MENARID Knowledge Exchange workshop 24th to 28th of March, Hammamet, Tunisia Par Nabil HAMADA
[IEEE 2013 10th International Multi-Conference on Systems, Signals & Devices (SSD) - Hammamet,...
Transcript of [IEEE 2013 10th International Multi-Conference on Systems, Signals & Devices (SSD) - Hammamet,...
1 2 3 4 5 6 7 8 91011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556576061
Abstract—Networks-on-chip represent promising communication
infrastructures for highly integrated circuits to deal with future
systems design challenges, such as interconnection difficulties,
design productivity, and power consumption. Given the
difficulties associated with the design of network on chip and the
necessary time to validate the simulation network topology, or
the choice of network topology before the implementation at RTL
level, the Transaction-level modeling has been proposed as a
framework which allows rapid exploration of several networks
topology containing different descriptions of system components.
The Transaction-level represents a modeling technique that is
intended to separate the specification of computation and
communication while providing efficient methods to implement
the various elements at different levels of abstraction. This paper
introduces a library of generic TLM models of router
components that allow constructing several types of router which
can be used later to construct several network on chip topologies
for performance evaluation purpose. It allows taking decision of
the best topology for a given traffic or application target. To
prove our approach, several router types are proposed based on
these models of components and are used to build a mesh and
STAR-RING network on chip at transaction level.
Keywords-Network on Chip; Transaction Level Modeling; TLM
router; Traffic pattern
I. INTRODUCTION
The Intercommunication platform of SoCs composed
by hundreds of cores will not be feasible using a single
shared bus or a hierarchy of buses since they both have proved
their poor scalability with system size. Moreover these two
have a limited shared bandwidth between all the cores attached
to them. To overcome these problems of complexity and
scalability, Network on Chip (NoC) appears to be an
appropriate possible solution proposed as a promising
replacement for buses and dedicated interconnects [1, 2, and 3].
This is essentially based on on-chip packet switched
communication thanks to on-chip integrated routers [4]. The
use of NoC offers several advantages such as communication
scalability, better performance, modularity and structured
communication schemes [5]. This communication platform
allows perfect flexibility and reusability in the design of
complex SoC applications and fits the natural communication
scheme of any given application.
These properties of Network-on-chip have been verified in
several research works and have been implemented in both
VHDL [6, 4] and systemC [7] description.
Given the difficulties that are associated with the
integration of complex systems, the designer is faced with new
challenges. Therefore it is necessary to validate the simulation
system before the implementation. The Transaction-level
modeling framework allows rapid exploration of several
architecture containing different descriptions of system
components. The Transaction-level represents a modeling
technique intended to separate the specification of computation
and communication while providing efficient methods for
implementing the various elements at different levels of
abstraction [8], [9].
The Transaction Level Modeling (TLM) approach has been
proposed ([10], [11], [12]) as a way to tackle firstly the
HW/SW co-design problems and secondly the early
architecture analysis. The two standards that are closely related
to TLM are the open SystemC initiative (OSCI) [13] and the
Open Core Protocol (OCP) [14], that standardizes a modeling
language and an intellectual property (IP) interface
respectively. TLM model therefore works only on function of
calling and transfer data packets. The idea is to represent as
closely as possible for the designer the overall behavior of the
circuit without going into details of the description of the
signals which are carried at Register Transfer Level (RTL).
There are many variations in the level of the chosen
abstraction. It is the same case with TLM. This latter has two
models which are PV and PVT.
. At PV level, the model contains all the necessary
information for the software development teams to work. The
PVT model includes time information that can include work on
performance analysis of the circuit, without excessively
penalizing the simulation time. In the next sction we will
present related work for communication on chip.
