amd at itc 2014

AMD AND OPTIMAL+ PARTNERING FOR TRANSFORMATION

AND MARGIN IMPROVEMENT

CARL BOWEN OCTOBER 2014

2 | AMD & OPTIMAL + | OCTOBER 2014

GLOBAL FOOTPRINT

Bangalore

Boxborough

Markham

Penang

Singapore

Sunnyvale

Austin Suzhou Shanghai Orlando

Beijing

Shenzhen

Established in 1969 and headquartered in Sunnyvale, California and Austin, Texas

2013 revenues of $5.3 billion

Operations in 31 countries, with more than 50 locations, including more than a dozen R&D facilities

Approximately 10,000 employees worldwide


Invented 64-bit X86

First multi-core CPU

Inventors of the APU

Founded HSA

2003 2004 2011 2012

BUILT ON A FOUNDATION OF INNOVATION & LEADERSHIP

Highest performing graphics cards

World record overclocking CPU designs

FIRST Gigapixel display uses 400 screens

Gaming supremacy inside every major next generation gaming console

First to design side-by-side 64-bit ARM® and x86-processors.


AGENDA

Landscape

Process

Outcomes and Next Steps

Challenge

Landscape


LANDSCAPE YIELD MANAGEMENT AT AMD

FORECAST: Signal the future

ATTACK: Drive the ceiling

PROTECT: Manage the floor

1 2 3 4 5 6 7 8 9 10 11 12

Yie

ld

time

FORECAST

“Attack”

“Protect”

FORECAST: Signal the future

ATTACK: Drive the ceiling

PROTECT: Manage the floor

• Focus on silicon characterization

• Optimized for “deep dive” unit level analysis

• Near real-time alert for excursions

• Isolate cause and direct corrective action

• Separate equipment from silicon contribution

Internal tools to address “ATTACK” dimension

Optimal+ to address “PROTECT” dimension


LANDSCAPE

Business Climate

• Reduced R&D footprint

• Gross margin pressure

• “Fabless” transition

• Immature process technology

Supply Chain

• Internal and external test facilities

• Assemble to demand strategy

• Steep product ramp leads to capacity constraints

INFLUENCES

Challenge


THE CHALLENGE

Historic NPI focus…

• NPI characterization and samples “delivery” approach vs. Operations

… leaves control gaps…

• We get “burned” with unplanned events that affect our supply

• Time to detect and react is unacceptable

• Difficult to separate equipment silicon causes

… presents opportunity

• Retest bin recovery optimization

• Simple test time optimization

• Capacity utilization

SOLUTIONS “PARADIGM”


CHALLENGE

Lack of standards

• Test data is “dirty”

Data transfers

problematic

• Time to data: 4-24 hours

No business intelligence

tools

• User expertise dependent

INTERNAL WEAKNESSES

Internal systems were recognized to be inadequate


CHALLENGE

Integrated data capture, transfer and presentation

• Time to visualize (includes data cleanse and augment)

Effective event detection methods

• rules covering Yield, SBL, Parameter Trend, Parameter Process Capability, Outlier Detection

Integrated UI

• Enables users to build, design and share customized templates for analysis

IDENTIFY THE NEED

Solution has to be more effective than what we already have

Process


PROCESS

Diverse Stakeholders

Product/ Test Engineering

Supply Chain/ Procurement

Factory (Operations)

Senior Management

Strongly held perceptions

We know our business

“Yes, but…”

Cheaper to build

STAKEHOLDER POSITIONS


PROCESS

Identify Champion

Know the Supporters and Resistors

Frequent “Face to Face” Communication

Stick to Facts and Data

Leverage Executive Sponsor

STAKEHOLDER MANAGEMENT

Manage perceptions!


PROCESS Cast a Vision • Integrated solution combining robust test data management with business intelligence to create “actionable data” and

enable users to drive improvement in yield and efficiency around test operations

• Differentiates between silicon, tester hardware and test program induces signals

• Near real-time monitor of yields with quick access to historical data

• Automated dashboards providing easy access to test data and its spatial representation

Methodology • Used to derive and implement process controls

• Engineers utilize “rules” to trigger end of run events

Engineers & Planning teams utilize dashboards to identify potential and gaps using “best in class” comparisons

• Action rules drive immediate response and correction of critical problems

• Dashboards and information rules drive definition and monitoring of continuous improvement projects

Benefits

• Improve yield, supply predictability and quality by reducing impact of excursions

• Improve capacity planning and leverage with suppliers by providing transparency in actual performance

• Improve engineering efficiency through improved business intelligence systems


PROCESS PHASES

PoC Pilot Decision Implementation J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J

11'Q3 11'Q4 12'Q1 12'Q2 12'Q3 12'Q4 13'Q1 13'Q2 13'Q3 13'Q4 14'Q1 14'Q2

Proof of Concept

•Demonstrate potential

Pilot

•Evaluate capability

Decision

•ROI

Implementation

•Extract value

Leverage a systematic process with continuous communication


PROCESS

Value Stream Value(s) Measure(s)

Product/ Yield Engineering

Product Cost & Reliability Yield, Silicon Quality, Hold Lot CT

Yield, SBL, Crit Param Cpk, Bin Recovery, Outlier Detection

Test Test Program & HW Stability Stability, Efficiency, Capability

Yield Gap (FTRd), IR, Site variation, OHT, Index Time, Bin Recovery

Planning Utilization & Feasibility Capacity Planning, Continuity of Supply

UPH, OHT, Effective Util, SBL

Factories Utilization & Stability Capacity Attainment, Supply Attainment

UPH, OHT, Effective Util, SBL

Procurement Price Negotiation Consumables Planning, Pricing Leverage

UPH, OHT, Effective Util, LB/ PC Usage

ESTABLISH “VALUE STREAMS” CONCEPT

Each Value Stream has a champion and a sponsor


PROCESS VALUE STREAM IMPLEMENTATION

Value

– Establish value stream from

practice from other companies.

