| << Chapter < Page | Chapter >> Page > |
ROAD MAP OF SUPERCOMPUTER.[‘The Data’, IEEE SPECTRUM, APRIL 2009]
Table I.3. Road Map of Super Computer.
| Year | Name | No.of processors | Capability | Cost |
| 1976 | Illiac IV | 64 | TeraFLOPS | 480kFLOPS/million dollar |
| 1976 | Cray 1 | 1 | TeraFLOPS | 17.78MFLOPS/ million dollars |
| 1988 | CRAY Y-MP | 8-vectorprocessor | TeraFLOPS | 111.5MFLOPS/ million dollars |
| 1997 | ASCI RED | 4510 | TeraFLOPS | 18.18182GFLOPS/million dollars |
| 2002 | EARTHSIMULATOR | 5120 | TeraFLOPS | 17.5GFLOPS/million dollars |
| 2004 | BLUEGENE/L | 65536 | 10TeraFLOPS | 2.8TFLOPS/million dollars |
| 2007 | PLAYSTATION3 CLUSTER† | 8playstation | 375TFLOPS/million dollars | |
| 2008 | NVIDIA TESLA* | 960cores | 439.1 TFLOPS/ million dollars | |
| 2008 | ROADRUNNER‡ | 19440 | 10sTeraFLOPS | 8.3TFLOPS/ million dollars |
*Based on the cost of building your own Tesla supercomputer out of four Tesla C1060 units. Complete instructions are available at nvidia.com
† Seven processors per node at 150GFLPS per Unit for a single processor( 11 processors per node for a dual processor). $400-plus per node.
‡ 6480 Opteron CPUs and 12960 Cell processors.
Supercomputers are massively parallel chips can be used to calculate protein folding, predict climate change and crack the encryption of hitherto-secure Web sites.
Nvidia advertises its new workstation, the Tesla, as a “personal supercomputer”. It clusters 4 Nvidia C1060 processing boards, each of which unites 240 graphic cores to process instructions at nearly teraflop speed.
Tesla does single-precision floating-point calculations using 8-bit bytes.
Roadrunner uses 64-bit floating integers.
Table I.4.Evolution of Microprocessor Chip
| Name | Technology | Data&Address Size | Packing Density | Number of Pins | Clock Rate | Instructions Per Second | Year&Price. |
| Intel 4004 | 10µm | 4-bit | 2300PMOS | 16 | 740kHz | 92kIPS | 1971 Rs 7,500.00 |
| Intel 8080 | 3 µm | 8-bit | 6000NMOS | 40 | 2MHz | 0.64MIPS | 1977 |
| Intel 8086 | 3 µm | 16-bit | 29,000 HMOS | 40 | 5 MHz | 0.33MIPS | 1978 |
| Intel 80286 | 1.5 µm | Data 16b Address 24b | 134,000 HMOS | 68 | 6-12.5 MHz | 1.8 MIPS | 1982 |
| Intel 80386 | 1 µm | 32-bit | 275,000 CMOS | 132 | 16 MHZ | 5 MIPS | 1985 |
| Intel 80486 | 800nmpipelined | 32-bit | 1.2million CMOS | 168 | 25 MHZ | 20 MIPS | 1989 |
| Pentium | 600nmSuperscalararchitecture | Data 64b Address32bMemory Size 2 32= 4GB | 3.1 million CMOS | ? | 60 MHZ | 100 MIPS | 1993 |
| Pentium II | 350nm | Same | 7.5 million CMOS | 370 | 233 MHZ | 400 MIPS | 1997 |
| Pentium III | 250nm | same | 9.5 CMOS | 370 | 450 MHZ | 1000 MIPS | 1999 |
| Pentium IV | 180nm | Data 64b Address 64b | 42 million CMOS(Nano electronics) | 478 | 2.2 GHz | 5GIPS | 2001 |
MOORE’s LAW: The packing density of CMOS on μP Chip would double every two years leading to an exponential growth in the complexity and speed of μP Chip.
Table I.5. Evolution of Computers.
| Generation | Components | Architecture | Logic | Programming | Language | ||||||
| 1 st (1947-ENIAC 1952- Univac I, Main-Frame) | Vacuum Tubes |
| 2 nd (1956-UNIVAC II, Main Frame | Discrete BJT |
| 3 rd (1964- PDP8,Mini-comp.,Table Top) | IC’s chips |
| 4 th (1977- APPLE II, Micro-comp., Desk Top, Personal Comp. | μP chips |
Table I.6. Evolution of Local Area Network
| Year | Standard | Name | Data transfer rate |
| 1973 | 802.1 | OSI (Open System Interconnection) | 10Mbps |
| 802.2 | Logical Link Control | 10Mbps | |
| 1983 | 802.3 | CSMA/CD (Carrier Sense Multiple Access/Collision Detect) | 10Mbps |
| 802.4 | Token Bus | 100Mbps | |
| 802.5 | Token Ring | 100Mbps | |
| 1999 | 802.11 | WLL (Wire Local Loop) | 1 Gbps |
| 2001 | 802.13 | Wi-Fi (Wireless internet Access |
Table I.7. Development in submarine cable.(Sources: Telegraphy Research: “History of the Atlantic Cable and Undersea Communications” by Bill Glover,Wikipedia).
| Region | Submarine Cable %BW utilization | LIT Submarine Cable Capacity |
| Asia | 69% | |
| Middle East | 73% | |
| Oceania | 56% | |
| Europe | 52% | |
| US&Europe | 51% | |
| Latin America&Caribbean | 77% | |
| Africa | 69% | |
| Transatlantic Cable | 15% | |
| US-Latin America | 37% | |
| Transpacific | 21% | |
| Europe-Asia | 91% | |
| Intra-Asia | 31% |
Submarine Cable Boom: In 2001 US $13.5 billion was spent. Due to Dot.com bust, only $2billion was spent during four lean years from 2004 to 2007. Even the during the lean periods, global demand never slaked growing at an average compound annual rate of 54% from 2002 to 2008. In Latin America and the Caribbean, traffic grew the most at more than 75%.
METALFE LAW: The value of a NETWORK grows as the square of the numbe of users.
Notification Switch
Would you like to follow the 'Solid state physics and devices-the harbinger of third wave of civilization' conversation and receive update notifications?
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|