Download - Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Transcript
Page 1: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Dataflow Process NetworksLee & Parks

Synchronous DataflowLee & Messerschmitt

Abhijit Davare

Nathan Kitchen

Page 2: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Dataflow

Data oriented Host Language

Java, C/C++ Coordination language

SDF, Cyclo-static Dataflow

Page 3: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Kahn Process Networks

Dataflow process networks are a special case of Kahn process networks.

P2

a, b, c A, B, C

unbounded

channels

P1 P3

Page 4: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Properties of Kahn Processes

The processes that we deal with are usually:ContinuousMonotonic

Which means that we can realistically compute them.

Page 5: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Continuous Processes

Formally:

Practically: A continuous process

will not wait for the end of an infinite input stream before producing an output.

)( FF

Page 6: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Monotonic Processes Adding to the input does not change the output

already produced. All continuous processes are monotonic.

monotonic not monotonic

P1, 2, 3 2, 4, 6

P1, 2 2, 4

P1, 2, 3 4, 2, 6

P1, 2 2, 4

Page 7: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Computing Continuous Processes Continuity means:

We can start producing the output stream before we have received all the input.

We can process an infinite stream with finite resources.

Networks of processes are determinate.

Page 8: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Nondeterminism

Pros: Compact system definition Incomplete specification OK

Cons: Loss of continuity Analysis more difficult

Page 9: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Nondeterminism (2) Nondeterminism can be added to KPN by:

Processes can test for empty channels Internal Nondeterminism Channel has multiple sources Channel has multiple sinks Processes have shared variables

rand()

x+=2

x=0 empty?

Page 10: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Streams Most relevant in real-time signal

processing Recursively defined or Channel oriented Be good at losing data, recycling memory,

and storing data.

Recursive Camp Channel Camp

Page 11: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Dataflow Processes

When a dataflow actor fires, it consumes inputs and produces outputs.

We get a dataflow process by repeatedly firing an actor.

2

2+ 4

Page 12: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Firing Rules for Actors

Dataflow processes are continuous if: Each actor firing is functional

(i.e., the outputs depend only on the inputs).

The firing rules are sequential. A firing rule is a set of patterns

that the inputs have to match.

Page 13: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Examples of Firing Rules

At least one token on each input: R1 = { [*], [*] }

The select channel chooses an input channel to take a token from: R1 = { [*], , [T] }

R2 = { , [*], [F] }

+[*]

[*]

[*]

[*]

F

T

Page 14: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Firing Rules in SDF

A synchronous dataflow actor has a single firing rule containing only wildcards—it is enabled by a fixed number of tokens.

downsample

4 1

R = { [*, *, *, *] }

Page 15: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Firing Rules with a Problem

At least one token on either input:

R1 = { [*, ] }, R2 = { [, *] } Which rule should be checked first? If we check

with a blocking read, we will deadlock.

These rules are not sequential.

merge merge

Page 16: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Sequential Firing Rules

If we can avoid deadlock by checking rules in the right order, the rules are sequential.

Example: the select actorR1 = { [*],, [T] }

R2 = { , [*], [F] }

We read the third input first, then we know which other input to read.

Page 17: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Implications of Sequentiality

All sequential processes are continuous, therefore determinate.

Networks of functional actors with sequential firing rules can be scheduled; we do not have to synchronize processes using blocking reads.

Page 18: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Execution models

For Dataflow Process Networks, execution independent of scheduling

1. Concurrent Processes

2. Dynamic Scheduling

3. Static Scheduling

Page 19: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

B

Concurrent Processes

C

B

A

Page 20: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

B

Concurrent Processes

C

B

A

hungry

Page 21: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

B

Concurrent Processes

C

B

A

hungry

suspended

hungry

Page 22: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

B

Concurrent Processes

C

B

A

hungrysuspended

hungry

suspended

hungry

enabled

Page 23: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Concurrent Processes(2)

Large overhead of context switching In general, increasing process granularity

decreases relative overhead

Page 24: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Dynamic Scheduling

Data dependencies can make static scheduling impossible

In particular, dynamic scheduling is necessary when number of input and output tokens for each actor cannot be defined a priori