A MODULAR AND GENERIC ROUTER TLM
MODEL FOR SPEEDUP NETWORK-ON-CHIP
TOPOLOGY GENERATION Nourddine Abid,
wissem Chouchene
Electronics and Micro-
Electronics Laboratory
Faculty of Sciences of Monastir
Monastir University, Tunisia
Email: [email protected]
Brahim Attia,
Electronics and Micro-Electronics
Laboratory
Faculty of Sciences of Monastir
Monastir University, Tunisia
Email: [email protected]
Abdelrim Zitouni and Rached
Tourki Electronics and Micro-
Electronics Laboratory
Faculty of Sciences of Monastir
Monastir University,
SSD'13 1569695475
1
2013 10th International Multi-Conference on Systems, Signals & Devices (SSD) Hammamet, Tunisia, March 18-21, 2013
978-1-4673-6457-7/13/$31.00 ©2013 IEEE
1 2 3 4 5 6 7 8 91011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556576061
II. RELATED WORKS
The idea of using TLM to describe communication in
network structures has been investigated in [15], [16] and [17],
where TLM is employed to model of the communication
between the adjacent switches on a packet’s path. While these
approaches can already improve simulation performance
significantly in comparison to RTL. A detailed description of
Methodologies of transaction level modeling with SystemC is
introduced in [18] and [19]. Communication mechanisms are
modeled as channels and presented to modules using SystemC
interface classes. Transaction requests take place by calling
interface functions of these channel models, which encapsulate
low-level details of the information exchange [18]. The authors
of [20] introduced and applied Parallel Discrete Event
Simulation techniques to build SoC simulation environment at
Transaction Level Model with Time (TLM/T) using NoC as its
interconnect architecture. In [21] a novel modeling technique,
called Result Oriented Modeling, which delivers the same
speed as TLM with fully accurate timing, is proposed. They
described the AMBA AHB bus architecture in this technique
and show its efficient simulation. Methodologies of transaction
level modeling have been already widely used in describing
electronic systems that are based on NoC. In [22], the writers
enumerated NoC characteristics to be modeled and expose
SystemC models' properties of a NoC system at several
abstraction levels. In [23], a scalable and parametrical NoC is
implemented in SystemC. The NoC is composed of resources
and routers connected by channels, with a mesh topology and
wormhole routing. In [24] and [25], the authors modeled a
MIMO-OFDM receiver at transaction level abstraction based
on NoC platform in SystemC, and then analyze and compare
the performance and complexity of the model for various
configurations in terms of throughput. Since the on-chip
interconnects topology is the key to determine the main
performance metrics of the NoC , the objective of this paper is
to build libraries of models of components that allow
constructing several type of router that can be used later to
construct several network on chip topologies for performance
evaluation propose. It allows taking decision of the best
topology for given traffic or application target. To prove our
approach, several routers type is proposed based in these
models of components and are used to build a mesh and
STAR-RING network on chip at transaction level. In the next
section will give a short introduction to TLM modeling, Mesh
topology, and detailed description of the STAR-RING
topology and routing algorithm used by it. After we will
present the TLM model proposed of router components.
Finally, the traffic generator and traffic analyzer are presented.
Section 4 presents a comparative study of the two network
topology in terms of performance evaluations using two traffic
models with the same network sizes. Finally, we conclude in
section 5 with remarks.
III. THE CONSIDERED PROPOSED TLM NETWORK ON CHIP
The connection is made through a bidirectional macro-
socket as shown in figure 1. A socket is a high level
mechanism that is introduced in the TLM 2.0 standard which
combines a port with an export to provide designers with the
facility of sending packet in two ways. Sockets belong to
protocol layer of interoperability TLM layers. The initiator and
target sockets group the TLM 2.0 interfaces including blocking
and non-blocking transport interfaces (transport layer) for both
forward and backward paths together into a single object.
Figure 1. TLM interconnection modules
The generic payload belonging to the application layer is
introduced to improve the interoperability of memory-mapped
bus models and facilitate the IP reuse of TLM2.0 IPs . It also
provides the extension mechanism which can be added to
generic payload object when generic payload attributes are not
adequate to model the full functionality of the architecture
(NoC). Moreover, it is capable of creating detailed models of
specific bus protocols, while reducing the implementation cost
and increasing the simulation speed of the whole design. The
main features of the generic payload are illustrated in figure 2.
