Vector Processor Rake-Finger Operation Execution Unit

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current vector processors require multiple instructions and increased clock cycles to perform rake-finger operations, leading to higher power consumption and inefficiency due to the need for separate instructions for rotation, FIR filtering, and despreading operations.

Innovation Solution

A vector execution unit is configured to perform rake-finger operations in a single instruction, integrating rotation, FIR filtering, and despreading without intermediate storage in the register, reducing clock cycles and power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If separate instructions are used for rotation, FIR filtering, and despreading operations, then each operation can be performed with dedicated hardware, but the number of clock cycles and power consumption increase

Engineering Contradiction:
Improveoperation accuracyVSAvoidclock cycles
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent combines rotation, FIR filtering, and despreading operations into a single unified instruction that executes all three operations in one clock cycle. The execution unit is designed with integrated functional blocks that perform these operations sequentially without requiring separate instructions, thereby reducing the total number of clock cycles from three to one while maintaining operation accuracy through dedicated hardware paths for each function.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The execution unit is designed as a universal processor capable of performing multiple functions (rotation, FIR filtering, despreading) within a single instruction cycle. This multi-functional design allows the same hardware unit to handle different signal processing operations without requiring separate dedicated hardware units for each operation, thus reducing both clock cycles and power consumption while maintaining reliability.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Reliability

If separate instructions are used for rotation, FIR filtering, and despreading operations, then each operation can be performed with dedicated hardware, but power consumption increases

Engineering Contradiction:
Improveoperation accuracyVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent merges rotation, FIR filtering, and despreading operations into a single unified instruction that executes all three operations in one clock cycle. By combining these operations, the patent reduces the number of times data must be loaded from and stored to memory, and reduces the number of instruction fetch and decode cycles, thereby significantly reducing power consumption while maintaining operation accuracy through dedicated hardware paths for each function within the unified instruction.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The unified instruction enables continuous processing where the output of one operation (rotation) is immediately fed to the next operation (FIR filtering) and then to the final operation (despreading) without interruption or memory access in between. This continuous action eliminates idle cycles and repeated memory access operations, reducing power consumption while ensuring each operation completes accurately with dedicated hardware support.

Inventive Principle:
Principle #20Continuity of useful action

3Reliability

If intermediate storage in register is used between operations, then data can be preserved between steps, but processing efficiency decreases

Engineering Contradiction:
Improvedata integrityVSAvoidprocessing efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent combines rotation, FIR filtering, and despreading operations into a single unified instruction that executes all three operations in one clock cycle with direct data flow between operations. This eliminates the need for intermediate storage in registers between operations, as the data flows directly through the execution unit's internal pathways from rotation to FIR filtering to despreading, thereby maintaining data integrity while significantly improving processing efficiency by removing memory access overhead.

Inventive Principle:
Principle #5Merging (Combining)

4Adaptability or versatility

If multiple instructions are used for rake-finger operations, then each operation can be controlled independently, but device complexity increases

Engineering Contradiction:
Improveoperation control flexibilityVSAvoidinstruction sequence complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The execution unit is designed as a universal processor capable of performing multiple functions (rotation, FIR filtering, despreading) within a single instruction cycle. This multi-functional design simplifies the instruction set by eliminating the need for separate instructions for each operation, thereby reducing device complexity and instruction sequence complexity while maintaining adaptability and versatility through configurable parameters within the unified instruction that allow independent control of each operation's characteristics.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3100158B1An instruction and method for fused rake-finger operation on a vector processor
Publication Date: 2018.08.29 QUALCOMM INC
  • EP3100158B1 patent drawingFigure 1
  • EP3100158B1 patent drawingFigure 2
  • EP3100158B1 patent drawingFigure 3

AI summary

Systems and methods for performing a rake-finger operation on a vector processor are described herein. In one embodiment, a method for rake-finger processing comprises loading samples from a register into an execution unit, performing a rake-finger operation on the samples in the execution unit, and writing results from the rake-finger operation to the register. Performing the rake-finger operation comprises performing a finite impulse response (FIR) filter operation, and performing a despread operation, wherein filtered samples from the FIR filter operation are input to the despread operation without going through the register.