Design and Implementation of SOC Bus Based on AMBA 4.0
2 likes751 views
This document details the design and implementation of an advanced peripheral bus (APB) bridge for the ARM Advanced Microcontroller Bus Architecture (AMBA) 4.0, facilitating communication between high-performance AXI and low-power APB domains. It describes key features such as clock domain crossing, a finite state machine, and the error response mechanisms to handle transfer interruptions. The research demonstrates the operation of this bridge, highlighting its ability to manage multiple peripherals and prioritize read requests during data transfers.
![IJSRD - International Journal for Scientific Research & Development| Vol. 1, Issue 7, 2013 | ISSN (online): 2321-0613
All rights reserved by www.ijsrd.com 1514
Abstract—ARM introduced the Advanced
Microcontroller Bus Architecture (AMBA) 4.0
specifications in March 2010, which includes Advanced
extensible Interface (AXI) 4.0. AMBA bus protocol has
become the de facto standard SoC bus. That means more
and more existing IPs must be able to communicate with
AMBA 4.0 bus. Based on AMBA 4.0 bus, we designed
an Intellectual Property (IP) core of Advanced Peripheral
Bus (APB) Bridge, which translates the AXI4.0-lite
transactions into APB 4.0 transactions. The bridge
provides an interface between the high-performance AXI
bus and low-power APB domain
Keywords-SoC; AMBA; AXI; APB
I. INTRODUCTION
Integrated circuits have entered the era of System-on-a-Chip
(SoC), which refers to integrating all components of a
computer or other electronic system into a single chip. It
may contain digital, analog, mixed-signal, and often radio-
frequency functions – all on a single chip substrate. With the
increasing design size, IP is an inevitable choice for SoC
design. And the widespread use of all kinds of IPs has
changed the nature of the design flow, making On-Chip
Buses (OCB) essential to the design. Of all OCBs existing in
the market, the AMBA bus system is widely used as the de
facto standard SoC bus.
On March 8, 2010, ARM announced availability of the
AMBA 4.0 specifications. As the de facto standard SoC bus,
AMBA bus is widely used in the high-performance SoC
designs. The AMBA specification defines an on-chip
communication standard for designing high-performance
embedded microcontrollers. The AMBA 4.0 specification
defines five buses/interfaces [1]:
1 Advanced extensible Interface (AXI)
2 Advanced High-performance Bus (AHB)
3 Advanced System Bus (ASB)
4 Advanced Peripheral Bus (APB)
5 Advanced Trace Bus (ATB)
AXI, the next generation of AMBA interfaces
Defined in the AMBA 4.0 specification, is targeted at high
performance, high clock frequency system designs and
includes features which make it very suitable for high
speed sub-micrometer interconnect:
1 separate address/control and data phases
2 support for unaligned data transfers using byte
strobes
3 burst based transactions with only start address
issued
4 issuing of multiple outstanding addresses
5 easy addition of register stages to provide
timing closure
II. TOP BLOCK
A. Block Diagram
In this study, we focused mainly on the implementation
aspect of an AXI4-Lite to APB Bridge. The APB Bridge
provides an interface between the high-performance AXI
domain and the low-power APB domain. It appears as a
slave on AXI bus but as a master on APB that can access up
to sixteen slave peripherals. Read and write transfers on the
AXI bus are converted into corresponding transfers on the
APB. The AXI4- Lite to APB bridge clock diagram is
shown in Figure. 1.
Fig. 1: Block Diagram
B. Signal Connections
Figure.2 shows the component signal connections. The
bridge uses:
Fig. 2: Signal Connections
Design and Implementation of SOC Bus Based on AMBA 4.0
R. Gopi Reddy1
M. Sanjeeva Reddy2
1
M. Tech. 2
Professor
1,2
Kakatiya Institute of Technology & science, Warangal](https://image.slidesharecdn.com/ijsrdv1i7031-140806064256-phpapp01/85/Design-and-Implementation-of-SOC-Bus-Based-on-AMBA-4-0-1-320.jpg)
![Design and Implementation of SOC Bus Based on AMBA 4.0
(IJSRD/Vol. 1/Issue 7/2013/0031)
All rights reserved by www.ijsrd.com 1515
1 AMBA AXI-Lite signals as described in the
AMBA AXI-Lite 4.0 protocol specification.
