Tiva TM C Series TM4C123x MCUs

Transcription

1 Tiva TM C Series TM4C123x MCUs Floating-Point Performance, Analog Integration and Best-in-Class Low-Power Consumption Floating Point Unit Ashish Ahuja Acknowledgements: Alex Bestavros

2 Agenda IEEE 754 Standard ARM Cortex TM -M4F Floating Point Unit Modes of Operation & Registers Instruction Set CMSIS DSP Library Getting Started DSP Library Functions DSP Library Performance TivaWare TM for C Series Peripheral Driver Library Sensor Library 2

3 Agenda IEEE 754 Standard ARM Cortex TM -M4F Floating Point Unit Modes of Operation & Registers Instruction Set CMSIS DSP Library Getting Started DSP Library Functions DSP Library Performance TivaWare TM for C Series Peripheral Driver Library Sensor Library 3

4 Introduction Floating-Point Floating-point is a way to represent real ( ) numbers on computers IEEE floating-point formats: Half (16-bit) Single (32-bit) Double (64-bit) Quadruple (128-bit) 4

5 IEEE Standard 754 Bit X Symbol Sign (s) Exponent (e) Fraction (f) 1 bit 8 bits 23 bits Decimal Value = (-1) s (1+f) 2 e-bias where: f = [(b -i )2 -i ] i ϵ (1,23) bias = 127 for single precision floating-point Symbol s e f Example X sign = (-1) 0 exponent = [ ] 2 = [134] 10 fraction = [ ] 2 = [ ] 10 = [1] 10 Decimal Value = (-1) s x (1+f) x 2 e-bias = [1] 10 x ([1] 10 + [ ] 10 ) x [ ] 10 = [ ] 10 x 128 = [ ] 10 5

6 ARM Cortex -M4F What s new? ANSI/IEEE Std compliant, IEEE Standard for Binary Floating-Point Arithmetic compliant Hardware support for conversion, addition, subtraction, multiplication with optional accumulate, division, and square-root Advantages Higher precision in control loops (can save energy in motors) Faster signal processing, lower latency & faster response Easier to integrate with tools such as MATLAB and LabVIEW Cortex M4F Block Diagram Image: 6

7 Floating Point Unit Combined multiply and accumulate (MAC) functions for increased precision 32 x 32 multiply accumulate (MAC) with 64-bit result Example End Applications Data compression, sensor array processing, statistical signal processing, multi-band graphic equalizers Measure, filter, compress real-world analog signals Control systems such as motor control, solar inverters, lighting control Digital signal control applications that demand an efficient, easy-to-use blend of control and signal processing capabilities Multiband Equalizers HVAC, Pump, Inverter, Compressor Motor Automation and Motor Control Solar Inverters Lighting Control Example End Applications 7

8 Floating Point Unit Conversions between fixed-point and floating-point data formats, and floating-point constant instructions 32-bit instructions for single-precision data-processing operations Single instruction, multiple data (SIMD) for 16-bit data types Decoupled three-stage pipeline Hardware support for denormals and all IEEE rounding modes Supports saturation math FPU may be disabled to conserve power 8

9 ISA Compatibility 9

10 Agenda IEEE 754 Standard ARM Cortex TM -M4F Floating Point Unit Modes of Operation & Registers Instruction Set CMSIS DSP Library Getting Started DSP Library Functions DSP Library Performance TivaWare TM for C Series Peripheral Driver Library Sensor Library 10

11 Modes of Operation Three modes of operation are provided to accommodate a variety of applications Full Compliance Mode FPU processes all operations according to the IEEE 754 standard in hardware Flush-to-Zero (FZ) Mode A tiny result, where the destination precision is smaller in magnitude than the minimum normal value before rounding, is replaced with a zero Setting the FZ bit in the FPSC register enables this mode Default NaN (DN) Mode The result of any arithmetic data processing operation that involves an input NaN, or that generates a NaN result, returns the default NaN. Setting the DN bit in the FPSC register enables the default NaN mode Example of NaN being returned 0 / 0 = NaN 11

