practical insights on selecting the right processor for your application

PRACTICAL INSIGHTS ON SELECTING THE RIGHT PROCESSOR FOR YOUR APPLICATION

It is a bit more complicated than

• • • •

Need a More Structured Approach – 10,000 Ft view Mission Requirements Operating Environment Component Availability Experience and Heritage

Harsh Environment – Getting There Launch • Acoustic Loading • Vibration • Random and Sine • Shock • Depressurisation – Venting • Electro Magnetic Interference

Harsh Environment – Operating there Operation • Temperature • Radiation • Reliability • Long Operating life • Electro Magnetic Interference

What are some of the Processors are Available Processor

Architecture

MIPS

Comments

LEON 3 FT ASIC Dual

SPARC V8

200

100 MHz

LEON 3 FT AISC Single

SPARC V8

75

66 MHz

LEON 3 FT FPGA RTAX

SPARC V8

20

25 MHz

RAD 750

Power PC

400

200 MHz

PC 7448

Power PC

3000

1267 MHz

ARM7-TDMI

ARM7

36

40 MHz Cubesat Not Rad Hard

Mongoose-V

MIPS R3000

8

10 - 15 MHz

Redundancy Architecture

Performance

Power Management

PCB Mounting

Processor Selection

Radiation

Operating Systems

Communication IF’s

Debugging Tools

Performance • Key Driver for achieving mission requirements • Million Instructions Per Second • Number of Instructions per Clock Cycle – Scalar / Vector / Superscalar • Floating Point Unit • Multicore • Hardware Acceleration / Peripherals • Direct Memory Access • Memory Bandwidth

Power Management • Drives the thermal management of the payload – Conversion losses • How many voltage rails are required – this complicates the power architecture - increasing losses and decreasing MTBF, Weight etc. • Does the processor have low power modes • Can we reduce the core frequency if desired – Dynamic Power Management • Voltage scaling

Communication IF’s

• How our equipment communicates externally • On Chip Peripherals for communication • Simple UART, SPI, I2C, Ethernet, PCIe • DMA Provided to reduce load on processor • Is the bandwidth sufficient for the application requirements

Operating Systems • Operating system enables scheduling of processes and processor resources • Application dependant • Hard Real Time • Soft Real Time • Do we need an operating system – Can it be achieved using a bare metal solution • Certification – DO178B, Sil3 • Examples in use in space missions • RTEMS, VxWorks, uC/Osii, eCOS, Linux (RTAI and Xluna)

Debugging Tools • Development Environment • Can we debug at the system level • Can we examine the processor state without impacting programme execution • Can we profile the application • Can we trace where the processor has been • Is it possible to analyse communication protocols on communication interfaces

PCB Mounting • • • •

Often the most over looked aspect Is it BGA, CGA or QFP Impacts reliability – BGA balls fail, Column Crack, How it mounts has a large impact on how we perform thermal management. • Number of IO required • Is the mounting method currently qualified for your manufacturing facility

Radiation Tolerance • Is the device latch up immune • Can it experience SEUs • What is the maximum Total Ionising Dose • Does the processor contain fault tolerant circuits e.g ECC / EDAC to correct for faults • Is there a watchdog provided for lock ups – often ths needs to be independent.

Redundancy Architecture

• Is Redundancy required ? • Inter or Intra module • Hot or Cold spared • Cost and Weight driver • Redundancy approach for SW • External / internal watchdog provided for SEFI

SW Techniques • For Space Applications there are some mitigation strategies we can follow • Redundancy at the Instruction level • Redundancy at task level • Redundancy at application level • ESA Handbook – Techniques for Radiation Effects Mitigation in ASICs and FPGAs is provides a good understanding for all engineers – does include a section on processors.

HW Design Cost

Component Cost

SW Design Costs

Technical Risks

Processor Selection

Programme Risks

Schedule

Design Reuse

Heritage & TRL

Costs • Goal is meeting mission requirements within the available budget. • Number of Cost drivers • Bill of Materials • Engineering Life Cycle costs – design reviews • Design / Verification / Integration costs for HW and SW • Certification programmes

Risks – Technical and Programmatic

• Any Programme will have a number of risks technical and programmatic • Technical risks relate to engineering aspects i.e. Quality • Programmatic relate to cost and timescale • Process • Identify candidate risks • Identify applicable standards, laws, agreements and company policies • Analyse Risks / Plan Risk Management / Evaluate • Risk Profile and Risk Strategy will result

Schedule • Not necessarily the domain of the project manager but also the Work Package Manager. • All projects want to deliver • On Schedule, On Cost, On Quality • Schedule will define the engineering and programmatic milestones – enables progress to be tracked • Crucially it will enable past performance to be used as an indicator of future performance • If you are slipping 6 months in 12 months for example, without addressing the issues via a recovery programme future performance will continue to slip 6 months in 12.

Heritage & TRL • Does the processor have any heritage in the application. Reduces risk of implementation or qualification issues • Technology Readiness Level • Determines the readiness level of the technology to be used in your application • ESA have 9 levels from concept to proven and used on a mission. • TRL development plan – shows the path taken to prove the system to the required level.

Design Reuse

• Enables faster time to market • Requires a development rules to enable reuse • Increases quality as it is used across several projects • Enables the estimation of new projects to be undertaken easier •

Conclusion • Technical and Programmatic considerations have been presented • Criteria's use depends upon your application • Thank you for listening Questions ?