We use in this work the mesh 2D topology with
deterministic routing function called XY routing. In this kind
of the routing, the transaction is routed in the X direction until
the destination address and the current address are equal. After,
the transaction is routed in the Y direction until it reaches the
destination router. In the X direction, the transaction is routed
to the east port or to the west port. In the Y direction, the
transaction can be routed to the north port or to south port.
Finally, the local port can route transaction to the north, south,
east, and west ports. For more information about the TLM
modeling of the mesh 2D can be found in [30].
Figure 2. TLM Generic_Payload
In this section we propose a new determinist routing
algorithm that allows obtaining a fixed diameter equal to two
hops regardless the network size and the STAR valence. The
STAR RING topology is characterized by a diameter of two
links independently of the total number of routers (N) as shown
in figure 3.
2
1 2 3 4 5 6 7 8 91011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556576061
Figure 3. TLM interconnection modules
The proposed STAR-RING topology was been build with two types of routers. The first one is used to build the ring topology and it is composed by four I/O ports numbered from 0 to 3 and positioned in a ring around the central node such presented in table I and named peripheral router.
TABLE I. I/O PORTS ATTRIBUTE
Reference Port Symbol
0 Local Port L
1 CLOCKWISE Port CW
2 COUNTER CLOCKWISE
Port
CCW
3 ACCROSS Port AC
The second router is the star router with high radix valence.
The number of its I/O ports is equal to the number of the
routers placed in the ring plus one. It has V+1 I/O ports from
numbered from 0 to V with their local port has the value V.
The figure 7 present a star router with 9 input ports. It uses a
local port and 8 ports to connect the star router with the 8
peripheral routers.
As shown in figure 4 each peripheral router uses 4 ports.
The first one is the local port where the IP core is connected.
The others ports are used to connect the peripheral router with
neighboring routers called respectively CW port and CCW
port, and AC port. The AC port establishes the connection
between peripheral router and the STAR router.
Figure 4. TLM interconnection modules
When a peripheral router initiate a transaction through its
local port to other router, the routing function of the initiator
router computes the address difference between initiator and
target address, if the difference are greater than 2 or smaller
than -2 the routing function send a request to the Across port
and the transaction are routed to the STAR router. It will be
routed later to the destination peripheral router. When the
difference is equal to 1 or 2 then the transaction will be routed
to the CW port. Finally if the difference is equal to -1 or -2 then
the transaction will be routed to the CCW port. The next figure
presents the pseudo code used to implement the routing
function for local port.
All the peripheral routers can send transaction to the STAR
router by using its unique address. The routing function used
by the STAR router is not the same. When the local port of the
STAR router wants to send transaction to a peripheral router, a
connection is made with the across port of the destination
router through the SATR router.
Figure 5. TLM interconnection modules
The routing function of the port K of STAR router cannot route packet only for routers K-2, K-1, K+1, and K+2. The central router is connected to all routers in the ring on its peripheral ports. Figure 6 presents the pseudo code used to implement the routing function sub routine for STAR router which is different to then used by the local port.
3
1 2 3 4 5 6 7 8 91011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556576061
Figure 6. TLM interconnection modules
In order to model TLM mesh 2D or STAR RING topologies, we present in figure 9 our extension including several attributes able to perform the used routing algorithm. Moreover this extension employs priority of the transaction that can be used to improve QoS. “Adress_Port_int” attribute corresponds to the source address of the initiator module providing the transaction (Master IP) while “Adress_Port_dest” refers to the destination address of the target module (Slave IP) accepting the transaction. In fact, those parameters indicate the essential factors of the communication network specified in the routing functions. To ensure the scalability of the 2D mesh or STAR RING, we specify a setting in the template entry that indicates the number of rows and columns or the valence needed for the automatic mapping instance of the network object.
