|
|
|
|
|
|
 |
|||
|
ARM Announces Vector Floating-Point Coprocessor for ARM9E-S Cores
ARM Announces Vector Floating-Point Coprocessor for ARM9E-S Cores New Floating-Point Coprocessor from ARM targets automotive and industrial control applications EMBEDDED PROCESSOR FORUM, San Jose, Calif. - June 12, 2001 - ARM [(LSE:ARM); (Nasdaq:ARMHY)], the industry's leading provider of 16/32-bit embedded RISC microprocessor solutions, today announced a new floating-point coprocessor, the ARM® VFP9-S[tm] vector floating-point coprocessor. The ARM VFP9-S coprocessor is a high-performance, low-power and small die area floating-point solution for the ARM9E-S[tm] family of synthesizable microprocessor cores and follows on from the previously announced ARM VFP10[tm] coprocessor for the ARM10[tm] family of microprocessor cores. Floating-point coprocessors are particularly suited to applications that have dynamic ranges that are greater than the range of a standard microprocessor core. The ARM VFP9-S coprocessor is initially targeted at the automotive and industrial control markets where large data ranges and small data values are predominant. For example, the ARM VFP9-S coprocessor can be used in car engine management systems where sensors measure tiny amounts of oxygen in the combustion cylinder. In order to accurately control fuel injection, the engine management processor must be able to quickly calculate the volume of fuel to be injected. The ARM VFP9-S coprocessor enables the system to quickly calculate the values needed and can then send this information to the fuel injection system in microseconds. Compared to standard integer emulation routines, the ARM VFP9-S coprocessor provides fast processing owing to its use of low-latency, single-cycle throughput, single and double precision operations. The ARM VFP9-S coprocessor includes novel short vector operations that allow performance processing with reduced instruction bandwidth. This delivers high- processing throughput with overlapping load and store operations which allows a single precision floating-point Finite Impulse Response (FIR) tap to be computed in less than 1.5 cycles. The ARM VFP9-S coprocessor consists of a maximum of approximately 95,000 gates, compared with an ARM9E[tm] integer microprocessor core that has around 65,000 gates. In a typical 0.18u implementation of an ARM966E-S[tm] with 16kb of instruction tightly coupled memory and 16kb of data tightly coupled memory, the total die size is around 4mm2, and the ARM VFP9-S coprocessor would add no more than an additional 1.5mm2 of die size. "In the past, floating-point processors have been considered an expensive addition to an SoC. However, with shrinking geometries and the associated manufacturing cost reductions, the cost of adding floating-point is becoming negligible," said Paul Garden, Product Marketing manager, ARM. "ARM has invested great efforts in producing the vector floating-point coprocessor for the ARM9E family and this product further demonstrates our commitment to providing high-performance, low power and small area floating-point solutions. Our partners will be able to offer greatly enhanced SoCs that will allow designers to build products with better calculating power for applications that demand superior dynamic range." Availability About ARM - ends -
|
|
|
|
|
|
|
|
Home | Feedback | Register | Site Map |
|
|
|
All material on this site Copyright © 2017 Design And Reuse S.A. All rights reserved. |