LINPACK
 | You can help expand this article with text translated from the corresponding article in German. (October 2019) Click [show] for important translation instructions.
|
| Original author(s) | Jack Dongarra, Jim Bunch, Cleve Moler, and Gilbert Stewart |
|---|---|
| Written in | Fortran |
| Type | Library |
| Website | netlib |
LINPACK is a software library for performing numerical linear algebra on digital computers.[1] It was written in Fortran by Jack Dongarra, Jim Bunch, Cleve Moler, and Gilbert Stewart, and was intended for use on supercomputers in the 1970s and early 1980s.[2][3] It has been largely superseded by LAPACK, which runs more efficiently on modern architectures.
LINPACK makes use of the BLAS (Basic Linear Algebra Subprograms) libraries for performing basic vector and matrix operations.
The LINPACK benchmarks appeared initially as part of the LINPACK user's manual. The parallel LINPACK benchmark implementation called HPL (High Performance Linpack) is used to benchmark and rank supercomputers for the TOP500 list.
World's most powerful computer by year
[edit]| Year(s) | Benchmark | Computer | Design | Casing | System | Successor | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Manufacturer | Designer | Release date | Units sold | Price | Dimensions | Weight | Power | Front-end | Operating system | CPU | Memory | Storage | MIPS | FLOPS | ||||
| 1951-1954 | - | MIT Whirlwind I | MIT | - | 1954 | 1 | - | - | - | - | - | - | - | - | - | - | - | - |
| 1955-1957 | - | IBM NORC | IBM | - | 1954 | 1 | - | - | - | - | - | - | - | - | - | - | - | - |
| 1958-1959 | - | AN/FSQ-7 | IBM | - | 1955 | 24 | - | - | 250 tons | up to 3 megawatts | - | - | 49,000 vacuum tubes @ 75,000 instructions per second | - | - | - | - | - |
| 1960 | - | IBM 7090 | IBM | - | - | - | - | - | - | - | - | - | - | - | - | - | - | - |
| 1960–1961 | - | UNIVAC LARC | - | - | - | - | - | - | - | - | - | - | - | - | - | - | - | - |
| 1961–1963 | - | IBM 7030 Stretch | IBM | Gene Amdahl | May 1961 | 9 | US$7,780,000 (equivalent to $70,550,000 in 2021) | - | 70,000 pounds (35 short tons; 32 t) | 100 kW @ 110 V | - | MCP | 64-bit processor | 2048 kilobytes (262,144 x 64 bits) | - | 1.2 MIPS | - | CDC 7600 |
| 1964–1968 | 3 megaflops | CDC 6600 | Control Data Corporation | Seymour Cray | September 1964 | 100 | US$2,370,000 (equivalent to $20,710,000 in 2021) | Height : 2,000 mm (79 in) Cabinet width: 810 mm (32 in) Cabinet length : 1,710 mm (67 in) Width overall : 4,190 mm (165 in) |
about 12,000 lb (6.0 short tons; 5.4 t) | 30 kW @ 208 V 400 Hz | - | SCOPE, KRONOS | 60-bit processor @ 10 MHz | Up to 982 kilobytes (131000 x 60 bits) | - | 2 MIPS | - | CDC 7600 |
| 1969–1975 | 10 megaflops | CDC 7600 | Control Data Corporation | Seymour Cray | June 1967 | 75 | US$62 - $155 thousands (monthly rent in 1968) | Height : 188 cm (74 in) Width: 302 cm (119 in) |
- | 95 kW @ 208 V 400 Hz | - | Chippewa, SCOPE, KRONOS | 60-bit processor @ 36 MHz | 3.84 Megabytes (up to 512000 60-bit words) | - | 15 MIPS | 36 MFLOPS | CDC Cyber |
| 1976-1982 | 136 megaflops | Cray-1 | Cray Research | Seymour Cray | 1975 | 100 | US$7.9 million in 1977 (equivalent to $35.3 million in 2021) | Height: 196 cm (77 in) Dia. (base): 263 cm (104 in) Dia. (columns): 145 cm (57 in) |
5.5 tons (Cray-1A) | 115 kW @ 208 V 400 Hz | Data General Eclipse | COS, UNICOS | 64-bit processor @ 80 MHz | 8.39 Megabytes (up to 1 048 576 words) | 303 Megabytes (DD19 Unit) | - | 160 MFLOPS | Cray X-MP |
| 1983–1985 | 713 megaflops | Cray X-MP/4 | Cray Research | Steve Chen | 1982 | - | US$15 million | 2.62 m (8.6 ft) x 1.96 m (6.4 ft) | 5.12 t (11,300 lb) | 345 kW | Most minicomputers of the time | COS, UNICOS | 4x Vector processor 64 bits @ 105 - 117 MHz | 128 megabytes | 38.4 gigabytes (32 disks) | 400 MIPS (4 CPU) | 800 MFLOPS (4 CPU) | Cray-2 |
| 1985–1987 | 1.95 gigaflops (peak) | Cray-2 | Cray Research | - | 1985 | 25 | - | - | - | - | - | - | Custom Vector Processors | - | - | - | - | Cray Y-MP |
| 1988–1989 | 2.144 gigaflops | Cray Y-MP/832 | Cray Research | - | - | - | - | - | - | - | - | - | - | - | - | - | - | - |
| 1990–1991 | 4.0 gigaflops (measured) | Fujitsu VP2000 | Fujitsu | - | - | - | - | - | - | - | - | - | - | - | - | - | - | - |
| 1992 | 20.0 gigaflops | NEC SX-3/44 | NEC | - | - | - | - | - | - | - | - | - | - | - | - | - | - | - |
| June 1993 | 59.7 gigaflops | Thinking Machines CM-5/1024 | Thinking Machines Corporation (TMC) | Danny Hillis | - | - | - | - | - | - | - | - | - | - | - | - | - | -
|
References
[edit]- ^ Dongarra, Jack J.; Moler, Cleve B.; Bunch, James R.; Stewart, Gilbert W (1979). LINPACK Users' Guide. SIAM. ISBN 9780898711721.
- ^ Matlis, Jan (2005-05-30). "Sidebar: The Linpack Benchmark". ComputerWorld.
- ^ Markoff, John (1991-09-22). "Technology; Measuring How Fast Computers Really Are". New York Times.
 | This computer-engineering-related article is a stub. You can help Wikipedia by expanding it. |