12 Register Bank The FPU provides an extension register file containing 32 single-precision registers. Sixteen 64-bit double-word registers, D0-D15 Thirty-two 32-bit single-word registers, S0-S31 A combination of registers from the above views Mapping between registers S<2n> maps to least significant half of D<n> S<2n+1> maps to most significant half of D<n> Compliers can use floating-point registers for floating-point or non-floating-point code. S0 S1 S2 S3 S4 S5 S30 S31 D0 D1 D2 D15 FPU Register Bank 12

13 Agenda IEEE 754 Standard ARM Cortex TM -M4F Floating Point Unit Modes of Operation & Registers Instruction Set CMSIS DSP Library Getting Started DSP Library Functions DSP Library Performance TivaWare TM for C Series Peripheral Driver Library Sensor Library 13

14 Instruction Set Operation Description Assembler Cycles Absolute Value of float VABS.F32 1 Addition floating point VADD.F32 1 Compare float with register or zero VCMP.F32 1 Convert between integer, fixed-point, half precision and float VCVT.F32 1 Divide floating-point VDIV.F32 14 Move immediate/ float to float-register VMOV 1 Multiply float VMUL.32 1 Negate float VNEG.F32 1 Pop float registers from stack VPOP.32 1+N Push float registers to stack VPUSH.32 1+N Square-root of float VSQRT.F32 14 Store single float register VSTR.32 2 Subtract float VSUB.F32 1 Note: Please refer ARM DDI 0439C ID Cortex M4 Technical Reference Manual for the complete instruction set. N = number of registers. 14

15 Agenda IEEE 754 Standard ARM Cortex TM -M4F Floating Point Unit Modes of Operation & Registers Instruction Set CMSIS DSP Library Getting Started DSP Library Functions DSP Library Performance TivaWare TM for C Series Peripheral Driver Library Sensor Library 15

16 HW CMSIS USER CMSIS DSP Library Cortex TM Microcontroller Software Interface Standard (CMSIS) TM Provides a single standard across all Cortex-M processor Enables code re-use across software projects Reduces time-to-market for new embedded applications CMSIS DSP Library A suite of common signal processing functions for use on Cortex-M processor based devices Includes separate functions for operating on 8-bit, 16-bit, 32-bit integer and 32-bit floating-point values. Developed & tested with MDK-ARM. Support for IAR, GCC & CCS coming soon Supports single public header file arm_math.h for Cortex- M4/M3/M0 with little endian and big endian Application Code CMSIS DSP Lib CMSIS Core Cortex MCU Core Source: CMSIS Documentation CMSIS\Documentation\DSP_Lib\html\index.html CMSIS DSP Library 16

17 Example CMSIS DSP Library Functions Filtering Transform & Controller Math Matrix & Interpolation Statistics Convolution Complex FFT & CIFFT Absolute Value Addition Mean Correlation Finite Impulse Response (FIR) Decimator Discrete Cosine Transform & IDCT Vector Addition Inverse Sum of Squares Real FFT & RIFFT Vector Multiplication Vector Multiplication Root Mean Square FIR Interpolator PID Control Dot Product Scalar Multiplication Standard Deviation FIR Lattice Sine & Cosine Negate Subtraction Variance FIR Sparse IIR Lattice Least Mean Square (LMS) Filter Normalized LMS Filter Vector Clark Transform (VCT) Vector Park Transform (VPT) IVCT & IVPT Square Root Vector & Scalar Multiplication Transpose Max Complex Conjugate Linear Interpolation Min Complex/ Real Multiplication Bilinear Interpolation Source: CMSIS Documentation CMSIS\Documentation\DSP_Lib\html\modules.html 17

18 CMSIS DSP Library Performance Cortex TM -M4 SIMD + FPU vs. Cortex M3 Fixed-point ~ 2x faster Floating-point ~ 10x faster DSP Library Benchmark: Cortex M3 v/s Cortex M4 Source: ARM CMSIS Partner Meeting Embedded World, Reinhard Keil 18