2 AMBA APB signals as described in the AMBA
APB 4.0 protocol specification.
C. AXI Handshake Mechanism
In AXI 4.0 specification, each channel has VALID and
READY signals for handshaking [2]
. The source asserts
VALID when the control information or data is available.
The destination asserts READY when it can accept the
control information or data. Transfer occurs only when both
the VALID and READY are asserted. Figure 3 show all
possible cases of VALID/READ handshaking. Note that
when source asserts VALID, the corresponding control
information or data must also be available at the same time.
The arrows in Figure 3 indicate when the transfer occurs. A
transfer takes place at the positive edge of clock. Therefore,
the source needs a register input to sample the READY
signal. In the same way, the destination needs a register
input to sample the VALID signal. Considering the situation
of Figure.3(c), we assume the source and destination use
output registers instead of combination circuit, they need
one cycle to pull low VALID/READY and sample the
VALID/RE ADY again at T4 cycle. When they sample the
VALID/REA DY again at T4, there should be another
transfer which is an error. Therefore source and destination
should use combinational circuit as output. In short, AXI
protocol is suitable register input and combinational output
circuit.
The APB bridge buffers address, control and data from
AXI4-Lite, drives the APB peripherals and returns data and
response signal to the AXI4-Lite. It decodes the address
using an internal address map to select the peripheral. The
bridge is designed to operate when the APB and AXI4-Lite
have independent clock frequency and phase. For every AXI
channel, invalid commands are not forwarded and an error
response generated. That is once an peripheral accessed does
not exist, the APB bridge will generate DE CERR as
response through the response channel (read or write). And
if the target peripheral exists, but asserts PSLVERR, it will
give a SLVERR response.
Fig. 3: Handshake Mechanism
III. CLOCK DOM AIN CROSSING (CDC)
A clock domain crossing (CDC), or simply clock crossing,
is when a signal crosses from o ne clock domain into
another [3]. If a signal does not assert long enough and is
not registered, it may appear asynchronous on the incoming
clock boundary. Meta stability happens when signal changes
within the setup/hold time window. Synchronizing a signal
that crosses into a higher clocked domain can be
accomplished by registering the signal through a flip-flop
that is clocked by the source domain, thus holding the signal
long enough to be detected by the higher clocked destination
domain. Synchronizing a signal traversing into a slower
clock domain is more cumbersome [4]. This typically
requires a register in each clock domain with a form of
feedback from the destination domain to the source domain,
indicating that the signal was detected.
A. Metastability
Metastability cannot be avoided, but a solution for handling the
metastable signal enables proper functioning of the design
[5].The metastability occurrences can be predicted by using the
mean time between failures (MTBF) formula (1)[6] : Where C1
and C2 are constants that depend on the technology used to
build the flip-flop; tMET is the duration of the metastable
output; and fclk and fdata are the frequencies of the
synchronous clock and the asynchronous input, respectively.
MTBF =
B. Synchronizer
Designers can use special metastable hardened flops for
increasing the MTBF [7]. For example, in Figure.4, a
synchronizer flop is used following the signal DB. So
instead of the metastable signal DB being used in the
function downstream as in Figure 4, the stable signal DB2 is
used in the logic downstream [8]. In the AXI4-Lite to APB
bridge, we use synchronizer block designs for communicate
between the AXI and APB clock domain.
Fig. 4: Two flip-flop synchronizer](https://image.slidesharecdn.com/ijsrdv1i7031-140806064256-phpapp01/85/Design-and-Implementation-of-SOC-Bus-Based-on-AMBA-4-0-2-320.jpg)
![Design and Implementation of SOC Bus Based on AMBA 4.0
(IJSRD/Vol. 1/Issue 7/2013/0031)
All rights reserved by www.ijsrd.com 1516
IV. FINITE STATE MACHINE FSM)
A finite state machine is a mathematical abstraction
sometimes used to design digital logic or computer
programs [9]. It is a behavior model composed of a finite
number of states, transitions between those states, and
actions, similar to a flow graph in which one can inspect the
way logic runs when certain conditions are met
[10]
. The
state transition diagram is a picture of our state machine
model. Figure 5 is the state transition diagram of our FSM.
