展开查看详情
1 .Analyzing 2TB of Raw Trace Data
from a Manufacturing Process:
A First Use Case of Apache Spark for
Semiconductor Wafers from Real Industry
Seungchul Lee, sclee@bistel.com
BISTel Inc.
Daeyoung Kim, dykim3@bistel.com
BISTel Inc.
#UnifiedAnalytics #SparkAISummit
2 .Contents
• Introduction to BISTel
– BISTel’s business and solutions
– Big Data for BISTel’s smart manufacturing
• Use cases of Apache Spark in manufacturing industry
– Trace Analyzer (TA)
– Map Analyzer (MA)
#UnifiedAnalytics #SparkAISummit 2
3 .Introduction to BISTel
#UnifiedAnalytics #SparkAISummit 3
4 .BISTel’s business areas
• Providing analytic solutions based on Artificial Intelligence (AI)
and Big Data to the customers for Smart Factory
#UnifiedAnalytics #SparkAISummit 4
5 .BISTel’s solution areas
• World-Class Manufacturing Intelligence through innovation
#UnifiedAnalytics #SparkAISummit 5
6 .BISTel’s analytic solution: eDataLyzer
#UnifiedAnalytics #SparkAISummit 6
7 .BISTel’s analytic solutions (MA)
• Map Pattern Clustering
– Automatically detect and classify map patterns with/without libraries
– Process thousands of wafers and give results in few minutes
Clustered
Defective
wafers
#UnifiedAnalytics #SparkAISummit 7
8 .BISTel’s analytic solutions (TA)
• Specialized Application for Trace Raw Data
– Extracts the vital signs out of equipment trace data
– Provide in-depth analysis which traditional methods cannot reach
Normal
Abnormal
#UnifiedAnalytics #SparkAISummit 8
9 .BISTel’s big data experiences
#UnifiedAnalytics #SparkAISummit 9
10 .BISTel’s big data experiences
- YMA Test using Spark - - Big data platforms comparison-
#UnifiedAnalytics #SparkAISummit 10
11 .Trace Analyzer (TA)
#UnifiedAnalytics #SparkAISummit 11
12 .Trace Data
• Trace Data is sensor data collected from processing equipment
within a semiconductor fab during a process run.
- Wafer -
- Semiconductor industry -
#UnifiedAnalytics #SparkAISummit 12
13 . Logical Hierarchy of the trace data
wafer
Wafer
Lot Visualization
Whole
process
Recipe Step
Recipe
Process
Recipe 1
Process
#UnifiedAnalytics #SparkAISummit 13
14 . An example of the trace data
Process Recipe Recipe step Lot Wafer Param1 Param2 Time
2015-01-20
021_LIT RecipeA 1 1501001 1 32.5 45.4
09:00:00
#UnifiedAnalytics #SparkAISummit 14
15 .Data attributes
• Base unit : one process and one parameters
• 1000 wafers
• Each wafer has 1000~2000 data points in a recipe step
• Some factors that make trace data huge volume
• # of parameters
• # of processes
• # of wafers
• # of recipe steps
• duration of the recipe step
#UnifiedAnalytics #SparkAISummit 15
16 .An example of the trace data – (2)
Parameter
# of # of Avg. Recipe Data # of
No. Fab per unit
processes recipe steps Process Time Frequency units
(max)
1 Array 109 10 16 mins 1Hz 288 185
2 CF 25 5 1min 1Hz 154 340
3 CELL 12 7 1min 1Hz 213 326
4 MDL 5 12 2mins 1Hz 32 154
• Some calculations
• For one process, one parameter and one wafer
• 16 * 10 * 60 sec * 1Hz = 9600 points
• Multi parameters, multi processes and multi wafers
• 9600 * 288 *185 * 109 * (# of wafers)
#UnifiedAnalytics #SparkAISummit 16
17 .Spark : Smart manufacturing
• Spark is a best way to process big data in batch analytics
• Distributing data based on parameter is suitable for using
Apache Spark.
