Silicon Graphics, Incorporated

Mips Technologies, Inc.

Cadence Design Systems, Inc.

Tensilica, Inc.

William A Huffman

Earl A Killian

Peter Y Hsu

Jeffrey Scott Kuskin

Henry Packard Moreton

Timothy J Van Hook

Joseph T Scanlon

Himanshu A Sanghavi

John Brennan

Joseph P Bratt

Gregory Michael Thorson

Steven C Miller

Randal Steven Passint

James Chinmo Kim

Daniel E Lenoski

Paul Kingsley Rodman

Michael S Woodacre

Eileen Margaret Peters Long

David H C Chen

Darin Stamenov Petkov

Peng Tu

Nupur B Andrews

Man K Tang

Steve J Ciavaglia

James Robert Kennedy

Fei Sun

Steven Lee Scott

Dror E Maydan

Mark Ronald Sikkink

Chandra Shekhar Joshi

Sachin P Ghanekar

Daniel L Kunkel

Monica R Nofal

Nan Ma

Michael L Anderson

Vernon W Swanson

Ram Gupta

Perter Hsu

Steve Ciavagia

Peter Yan-Tek Hsu

Susan Garcia

A fused-multiply-add system is disclosed. The fused-multiply-add system includes a multiplier to multiply first and second operands and to provide at least one product. The fused-multiply-add system also includes an alignment shifter for aligning a third operand with the at least one product to provide an aligned third operand. The fused-multiply-add system also includes an adder and a subtractor coupled to the multiplier and the alignment shifter for performing two asymmetrical additions in parallel paths. The fused-multiply-add system also includes at least one leading zero counter for counting a number of leading zero bits provided by at least one of the adder and the subtractor to provide at least one normalization shift amount. Finally, the fused-multiply-add system includes a multiplexer coupled to the adder and the subtractor for providing an appropriate output based upon a sign bit.

Optimized fused-multiply-add method and system

A method is provided to narrow down the exponent range throughout most part of the division and square root calculations, to make both software assistance and precision extension unnecessary. The method adjusts the exponent at the end of the calculation to reach IEEE-754 results.

Accurate floating-point calculation method and device

A microprocessor includes a memory interface to obtain data envelopes of a first length, and control logic to implement an instruction to load an initial data envelope of a stream of data values into a buffer, each data value having a second length shorter than the first length, the stream of data values being disposed across successive data envelopes at the memory interface. Another instruction merges current contents of the buffer and the memory interface such that each invocation loads one of the data values into a first register, and moves at least a remainder of the current contents of the memory interface into the buffer for use in a successive invocation. Another instruction loads a reversed representation of a set of data values obtained via the memory interface into a second register. Another instruction implements an FIR computation including a SIMD operation involving multiple data values of the stream and the reversed representation.

Audio digital signal processor

A technique that improves both processor performance and associated data bandwidth through user-defined interfaces that can be added to a configurable and extensible microprocessor core. These interfaces can be used to communicate status or control information and to achieve synchronization between the processor and any external device including other processors. These interfaces can also be used to achieve data transfer at the rate of one data element per interface in every clock cycle. This technique makes it possible to design multiprocessor SOC systems with high-speed data transfer between processors without using the memory subsystem. Such a system and design methodology offers a complete shift from the standard bus-based architecture and allows designers to treat processors more like true computational units, so that designers can more effectively utilize programmable solutions rather than design dedicated hardware. This can have dramatic effects not only in the performance and bandwidth achieved by designs, but also in the time to market and reuse of such designs.

Method and apparatus for providing user-defined interfaces for a configurable processor

The present invention provides extended precision in SIMD arithmetic operations in a processor having a register file and an accumulator. A first set of data elements and a second set of data elements are loaded into first and second vector registers, respectively. Each data element comprises N bits. Next, an arithmetic instruction is fetched from memory. The arithmetic instruction is decoded. Then, the first vector register and the second vector register are read from the register file. The present invention executes the arithmetic instruction on corresponding data elements in the first and second vector registers. The resulting element of the execution is then written into the accumulator. Then, the resulting element is transformed into an N-bit width element and written into a third register for further operation or storage in memory. The transformation of the resulting element can include, for example, rounding, clamping, and/or shifting the element.