The state machine operates through the following states:
IDLE. This is the default state of the FSM.
WRITE_SETUP. When a write transfer request is
asserted, the FSM moves into the WRITE_SETUP
state.
READ_SETUP. When a read transfer request is
asserted, the FSM moves into the READ_SETUP
state.
WRIET_ACCESS. The enable signal, PENABLE,
is asserted in the WRITE_ACCESS state.
READ_ACCESS. The enable signal, PENABLE, is
asserted in the READ_ACCESS state.
WRITE_WAIT. When the AXI write response
channel is not ready for receiving signal BRESP,
then stay in WRITE_WAIT state.
READ_WAIT. When the AXI read data channel is
not ready for receiving signal RRESP, then stay in
READ_WAIT state.
States READ_WAIT and WRITE_WAIT are added,
because the APB is not pipelined, wait states are added
during transfers between the APB and AXI interface.
System Verilog source code sample of the enumerated type
encoded FSM is given below
enum logic [2:0] {IDLE = 3'b000, READ_SETUP =
3'b001, WRITE_SETUP = 3'b010, READ_ACCESS
= 3'b011, WRITE_ACCESS = 3'b100, READ _WAIT
= 3'b101, WRITE_WAIT = 3'b110} Current State,
Next State
Fig. 5: State Transition Diagram
According AXI specification, the read address channel,
write address channel and write data channel are completely
independent. Each channel has a set of forward signals and a
feedback signal for handshaking. A read and a write
requests may be issued simultaneously
(AWVALID/WVALID and ARVALID are asserted high
simultaneously) from AXI4-Lite, the AXI4-Lite to APB
bridge will give more priority to the read request than to the
write request. That is, when both write and read requests are
valid, the write request is initiated on APB after the read is
requested on APB.
V. SIMULATION & IMPLEMENTATION
The timing diagrams shown above illustrate the AXI4-Lite
to APB bridge operation for various read and write transfers.
It shows that when both read and write requests are active,
read is given more priority.
VI. CONCLUSION
In this study, we provide an implementation of AXI4-Lite
to APB Bridge which has the following features:
32-bit AXI slave and APB master interfaces.
PCLK clock domain completely independent of
ACLK clock domain.
Support up to 16 APB peripherals.
Support the PREADY signal which translates to
wait states on AXI.
An error on any transfer results in SLVERR as the
AXI read/write response.
REFERENCES
[1] ARM, "AMBA Protocol Specification 4.0",
www.arm.com, 2010.
[2] Ying-Ze Liao, "System Design and Implementation
of AXI Bus", National Chiao Tung University,
October 2007.
[3] Clifford E. Cummings, "Coding And Scripting
Techniques For FSM Designs With Synthesis-
Optimized, Glitch-Free Outputs," SNUG (Synopsys
Users Group Boston, MA 2000) Proceedings,
September 2000.
[4] Clifford E. Cummings, “Synthesis and Scripting
Techniques for Designing Multi-Asynchronous
Clock Designs,” SNUG 2001
[5] Chris Spear, "System Verilog for Verification,](https://image.slidesharecdn.com/ijsrdv1i7031-140806064256-phpapp01/85/Design-and-Implementation-of-SOC-Bus-Based-on-AMBA-4-0-3-320.jpg)
![Design and Implementation of SOC Bus Based on AMBA 4.0
(IJSRD/Vol. 1/Issue 7/2013/0031)
All rights reserved by www.ijsrd.com 1517
2nd
Edition", Springer, ww.springeronline.com, 2008.
[6] Lahir, K., Raghunathan A., Lakshminarayana G.,
“LOTTERYBUS: a new high-performance
communication architecture for system-on-chip
designs,” in Proceedings of Design Automation
Conference, 2001.
