 transistors, began to be produced around 1974, for computer main memories and second-generation microprocessors.
ULSI, WSI, SOC and 3D-IC
To reflect further growth of the complexity, the term ULSI that stands for "Ultra-Large Scale Integration" was proposed for chips of complexity of more than 1 million transistors.
Wafer-scale integration (WSI) is a system of building very-large integrated circuits that uses an entire silicon wafer to produce a single "super-chip". Through a combination of large size and reduced packaging, WSI could lead to dramatically reduced costs for some systems, notably massively parallel supercomputers. The name is taken from the term Very-Large-Scale Integration, the current state of the art when WSI was being developed.
System-on-a-Chip (SoC or SOC) is an integrated circuit in which all the components needed for a computer or other system are included on a single chip. The design of such a device can be complex and costly, and building disparate components on a single piece of silicon may compromise the efficiency of some elements. However, these drawbacks are offset by lower manufacturing and assembly costs and by a greatly reduced power budget: because signals among the components are kept on-die, much less power is required (see Packaging, above).
Three Dimensional Integrated Circuit (3D-IC) has two or more layers of active electronic components that are integrated both vertically and horizontally into a single circuit. Communication between layers uses on-die signaling, so power consumption is much lower than in equivalent separate circuits. Judicious use of short vertical wires can substantially reduce overall wire length for faster operation.
Wafer-scale integration (WSI) is a system of building very-large integrated circuits that uses an entire silicon wafer to produce a single "super-chip". Through a combination of large size and reduced packaging, WSI could lead to dramatically reduced costs for some systems, notably massively parallel supercomputers. The name is taken from the term Very-Large-Scale Integration, the current state of the art when WSI was being developed.
System-on-a-Chip (SoC or SOC) is an integrated circuit in which all the components needed for a computer or other system are included on a single chip. The design of such a device can be complex and costly, and building disparate components on a single piece of silicon may compromise the efficiency of some elements. However, these drawbacks are offset by lower manufacturing and assembly costs and by a greatly reduced power budget: because signals among the components are kept on-die, much less power is required (see Packaging, above).
Three Dimensional Integrated Circuit (3D-IC) has two or more layers of active electronic components that are integrated both vertically and horizontally into a single circuit. Communication between layers uses on-die signaling, so power consumption is much lower than in equivalent separate circuits. Judicious use of short vertical wires can substantially reduce overall wire length for faster operation.
Integrated circuit packaging
Early USSR made integrated circuit
The earliest integrated circuits were packaged in ceramic flat packs, which continued to be used by the military for their reliability and small size for many years. Commercial circuit packaging quickly moved to the dual in-line package (DIP), first in ceramic and later in plastic. In the 1980s pin counts of VLSI circuits exceeded the practical limit for DIP packaging, leading to pin grid array (PGA) and leadless chip carrier (LCC) packages. Surface mount packaging appeared in the early 1980s and became popular in the late 1980s, using finer lead pitch with leads formed as either gull-wing or J-lead, as exemplified by small-outline integrated circuit -- a carrier which occupies an area about 30 – 50% less than an equivalent DIP, with a typical thickness that is 70% less. This package has "gull wing" leads protruding from the two long sides and a lead spacing of 0.050 inches.
Small-outline integrated circuit (SOIC) and PLCC packages. In the late 1990s, PQFP and TSOP packages became the most common for high pin count devices, though PGA packages are still often used for high-end microprocessors. Intel and AMD are currently transitioning from PGA packages on high-end microprocessors to land grid array (LGA) packages.
Ball grid array (BGA) packages have existed since the 1970s. Flip-chip Ball Grid Array packages, which allow for much higher pin count than other package types, were developed in the 1990s. In an FCBGA package the die is mounted upside-down (flipped) and connects to the package balls via a package substrate that is similar to a printed-circuit board rather than by wires. FCBGA packages allow an array of input-output signals (called Area-I/O) to be distributed over the entire die rather than being confined to the die periphery.
Traces out of the die, through the package, and into the printed circuit board have very different electrical properties, compared to on-chip signals. They require special design techniques and need much more electric power than signals confined to the chip itself.
When multiple dies are put in one package, it is called SiP, for System In Package. When multiple dies are combined on a small substrate, often ceramic, it's called an MCM, or Multi-Chip Module. The boundary between a big MCM and a small printed circuit board is sometimes fuzzy.
Subscribe to:
Posts (Atom)