Figure 7. TLM interconnection modules
The mapping of the network instance is characterized by the interconnection of different types of routers as presented in figure 4 and figure 7. There two type of router in the STAR RING topology and 4 types of routers distributed symmetrically on the network and have the same functionality. The local port of each router is connected with an IP. As shown in Figure 11, the IP is modeled in the form of a macro- block composed by a traffic generator and traffic analyzer.
Figure 8. TLM interconnection modules
The traffic generator use an initiator socket connected to a target socket to the input port of the router. The traffic analyzer has a target socket connected to an initiator socket to the output port of the router. We note that each router processing type depends on its position with regards to the limits of the network, as shown in Figure 10. For example, the central switch has all five ports. However, each corner switch has only three ports.
Figure 9. Our TLM extension
Transaction generator: (initiator module: TG) is used to inject transactions into the network to be emulated according to the configure information. The TG generates transactions under various traffic patterns and injunction rate. Each transaction consists of a transaction command, a source address, and destination address fields. The transaction length is parameterized. The traffic patterns include stochastic traffic in uniform and non-uniform distributions (complementary traffic, hot spot traffic).
Transaction analyzer: (target module: TA) performs the following functions: When the transaction reaches the destination node, it will be transmitted into the packet target module.
In order to analyze the transmission delay, each transaction carries an attribute “delay” time stamp. The transaction generator and analyzer modules are synchronized in the simulation system with transport function call. The TA takes the time stamp out and calculates the transmission delay when the transaction reaches the destination. The TA also calculates the total transaction number and the total simulation time. These data are saved in a trace file owner to each TA.
Router: The first router type used in the mesh simulation has 5 input-output ports. Each port has two types of sockets, an initiator socket and a target socket. Indeed, each initiator socket is directly linked to a routing function, responsible for the switching of transactions. At the same time, an arbiter is placed
4
1 2 3 4 5 6 7 8 91011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556576061
upstream of each target socket in order to manage competing requests. The routing and arbiters functions, comply with the principle of internal connectivity through the mapping between the initiator sockets of the routing side, and the target sockets for the arbitration side.
Arbitration function: it is able to manage the competing request from initiators, which seek to achieve the same target module. The transport method can indeed be called simultaneously by multiple processes. This implies that a hardware implementation of the routing node must have the computing resources and infinite memory in order to simultaneously process an arbitrary number of transactions. To remedy this problem, a mutual exclusion (mutex) can be exploited in an arbitration module to manage the flow of data. The Queues can be used for the purpose of synchronization. This allows having a model with contention, which describes a realistic operation ready for synthesis.
Figure 10. Type of Network routers
In our case, the arbitration module encapsulates the functionality of a mutual exclusion from the class library SystemC scx_mutex_w_policy and a queue. These arbitrations are modeled following the approach first-come first-served. In this arbiter architecture, any request relating to a transaction, activates the verification operation to access to the shared resource. The access is permitted if the resource is free. The arbiter selects the socket initiator for the process requesting access and lock by calling the lock () function. However, any requests from other, their transactions will be temporarily buffered in the queue. Once the task of the current process is completed, the arbiter releases the access by calling the unlocked () function. This allows the arbiter to get the first transaction from queue and then notify it in order to be eligible. We note that the used arbitration has a generic parameter in terms the target number of sockets. Indeed, each arbiter has a target socket array in input and a single output socket.
Routing function: The routing functions are inherited from sc_module, and each of them has a specific function. Each function performs a comparison between the source address of the router where it is located and the destination address encapsulated in the transactions from the input ports. The difference between the properties of all modules routing is the decode_address method implementation. The Decode_address function is taken as a parameter; the destination address procures the extension of transaction and returns the index of the socket initiator initiator_socket from the table.
Figure 11. Network components
However, the index is returned, acheminemnt of the transaction is carried out by blocking transport interfaces b_transport.