[7] Sanghun Lee, Chanho Lee, Hyuk-Jae Lee, “A new
multi-channel on-chip-bus architecture for system-
on-chips,” in Proceedings of IEEE international SOC
Conference, September 2004.
[8] Martino Ruggiero, Rederico Angiolini, Francesco
Poletti, Davide Bertozzi, Luca 86
[9] Benini, Roberto Zafalon, “Scalability Analysis of
Evolving SoC Interconnect Protocols,” Int.
Symposium on System-on-Chip, 2004. Lukai Cai,
Daniel Gajski, “Transaction level modeling: an
overview,” in Proceedings of the 1st
IEEE/ACM/IFIP international conference on
Hardware/software codesign and system synthesis,
October 2003.
[10] Min-Chi Tsai, “Smart Memory Controller Design for
Video Applications,” Master thesis: National Chiao
Tung University, July 2006.](https://image.slidesharecdn.com/ijsrdv1i7031-140806064256-phpapp01/85/Design-and-Implementation-of-SOC-Bus-Based-on-AMBA-4-0-4-320.jpg)
Design and Implementation of SOC Bus Based on AMBA 4.0
- 1. IJSRD - International Journal for Scientific Research & Development| Vol. 1, Issue 7, 2013 | ISSN (online): 2321-0613 All rights reserved by www.ijsrd.com 1514 Abstract—ARM introduced the Advanced Microcontroller Bus Architecture (AMBA) 4.0 specifications in March 2010, which includes Advanced extensible Interface (AXI) 4.0. AMBA bus protocol has become the de facto standard SoC bus. That means more and more existing IPs must be able to communicate with AMBA 4.0 bus. Based on AMBA 4.0 bus, we designed an Intellectual Property (IP) core of Advanced Peripheral Bus (APB) Bridge, which translates the AXI4.0-lite transactions into APB 4.0 transactions. The bridge provides an interface between the high-performance AXI bus and low-power APB domain Keywords-SoC; AMBA; AXI; APB I. INTRODUCTION Integrated circuits have entered the era of System-on-a-Chip (SoC), which refers to integrating all components of a computer or other electronic system into a single chip. It may contain digital, analog, mixed-signal, and often radio- frequency functions – all on a single chip substrate. With the increasing design size, IP is an inevitable choice for SoC design. And the widespread use of all kinds of IPs has changed the nature of the design flow, making On-Chip Buses (OCB) essential to the design. Of all OCBs existing in the market, the AMBA bus system is widely used as the de facto standard SoC bus. On March 8, 2010, ARM announced availability of the AMBA 4.0 specifications. As the de facto standard SoC bus, AMBA bus is widely used in the high-performance SoC designs. The AMBA specification defines an on-chip communication standard for designing high-performance embedded microcontrollers. The AMBA 4.0 specification defines five buses/interfaces [1]: 1 Advanced extensible Interface (AXI) 2 Advanced High-performance Bus (AHB) 3 Advanced System Bus (ASB) 4 Advanced Peripheral Bus (APB) 5 Advanced Trace Bus (ATB) AXI, the next generation of AMBA interfaces Defined in the AMBA 4.0 specification, is targeted at high performance, high clock frequency system designs and includes features which make it very suitable for high speed sub-micrometer interconnect: 1 separate address/control and data phases 2 support for unaligned data transfers using byte strobes 3 burst based transactions with only start address issued 4 issuing of multiple outstanding addresses 5 easy addition of register stages to provide timing closure II. TOP BLOCK A. Block Diagram In this study, we focused mainly on the implementation aspect of an AXI4-Lite to APB Bridge. The APB Bridge provides an interface between the high-performance AXI domain and the low-power APB domain. It appears as a slave on AXI bus but as a master on APB that can access up to sixteen slave peripherals. Read and write transfers on the AXI bus are converted into corresponding transfers on the APB. The AXI4- Lite to APB bridge clock diagram is shown in Figure. 1. Fig. 1: Block Diagram B. Signal Connections Figure.2 shows the component signal connections. The bridge uses: Fig. 2: Signal Connections Design and Implementation of SOC Bus Based on AMBA 4.0 R. Gopi Reddy1 M. Sanjeeva Reddy2 1 M. Tech. 2 Professor 1,2 Kakatiya Institute of Technology & science, Warangal