• Easy deployment and scalability when it comes to providing the
solutions to our customers
#UnifiedAnalytics #SparkAISummit 17
18 .Naïve way: applying spark to TA
#UnifiedAnalytics #SparkAISummit 18
19 .How to apply Spark to TA?
traceDataSet = config.getTraceRDDs().mapToPair(t->{
String recipeStepKey = TAUtil.getRecipeStepKey(t); #use recipe step as key
return new Tuple2<String,String>(recipeStepKey,t);
}).groupByKey();
traceDataSet.flatMap(t->{
Map<String,TraceDataSet> alltraceData = TAUtil.getTraceDataSet(t);
...
TAUtil.seperateFocusNonFocus(alltraceData,focus,nonFocus); #separate data
ta.runTraceAnalytic(focus,nonFocus,config); # calling the TA core
...
});
20 .Most cases in manufacturing industry
• In real industry, most parameters have small number of data points.
(Most case : 1Hz)
• In addition, the number of wafers to be analyzed is not massive.
(up to 1,000 wafers)
• Therefore the total number of data points in a process can be easily
processed in a core
21 .Issues in manufacturing industry
• Last year, I have got an email indicating that..
#UnifiedAnalytics #SparkAISummit 21
22 .Big parameter
• Tools with high frequency or high recipe time can produce huge
volume for single parameter
• Requirements in industry
• For one parameter
• 400,000 wafers
• 20,000 data points.
#UnifiedAnalytics #SparkAISummit 22
23 .Limitations of the Naïve TA
traceDataSet = config.getTraceRDDs().mapToPair(t->{
String recipeStepKey = TAUtil.getRecipeStepKey(t); #use recipe step as key
All the data points based
return new Tuple2<String,String>(recipeStepKey,t);
}).groupByKey();
on the key are pushed
into one core by shuffling
For(Tuple<String,Iterable<String> recipeTrace : allTraceData){
TraceDataSet ftds = new TraceDataSet();
Iterable<String> oneRecipe = recipeTrace._2();
for(String tr : oneRecipe){
TraceData td = TAUtil.convertToTraceData(tr);
ftds.add(td);
} Java object holds too
} many data points
#UnifiedAnalytics #SparkAISummit 23
24 .Needs for new TA spark
• Naïve TA Spark version cannot process massive data points.
• Nowadays, new technology enhancements enable data capture at
much higher frequencies.
• TA for “big parameter” version is necessary.
#UnifiedAnalytics #SparkAISummit 24
25 .Our idea is that..
• Extracting the TA core logic
– Batch mode
– Key-based processing
– Using .collect() to broadcast variables
– Caching the object
#UnifiedAnalytics #SparkAISummit 25
26 .Batch
First element : process, recipe
step, parameter and batch ID
• Preprocessing trace data Second element : lot, wafer and
trace values
JavaPairRDD<String, List<String>> traceDataRDD
= TAImpl.generateBatch(traceData)
• Key-based processing Summary
• Base unit : process key or recipe step key statistics
.
.
.
•Param A
#UnifiedAnalytics #SparkAISummit 26
27 .Collect() : TA Cleaner
• Filtering out traces that have unusual duration of process time.
• Use the three main Spark APIs
– mapToPair : extract relevant information
– reduceByKey : aggregating values based on the key
– collect : send the data to the driver
#UnifiedAnalytics #SparkAISummit 27
28 .Collect() : TA Cleaner – (2)
• traceData.mapToPair()
• Return
• key : process
• value : wafer and its length
Worker Worker Worker Worker
wafer value wafer value wafer value wafer value
1 65 1 83 1 34 1 71
2 54 2 54 2 77 2 80
… … … … … … … …
#UnifiedAnalytics #SparkAISummit 28
29 .Collect() : TA Cleaner – (3)
• reduceByKey()
• Aggregating contexts into one based on the process key
wafer value wafer value
1 65 1 88
2 54 2 92
… … … …
wafer value
Shuffling
1 153
2 146
… …
#UnifiedAnalytics #SparkAISummit 29
热门城市