IV. PERFORMANCE EVALUATION
The evaluation of interconnection networks requires the
definition of representative workload models. Our evaluation metric for analyzing the proposed TLM NoC are latency and throughput. Transport latency is defined as the time (in clock cycles) that elapses between the occurrence of a message header injection into the network at the source node and the occurrence of a tail flit reception at the destination node [26]. In order to reach the destination node from some starting source node, flits must travel through a path consisting of a set of switches and interconnect which are called stages. Depending on the source/destination pair and the routing algorithm, each message may have a different latency. Therefore, for a given packet Pi, the latency Li is defined as:
Li = receiving time (tail flit of Pi) (1)
– Sending time (header flit of Pi).
The average packet latency is used as a performance metric in our evaluation. Let F be the total number of packets reaching their destination and let Li be the latency of packet Pi, where i ranges from 1 to F. The average packet latency, Lavg, is then calculated according to the following equation:
FLLF
i iavg /1∑ =
= (2)
The throughput packet tells the rate that traffic packet can be sent across the network. For packet passing system, the throughput packet, TP, is defined as follows [24]:
( )*( )
( )*( )
Total messages completed Message lengthTP
Number of IP blocks Total time=
(3)
where Total packets completed refers to the number of whole packets that successfully arrive at their destination PCs, length Packet is measured in flits, Number of IP blocks is the number of functional IP blocks involved in the communication, and the total time is the time (in clock cycles) that elapses between the occurrence of the first packet generation and the last packet reception. Thus, throughput is measured as the fraction of the maximum load that the network is capable of physically handling. An overall throughput of TP = 1 corresponds to all end nodes receiving one flit every cycle.
5
1 2 3 4 5 6 7 8 91011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556576061
Accordingly, throughput is measured in flits/cycle/PC. Throughput signifies the maximum value of the accepted traffic and it is related to the peak data rate sustainable by the system [27].
The workload or traffic model is basically defined by three parameters: distribution of destinations, injection rate, and message length. The traffic pattern indicates the destination for the next message at each node. We use the most frequently used traffic pattern; the uniform distribution [28]. In this distribution, the probability of node i sending a message to node j is the same for all i and j, i ≠ j [29]. The case of nodes sending packets to themselves is excluded because we are interested in the packet transfers that use the network [28].
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1125
130
135
140
145
150
uniform traffic
Latency (flit/cycle/IP)
Injection rate (flits/cycle)
STAR-RIN
Mesh2D
Figure 12. Average transaction latency under Uniform traffic
Figure 12 and 13 depict the average packet latency and the throughput packet versus normalized accepted traffic for the 3 x 3 mesh 2D network sizes using the XY dimension-order routing, and STAR RING topology with a valence equal to 8 using the proposed routing algorithm, with a uniform distribution of message destinations.
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10.01
0.02
0.03
0.04
0.05
0.06
0.07
Uniform traffic
Injection rate(flits/cycle)
throughput (flit/cycle/IP)
STAR-RING
mesh2D
Figure 13. Throughput under Uniform traffic
Figure 12 clearly indicates that 3 x 3 mesh network saturates at 90 %, the START RING network saturates at 80%. The STAR RING provides a better result in terms of latency than the mesh network. Figure 13 shows the variation of throughput under different injection rates. The mesh
throughput is better than the START RING network starting from 80% of injection rate because it has the lowest sensitivity to the packet drops. For injection rate greater than 80%, the latency and throughput of the mesh network is better than STAR RING network.
The second traffic pattern used is the hot spot traffic. In this type of traffic, each traffic generator sends a set of packets to other traffic analyzer with an equal probability except for a specific node (Called Hotspot) which receives packets with a greater probability. The percentage of additional packets that a hotspot node receives compared to other nodes is indicated after a hotspot name (Hotspot 100%).