- 2. Design and Implementation of SOC Bus Based on AMBA 4.0 (IJSRD/Vol. 1/Issue 7/2013/0031) All rights reserved by www.ijsrd.com 1515 1 AMBA AXI-Lite signals as described in the AMBA AXI-Lite 4.0 protocol specification. 2 AMBA APB signals as described in the AMBA APB 4.0 protocol specification. C. AXI Handshake Mechanism In AXI 4.0 specification, each channel has VALID and READY signals for handshaking [2] . The source asserts VALID when the control information or data is available. The destination asserts READY when it can accept the control information or data. Transfer occurs only when both the VALID and READY are asserted. Figure 3 show all possible cases of VALID/READ handshaking. Note that when source asserts VALID, the corresponding control information or data must also be available at the same time. The arrows in Figure 3 indicate when the transfer occurs. A transfer takes place at the positive edge of clock. Therefore, the source needs a register input to sample the READY signal. In the same way, the destination needs a register input to sample the VALID signal. Considering the situation of Figure.3(c), we assume the source and destination use output registers instead of combination circuit, they need one cycle to pull low VALID/READY and sample the VALID/RE ADY again at T4 cycle. When they sample the VALID/REA DY again at T4, there should be another transfer which is an error. Therefore source and destination should use combinational circuit as output. In short, AXI protocol is suitable register input and combinational output circuit. The APB bridge buffers address, control and data from AXI4-Lite, drives the APB peripherals and returns data and response signal to the AXI4-Lite. It decodes the address using an internal address map to select the peripheral. The bridge is designed to operate when the APB and AXI4-Lite have independent clock frequency and phase. For every AXI channel, invalid commands are not forwarded and an error response generated. That is once an peripheral accessed does not exist, the APB bridge will generate DE CERR as response through the response channel (read or write). And if the target peripheral exists, but asserts PSLVERR, it will give a SLVERR response. Fig. 3: Handshake Mechanism III. CLOCK DOM AIN CROSSING (CDC) A clock domain crossing (CDC), or simply clock crossing, is when a signal crosses from o ne clock domain into another [3]. If a signal does not assert long enough and is not registered, it may appear asynchronous on the incoming clock boundary. Meta stability happens when signal changes within the setup/hold time window. Synchronizing a signal that crosses into a higher clocked domain can be accomplished by registering the signal through a flip-flop that is clocked by the source domain, thus holding the signal long enough to be detected by the higher clocked destination domain. Synchronizing a signal traversing into a slower clock domain is more cumbersome [4]. This typically requires a register in each clock domain with a form of feedback from the destination domain to the source domain, indicating that the signal was detected. A. Metastability Metastability cannot be avoided, but a solution for handling the metastable signal enables proper functioning of the design [5].The metastability occurrences can be predicted by using the mean time between failures (MTBF) formula (1)[6] : Where C1 and C2 are constants that depend on the technology used to build the flip-flop; tMET is the duration of the metastable output; and fclk and fdata are the frequencies of the synchronous clock and the asynchronous input, respectively. MTBF = B. Synchronizer Designers can use special metastable hardened flops for increasing the MTBF [7]. For example, in Figure.4, a synchronizer flop is used following the signal DB. So instead of the metastable signal DB being used in the function downstream as in Figure 4, the stable signal DB2 is used in the logic downstream [8]. In the AXI4-Lite to APB bridge, we use synchronizer block designs for communicate between the AXI and APB clock domain. Fig. 4: Two flip-flop synchronizer