19 Using CMSIS Examples with CCS Examples Maximum, Minimum, Mean, Standard Deviation, Variance and Matrix functions Complex FFT, Complex-by-Complex Multiplication functions Multiply and Add functions to perform the dot product functions FIR lowpass filter function Complex FFT, Complex Magnitude, and Maximum functions Biquad cascade functions Matrix Transpose, Matrix Multiplication, and Matrix Inverse functions Normalized LMS Filter, Finite Impulse Response (FIR) Filter, and basic math functions Cosine, Sine, Vector Multiplication, and Vector Addition functions Source: CMSIS Documentation CMSIS\Documentation\DSP_Lib\html\examples.html Application Note click on Technical Docs tab 19

20 Agenda IEEE 754 Standard ARM Cortex TM -M4F Floating Point Unit Modes of Operation & Registers Instruction Set CMSIS DSP Library Getting Started DSP Library Functions DSP Library Performance TivaWare TM for C Series Peripheral Driver Library Sensor Library 20

21 TivaWare TM for C Series Over 250 APIs for graphical display support Over 175 APIs for USB device, USB host, or USB On-The-Go (OTG) applications Over 50 APIs for accel, gyro, and magnetometer sensor fusion. Provides open source Ethernet and RTOS options Provides an extensive array of third-party packages Graphics Library USB Library Sensor Library Peripheral Driver Library Open Source RTOS Boot Loader and In-System Programming Support Open Source Stacks TivaWare TM for C Series Software Utilities: Checksum Security Code Examples Third- Party Packages Over 700 APIs for peripheral initialization and control functions Provides in-field programmability Over 150 APIs providing optimized commonly used functions such as CRC checking and AES tables Over 50 apps providing an extensive array of source code samples 21

22 Peripheral Driver Library TivaWare TM for C Series FREE download from TI s website Visit Driver library (driverlib) APIs Function & Marco Declaration: fpu.h Function Definition: fpu.c C:\ti\TivaWare_C_Series-1.0\driverlib Code example(s) for FPU C:\ti\TivaWare_C_Series-1.0\examples List of FPU API s void FPUDisable (void) void FPUEnable (void) void FPUFlushToZeroModeSet (uint32_t ui32mode) void FPUHalfPrecisionModeSet (uint32_t ui32mode) void FPULazyStackingEnable (void) void FPUNaNModeSet (uint32_t ui32mode) void FPURoundingModeSet (uint32_t ui32mode) void FPUStackingDisable (void) void FPUStackingEnable (void) User guide C:\ti\TivaWare_C_Series-1.0\docs 22

23 Sensor Library USER MCU Sensor library is: A set of drivers for I 2 C connected sensors Three layers: Transport, Sensor and Processing Pre-complied on multiple tool chains Extendable to a variety of sensors Application Code All sensors provide a GetFloat() function Returns data in standard international units Reduces time-to-market for new embedded applications Utilizes the FPU capability of TM4C123x MCUs Keep math simple & true to textbook formulas TivaWare TM Sensor Lib Peripheral Driver Lib Sensors HW Example APIs: Initialize the I 2 C driver I2CMInit() Initialize the TMP006 TMP006Init() Read data from the TMP006 TMP006DataRead() CM4 Core with FPU How Sensor Library Fits in Your Application? 23

24 SensorHub Booster Pack Add sensor aggregation capabilities to your designs Motion tracking: gyro, accelerometer & compass Environmental monitoring: pressure, humidity, ambient temperature, ambient light & infrared sensor Designed for Tiva C Series LaunchPad Supports multiple tool chains CCS, Keil, IAR & Sourcery Code Bench Get started with Sensor Fusion technology in 10 minutes or less Air Mouse Example: USB Composite HID keyboard & mouse Viterbi-like gestures Wireless ready with RemoTI Tiva TM SensorHub Booster Pack Onboard temperature, pressure, motion, humidity, ambient light & IR sensors Tiva TM Launchpad EK-TM4C123GXL Low-Cost Versatile Evaluation Platform The Sensibility of Sensors 24

25 More Information Technical Documents : click on Technical Docs tab Order Samples : click on Products tab Tools & Software : click on Tools & Software tab Technical Support : click on Community & Support tab Tiva TM Launchpad EK-TM4C123GXL Low-Cost Versatile Evaluation Platform 25