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
200
400
600
800
1000
1200
Hot spot traffic
Latency (flit/cycle/IP)
Injection rate (flits/cycle)
STAR-RING
mesh2D
Figure 14. Average transaction latency under hotspot traffic
Figure 14 and 15 show the average packet latency and the throughput packet versus normalized accepted traffic for the 3 x 3 mesh 2D network sizes using the XY dimension-order routing, and STAR RING topology with a valence equal to 8 using proposed routing algorithm, with hot spot distribution of message destinations. The STAR router is the specific router for STAR RING network and the router (11) is the specific router for 3x3 mesh 2D network.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Hot spot traffic
Injection rate (flits/cycle)
throughput (flit/cycle/IP)
STAR-RING
mesh2D
Figure 15. Throughput under hotspot traffic
The mesh network outperforms STAR RING network in terms of latency and throughput the period of injection rate variation.
6
1 2 3 4 5 6 7 8 91011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556576061
V. CONCLUSION
A library of generic TLM models of router components
has been constructed. Two different topologies were
selected and analyzed using specific routing and switching
methods with systemC TLM2.0. A comparison study
aimed at exploring different topologies of networks-on-
chip has been conducted at transactional level modeling.
Simulations have been carried out to compare the metrics
of performance such as latency, throughput and network
load. The mesh 2D topology has a little increase of
throughput when compared to STAR RING network
topology in uniform traffic pattern but has a higher latency
in the same pattern of traffic. Mesh 2D network
outperforms STAR RING topology in the most of the
injection rate variation in hot spot traffic pattern because
of its regular and stable behavior. STAR RING topology
generally performs badly in most of the considered
parameters.
Other architectures will be studied and simulated for
different parameters using our library components.
REFERENCES
[1] L. Benini and G. De Micheli. Networks on chip: a new soc
paradigm. IEEE Computer, January 2002
[2] S. Kolson, A. Jantsch and al. A NoC architecture and design
methodology. Proc. Annual Symp. VLSI, 2002.
[3] W. Dally and B. Towles. Route packets not wires: On chip
inttercnnection networks. Design Automation Conference. pp.
684-689, 2001.
[4] F. Moraes, N. Calazans, A. Mello, M. Leandro, and L. Ost.
Hermes. An infrastructure for low area overhead
packetswitching networks on chip. Integr. VLSI J., Vol.38,
Issue1. pp. 69-93,2004.
[5] S. Kumar and al. A network on chip architecture and design
methodology. Proc. Symposium on VLSI, editor, Proc.
Symposium on VLSI. pp. 117-124, April 2002.
[6] J.-Y. Nollet, V. Marescaux, T. Verkest, D. Vernalde, S.
Lauwereins, R. Bartic and T.A. Mignolet. Highly scalable
network on chip for reconfigurable systems. IEEE International
Symposium on System-on-Chip. pp. 79-82, November 2003.
[7] D. Bertozzi, A. Jalabert, S. Murali, R. Tamhankar, S. Stergiou,
L. Benini, and G. De Micheli. Noc synthesisfilow for
customized domain specific multiprocessor systems-on-chip.
IEEE Transaction on Parallel and Distributed Systems. Vol. 16.
pp.113-129, February 2005.
[8] Cai, L. and D. Gajski. Transaction Level Modeling: An
Overview. International Conference on Hardware/Software
Codesign and System Synthesis (CODES+ISSS). pp. 19-24,
2003.
[9] A. Donlin. Transaction Level Modeling: Flows and Use Models.
International Conference on Hardware/Software Codesign and
System Synthesis (CODES+ISSS). pp. 75-80, 2004.
[10] L. Cai and D. Gajski: Transaction Level Modeling: An
Overview.International Conference on Hardware/Software
Codesign and System Synthesis. October 2003.
[11] L. Benini and.al.: SystemC Cosimulation and Emulation of
Multiprocessor SoC Design. IEEE Computer; April 2003.
[12] A. Donlin. Transaction Level Modeling: Flows and Use Models.
Int. Conf. Hardware/Software Codesign and System Synthesis;
September 2004.
[13] Open SystemC Initiative OSCI: SystemC documentation, OSCI,
www.systemc.org; 2004.