– Establish measurement KPIs.

Define Preliminary Criteria

– Establish performance baseline

– Define excursion triggering criteria.

– Align with PDG/HWI/Factory

– Criteria setup in O+.

Working Model & review forum

– Define R&R for excursion. (Including

both NPI & product mature phase)

Trial Run

– System usage training

– Issue detect and feedback.

Production Data Collection / Optimization

– System / Working Model Optimization based on

Trial run result

– Finalize the working model

Define

Value

Define

Criteria

Establish

Working

Model

Training & Trial Run

W20 W23 W25 W27

Production

Data Collection

W39

In-house Deployment (w23-w24)

OSAT Deployment (w19-w25)

Basic User Training (w24) Extended Training (w29)

Monthly Read Point

Test Value

Q4

Apply Lean Six Sigma Principles

Outcomes


OUTCOMES SOME FACTS

11 Test Facilities

(4 eTest)

415 Testers (73 eTest)

69 distinct products

High Volume 2.1M runs

325M units

10 min avg time to Portal

~100 users avg 1k queries/ day


OUTCOMES GOOD UNITS PER HOUR

Baseline Gain

SORT FinalTest

95% 46%

Almost doubled SORT capacity in three months!


OUTCOMES G2ND (GROSS TO NET DELTA)

Baseline Gain

SORT FinalTest

57% 25%

Gap between gross and net only identified through O+


OUTCOMES TEST TIME

G2ND measures gap between GROSS and NET test time


OUTCOMES RETEST PERCENT

Baseline Gain

SORT FinalTest

65% 56%

O+ provided easy quantification of retest impact


OUTCOMES YIELD GAP (LAST – FIRST YIELD)

Baseline Gain

SORT FinalTest

65% 46%

Isolate hardware contribution


OUTCOMES

Business and process reset

Reduce cost

Reduce customization

Reduce supply disruptions

Culture “we know best” “what can we learn”

“we aren’t well” “we are sick”

“we can’t afford to do this” “we can’t afford to NOT do this”

Decision Making

Gap analysis ROI

Manage the gap (JMP)

TRANSFORMATION THEMES


WHAT’S NEXT?

Most units can be tested less

Leverage Sequoia and PAT algorithms

Separate populations into “Test Less”, “Test More”

“TEST AS NEEDED”

GDBN

Cluster

NNR

Clean “Test Less”

Dirty “Test More”

Health Score


CONCLUSION

Biggest hurdle was getting started

Both AMD and Optimal+ underestimated effort AND impact

Near immediate results drove cultural change

Emphasis shifts to drive alternate test methodology

Questions?

Backup


LANDSCAPE INFLECTION POINTS

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Acquired ATI

Global Foundries

Converged Yield

Management

APU SOC’s

Semi-Custom Product Ramp

Optimal + Decision


Deep Dive Analysis

Internal Tools

Optimal + Product and Process Monitoring

Silicon Focus

Operation Focus

Characterization

• Process Margining

• IP Validation

• Product Margining

Product Monitor

• WAT Trend

• Statistical Bin Limit

• Parametric Trend

• Test Time

Operation Monitor

• Continuity

• Statistical Bin Limit

• Parametric Trend

• Utilization

DATA SYSTEMS INTERPLAY

Internal and external systems are complimentary – supporting different purposes

• Internal tools focus on the silicon to drive the ceiling

•Optimal+ focus on near time monitoring of test data for anomaly detection to manage the floor


DISCLAIMER & ATTRIBUTION

The information presented in this document is for informational purposes only and may contain technical inaccuracies, omissions and typographical errors.

The information contained herein is subject to change and may be rendered inaccurate for many reasons, including but not limited to product and roadmap changes, component and motherboard version changes, new model and/or product releases, product differences between differing manufacturers, software changes, BIOS flashes, firmware upgrades, or the like. AMD assumes no obligation to update or otherwise correct or revise this information. However, AMD reserves the right to revise this information and to make changes from time to time to the content hereof without obligation of AMD to notify any person of such revisions or changes.

AMD MAKES NO REPRESENTATIONS OR WARRANTIES WITH RESPECT TO THE CONTENTS HEREOF AND ASSUMES NO RESPONSIBILITY FOR ANY INACCURACIES, ERRORS OR OMISSIONS THAT MAY APPEAR IN THIS INFORMATION.

AMD SPECIFICALLY DISCLAIMS ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE. IN NO EVENT WILL AMD BE LIABLE TO ANY PERSON FOR ANY DIRECT, INDIRECT, SPECIAL OR OTHER CONSEQUENTIAL DAMAGES ARISING FROM THE USE OF ANY INFORMATION CONTAINED HEREIN, EVEN IF AMD IS EXPRESSLY ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.

ATTRIBUTION

© 2013 Advanced Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow logo and combinations thereof are trademarks of Advanced Micro Devices, Inc. in the United States and/or other jurisdictions. Other names are for informational purposes only and may be trademarks of their respective owners.

amd at itc 2014

Software