Providing extended precision in SIMD vector arithmetic operations

The present invention provides alignment and ordering of vector elements for SIMD processing. In the alignment of vector elements for SIMD processing, one vector is loaded from a memory unit into a first register and another vector is loaded from the memory unit into a second register. The first vector contains a first byte of an aligned vector to be generated. Then, a starting byte specifying the first byte of an aligned vector is determined. Next, a vector is extracted from the first register and the second register beginning from the first bit in the first byte of the first register continuing through the bits in the second register. Finally, the extracted vector is replicated into a third register such that the third register contains a plurality of elements aligned for SIMD processing. In the ordering of vector elements for SIMD processing, a first vector is loaded from a memory unit into a first register and a second vector is loaded from the memory unit into a second register. Then, a subset of elements are selected from the first register and the second register. The elements from the subset are then replicated into the elements in the third register in a particular order suitable for subsequent SIMD vector processing.

Alignment and ordering of vector elements for single instruction multiple data processing

A method and system for managing memory in a multiprocessor system includes defining the plurality of processor coherence domains within a system coherence domain of the multiprocessor system. The processor coherence domains each include a plurality of processors and a processor memory. Shared access to data in the processor memory of each processor coherence domain is provided only to elements of the multiprocessor system within the processor coherence domain. Non-shared access to data in the processor memory of each processor coherence domain is provided to elements of the multiprocessor system within and outside of the processor coherence domain.

Method and system for managing memory in a multiprocessor system

A memory access arbitration scheme is provided where transactions to a shared memory are stored in an arbitration queue. A collapsible queuing structure and method are provided, such that once a transaction is serviced, higher order entries ripple down in the queue to make room for new entries while maintaining an oldest to newest relationship among the queue entries. A queuing circuit having a plurality of registers interconnected by 2:1 multiplexers is also provided. The circuit is arranged such that each register receives either its own current contents or the contents of a higher order register during each register write cycle.

Queue circuit and method for memory arbitration employing same

A processor can achieve high code density while allowing higher performance than existing architectures, particularly for Digital Signal Processing (DSP) applications. In accordance with one aspect, the processor supports three possible instruction sizes while maintaining the simplicity of programming and allowing efficient physical implementation. Most of the application code can be encoded using two sets of narrow size instructions to achieve high code density. Adding a third (and larger, i.e. VLIW) instruction size allows the architecture to encode multiple operations per instruction for the performance critical section of the code. Further, each operation of the VLIW format instruction can optionally be a SIMD operation that operates upon vector data. A scheme for the optimal utilization (highest achievable performance for the given amount of hardware) of multiply-accumulate (MAC) hardware is also provided.

Vector co-processor for configurable and extensible processor architecture

The present invention provides an interface and method for synchronous data transfer between domains clocked at different frequencies. The interface includes a first latch for receiving data from a first domain clocked at one frequency when the first latch is selected and a second latch for receiving data from the first domain when the second latch is selected. A third latch is provided for transferring data from either the first latch or the second latch to the second domain when the second domain is clocked.

Interface for synchronous data transfer between domains clocked at different frequencies

Load/store operation of memory misaligned vector data using alignment register storing realigned data portion for combining with remaining portion

The present invention provides extended precision in SIMD arithmetic operations in a processor having a register file and an accumulator. A first set of data elements and a second set of data elements are loaded into a first vector register and a second vector register, respectively. Each data element comprises N bits. Next, an arithmetic instruction is fetched from memory. The arithmetic instruction is decoded. Then, a first vector register and a second vector register are read from the register file. The present invention then executes the arithmetic instruction on corresponding data elements in the first and second vector registers. The result of the execution is then written into the accumulator. Then, each element in the accumulator is transformed into an N-bit width element and stored into the memory.

Method for providing extended precision in SIMD vector arithmetic operations

Growing community of inventors

Los Gatos, CA, United States of America

William A Huffman