From WikiChip
Difference between revisions of "mathstar/arrix"
Line 18: | Line 18: | ||
| proc = 130 nm | | proc = 130 nm | ||
| proc 2 = 90 nm | | proc 2 = 90 nm | ||
− | |||
− | |||
| tech = CMOS | | tech = CMOS | ||
| clock min = 400 MHz | | clock min = 400 MHz |
Revision as of 23:11, 27 June 2016
Arrix | |
Developer | MathStar |
Manufacturer | TSMC |
Type | Programmable logic device |
Introduction | September 25, 2006 (announced) November 29, 2006 (launch) |
Production | 2006-2009 |
Word size | 16 bit 2 octets
4 nibbles |
Process | 130 nm 0.13 μm , 90 nm1.3e-4 mm 0.09 μm
9.0e-5 mm |
Technology | CMOS |
Clock | 400 MHz-1,000 MHz |
Package | FCBGA-896 |
Socket | FCBGA-896 |
Succession | |
← | |
Builder |
Arrix was a family of field-programmable object array designed by MathStar and introduced in late 2006. MathStar continued manufacturing this until 2009.
Contents
Architecture
- Main article: field-programmable object array
Arrix is a family of field-programmable object array (FPOA). The chips can operate, depending on model, from 400 MHz to 1 GHz. At the time their performance exceeded similar budget FPGAs. Arrix are a grid of 400 logic objects. Additionally the chips also feature:
- 2x Bi-directional 500MHz DDR 16-bit LVDS ports
- 96x GPIO (synchronously or asynchronously at up to 100 MHz)
- 12x banks of 500MHz internal SRAM memory banks
- 2x 266 MHz 36-bit DDR (72-bits per cycle) RLDRAM II controllers
The chips are arranged as a grid of 400 logic objects made up of:
- 256x ALU
- 16 bit data, 5 bits control, 32 operations
- 16-bit add/sub/shift/rotate/AND/OR/XOR
- circuitry for cascading status bits for making larger word
- 80x RFs
- 128 Byte, dualport RAM or FIFO
- 64x MACx
- 16x16 single clock cycle multiplier
- 32 bit intermediate result, signed or unsigned
- 40 bit accumulator, 256 accumulations before overflow
- 12x Internal RAM banks
- 768 x 76 bits
Members
MOA2400D Series
Model | Frequency | OBjects | Package | Temp Range |
---|---|---|---|---|
MOA2400D-10 | 1,000 MHz | 400 | FCBGA-896 | Commercial |
MOA2400D-10R | 1,000 MHz | 400 | FCBGA-896 RoHS | Commercial |
MOA2400D-09 | 900 MHz | 400 | FCBGA-896 | Industrial |
MOA2400D-09R | 900 MHz | 400 | FCBGA-896 RoHS | Industrial |
MOA2400D-08 | 800 MHz | 400 | FCBGA-896 | Industrial |
MOA2400D-08R | 800 MHz | 400 | FCBGA-896 RoHS | Industrial |
MOA2400D-06 | 600 MHz | 400 | FCBGA-896 | Industrial |
MOA2400D-06R | 600 MHz | 400 | FCBGA-896 RoHS | Industrial |
MOA2400D-04 | 400 MHz | 400 | FCBGA-896 | Commercial |
MOA2400D-04R | 400 MHz | 400 | FCBGA-896 RoHS | Commercial |
Documents
Product Brief
- Arrix Family Product Brief, April 2007
- Arrix Family Product Brief, August 2006
- MOA1400D Product Brief, February 2006
IP Cores
- Color Space Conversion (RGB to YCRCB) for Arrix FPOA
- Flat Field Correction for Arrix FPOA
- JPEG 2000 Encoder for Arrix FPOA
- MPEG2 Multi-channel Decoder for FPOA
Software
See also
- field-programmable object array
- Ambric's Am2000
Facts about "Arrix - MathStar"
designer | MathStar + |
first announced | September 25, 2006 + |
first launched | November 29, 2006 + |
full page name | mathstar/arrix + |
instance of | integrated circuit family + |
main designer | MathStar + |
manufacturer | TSMC + |
name | Arrix + |
package | FCBGA-896 + |
process | 130 nm (0.13 μm, 1.3e-4 mm) + and 90 nm (0.09 μm, 9.0e-5 mm) + |
socket | FCBGA-896 + |
technology | CMOS + |
word size | 16 bit (2 octets, 4 nibbles) + |