Can be done in hardware or software Actors activated when inputs available

Page 25: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Static Scheduling

Must have number of input & output tokens predetermined for all actors

Best schedule chosen among several based on objective (e.g. minimum buffer size, minimum code size)

Used for:Code generation by code stitchingHW synthesis

Page 26: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Formalism for Static Scheduling

Rank: Number of linearly independent vectors

1

2

4

3

4080

1200

0120

0063

T

1

2

3

43

6

2

8

4

1

2

1

Page 27: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Formalism for Static Scheduling

Rank: Number of linearly independent vectors

1

2

4

3

4080

1200

0120

0063

T

1

2

3

43

6

2

8

4

1

2

1

Full rankNo periodic solution

Page 28: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Formalism for Static Scheduling

Rank: Number of linearly independent vectors

4080

2200

0120

0063

T

1

2

4

3

1

2

3

43

6

2

8

4

1

2

2

2

2

1

2

Jq

Page 29: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Correctness of SDF graphs

1 2

3

A

BC

1 1

1

11

2

Page 30: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Correctness of SDF graphs

1 2

3

A

BC

1 1

2

11

2

Page 31: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Correctness of SDF graphs

1 2

3

A

BC

1 1

1

11

1

Page 32: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Delays

On startup, need to make sure that arcs have enough tokens to activate nodes

b(i) represents number of tokens in buffer i

b(n+1) = b(n) + T v(n) b(0) affects which startup options are legal

Page 33: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Single Processor Static Scheduling

Class S algorithms:Given a firing vector (q) and the initial number

of tokens in each bufferSelect each runnable node and update b(n)

iterativelyList of these selected nodes forms schedule If schedule cannot be met, deadlock has

occurred

Page 34: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Parallel Processor Static Scheduling

Exploit concurrency to increase throughput

In this example, nodes 1 and 3 can be scheduled at the same time on different processors

Periodic Admissible Schedules:

{1,3,1,2}

{3,1,1,2}

{1,1,3,2}

Page 35: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Static Buffering

Minimize execution time by having memory locations embedded as constants in the code, not variables

iq = KN Where

i = Number of tokens emitted per firing q = Number of firings in one period N = Size of buffer K = an integer

Page 36: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Functional Behavior & Hierarchy

Dataflow Process Networks allow hierarchy

Functionality may be present at the higher level

Delay, internal state, and unbalanced subsystems can cause nodes to be non-functional

Page 37: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Dataflow and Functional Languages Actors correspond to first-order functions:

Processes are higher-order functions (they take functions as arguments):

Process F results from applying first-order function f to a stream that starts with R and continues with X.

yxyxf ),(

)(:)():( XRXR FfF

Page 38: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Language Features

RecursionIterationCarrying state

PolymorphismParallelism

Page 39: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Recursion Not needed in dataflow to carry state

(we have feedback loops) Can be used with high-order functions

for compact models (e.g., FFT)Expensive unless unrolled at setup time

Page 40: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Polymorphism

Tokens can have arbitrary typeArrays as sequence of tokensArrays as single tokens

In PtolemyOne actor can operate on several types

add doubles add ints (without converting to doubles)

Page 41: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Parallelism

Functional languages Explicit Thwarted by recursion Use higher-order functions for parallelism instead

Dataflow Implicit Use higher-order functions as syntactic sugar

Unrolled at setup time Still parallel!

Page 42: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Credits

Sean Connery as James Bond: http://www.speakeasy.org/~wvt3rd/ BONDLIT1.HTM

Screen Beans © 1995, 1996, 1997, 1998, 1999, 2000 A Bit Better Corporation

Screen Beans is a registered trademark of A Bit Better Corporation. Figures from “Dataflow Process Networks,” Lee & Parks, Proc.

IEEE, May 1995, and “Synchronous Data Flow,” Lee & Messerschmitt, Proc. IEEE, Sep 1987.

All episodes are filmed before a live studio audience. No animals were harmed during the production of this presentation.

Page 43: Dataflow Process Networks Lee & Parks Synchronous Dataflow Lee & Messerschmitt Abhijit Davare Nathan Kitchen.

Tagged Token Model

Tokens have tags + values

Out of order execution Channels are not FIFO Actors fire when input tags

match Graph would deadlock as

dataflow More expressive Limited value