[14] OCP-IP: Open Core Protocol, OCP 2.1 Specification. OCP-IP.
www.ocpip.org; 2005.
[15] J. Xi and P. Zhong, “A Transaction-Level NoC Simulation
Platform with Architecture-Level Dynamic and Leakage Energy
Models. Proc. Of the 16th ACM Great Lakes Symposium on
VLSI (GLSVLSI), April 2006.
[16] S. Lee, S.-R. Yoon, J. Lee, M. L. Huang, and S.-C. Park.
Transaction level model simulator for NoC-based MPSoC
platform. Proc. of the 6th WSEAS International Conference on
Instrumentation, Measurement, Circuits and Systems (IMCAS),
January 2007.
[17] S. G. Pestana, E. Rijpkema, A. Radulescu, K. Goossens, and O.
P. Gangwal. Cost-performance trade-offs in networks on chip: A
simulationbased approach. Proc. of the Conference on Design,
Automation and Test in Europe (DATE), February 2004
[18] T. Grotker, S. Liao, G. Martin, and S. Swan, System Design
with SystemC. Kluwer Academic Publishers: Massachusetts,
2002, pp. 131151
[19] F. Ghenassia, Transaction Level Modeling with SystemC : TLM
Concepts and Applications for Embedded Systems, Springer:
Netherlands, 2005, pp.23-95.
[20] E. Viaud, F. P'echeux, and A. GreinerAn, "Efficient TLM/T
Modeling and Simulation Environment Based on Conservative
Parallel Discrete Event Principles," Proceedings of the
conference on Design, automation and test in Europe, Munich,
Germany, 2006, pp. 94-99
[21] G. Schimer and R. D'omer, "Fast and Accurate Transaction
Level Models using Result Oriented Modeling. Proceedings of
the 2006 IEEE/ACM international conference on Computer-
aided design, San Jose, Califomia, pp. 363-368, 2006.
[22] S. H. Sfar, I. E. Bennour, and R. Tourki. Transaction level
modeling of an OSI-like layered NoC. International Conference
on Design and Test of Integrated Systems in Nanoscale
Technology, Gammarth, Tunisia, pp.404-408, 2006.
[23] A. Portero, R. Pia and J. Carrabina. SystemC Implementation of
a NoC. IEEE International Conference on Industrial
Technology, Hong Kong, China, pp. 1132-1135, 2005.
[24] Sung-Rok Yoon and Sin-Chong Park. Case Study on
Transaction Level Modeling of NoC based IEEE 802.11n. Asia-
Pacific Conference on Communications, Busan, Korea, pp. 1-4,
2006.
[25] Sung-Rok Yoon, Jin Lee and Sin-Chong Park. Transaction level
analysis of NoC based coded MIMO-OFDM receiver. IEEE
Wireless Communications and Networking Conference, Las
Vegas, USA, pp.1794-1799, 2006,
[26] P. P. Pande, C. Grecu, A. Ivanov and R. Saleh, "Design of a
switch for network on chip applications. Proc. International
Symposium on Circuits and Systems, pp. 217–220, 2003.
[27] P. P. Pande, C. Grecu, M. Jones, A. Ivanov, and R. Saleh,
"Performance evaluation and design trade-offs for network-on-
chip interconnect architectures," IEEE Trans. Computer, Vol.
54-8, pp.1025–1040, August 2005.
[28] J. Duato, S. Yalamanchili, and L. Ni, “Interconnection networks:
an engineering approach. Morgan Kaufmann Publishers, 2003.
[29] D. A. Reed and D. C. Grunwald, "The performance of
multicomputer interconnection networks," IEEE Trans.
Computer, Vol. 20, Issue 6, pp. 63–73, June 1987.
[30] A. Noureddine, C. wissem, and A. Brahim, “Design and
Performance Evaluation of On Chip Network with Transaction
Level Modeling. Proceedings of the conference International
Conference Microelectronics (ICM), DEC 2011
7