- 3. Design and Implementation of SOC Bus Based on AMBA 4.0 (IJSRD/Vol. 1/Issue 7/2013/0031) All rights reserved by www.ijsrd.com 1516 IV. FINITE STATE MACHINE FSM) A finite state machine is a mathematical abstraction sometimes used to design digital logic or computer programs [9]. It is a behavior model composed of a finite number of states, transitions between those states, and actions, similar to a flow graph in which one can inspect the way logic runs when certain conditions are met [10] . The state transition diagram is a picture of our state machine model. Figure 5 is the state transition diagram of our FSM. The state machine operates through the following states: IDLE. This is the default state of the FSM. WRITE_SETUP. When a write transfer request is asserted, the FSM moves into the WRITE_SETUP state. READ_SETUP. When a read transfer request is asserted, the FSM moves into the READ_SETUP state. WRIET_ACCESS. The enable signal, PENABLE, is asserted in the WRITE_ACCESS state. READ_ACCESS. The enable signal, PENABLE, is asserted in the READ_ACCESS state. WRITE_WAIT. When the AXI write response channel is not ready for receiving signal BRESP, then stay in WRITE_WAIT state. READ_WAIT. When the AXI read data channel is not ready for receiving signal RRESP, then stay in READ_WAIT state. States READ_WAIT and WRITE_WAIT are added, because the APB is not pipelined, wait states are added during transfers between the APB and AXI interface. System Verilog source code sample of the enumerated type encoded FSM is given below enum logic [2:0] {IDLE = 3'b000, READ_SETUP = 3'b001, WRITE_SETUP = 3'b010, READ_ACCESS = 3'b011, WRITE_ACCESS = 3'b100, READ _WAIT = 3'b101, WRITE_WAIT = 3'b110} Current State, Next State Fig. 5: State Transition Diagram According AXI specification, the read address channel, write address channel and write data channel are completely independent. Each channel has a set of forward signals and a feedback signal for handshaking. A read and a write requests may be issued simultaneously (AWVALID/WVALID and ARVALID are asserted high simultaneously) from AXI4-Lite, the AXI4-Lite to APB bridge will give more priority to the read request than to the write request. That is, when both write and read requests are valid, the write request is initiated on APB after the read is requested on APB. V. SIMULATION & IMPLEMENTATION The timing diagrams shown above illustrate the AXI4-Lite to APB bridge operation for various read and write transfers. It shows that when both read and write requests are active, read is given more priority. VI. CONCLUSION In this study, we provide an implementation of AXI4-Lite to APB Bridge which has the following features: 32-bit AXI slave and APB master interfaces. PCLK clock domain completely independent of ACLK clock domain. Support up to 16 APB peripherals. Support the PREADY signal which translates to wait states on AXI. An error on any transfer results in SLVERR as the AXI read/write response. REFERENCES [1] ARM, "AMBA Protocol Specification 4.0", www.arm.com, 2010. [2] Ying-Ze Liao, "System Design and Implementation of AXI Bus", National Chiao Tung University, October 2007. [3] Clifford E. Cummings, "Coding And Scripting Techniques For FSM Designs With Synthesis- Optimized, Glitch-Free Outputs," SNUG (Synopsys Users Group Boston, MA 2000) Proceedings, September 2000. [4] Clifford E. Cummings, “Synthesis and Scripting Techniques for Designing Multi-Asynchronous Clock Designs,” SNUG 2001 [5] Chris Spear, "System Verilog for Verification,
- 4. Design and Implementation of SOC Bus Based on AMBA 4.0 (IJSRD/Vol. 1/Issue 7/2013/0031) All rights reserved by www.ijsrd.com 1517 2nd Edition", Springer, ww.springeronline.com, 2008. [6] Lahir, K., Raghunathan A., Lakshminarayana G., “LOTTERYBUS: a new high-performance communication architecture for system-on-chip designs,” in Proceedings of Design Automation Conference, 2001. [7] Sanghun Lee, Chanho Lee, Hyuk-Jae Lee, “A new multi-channel on-chip-bus architecture for system- on-chips,” in Proceedings of IEEE international SOC Conference, September 2004. [8] Martino Ruggiero, Rederico Angiolini, Francesco Poletti, Davide Bertozzi, Luca 86 [9] Benini, Roberto Zafalon, “Scalability Analysis of Evolving SoC Interconnect Protocols,” Int. Symposium on System-on-Chip, 2004. Lukai Cai, Daniel Gajski, “Transaction level modeling: an overview,” in Proceedings of the 1st IEEE/ACM/IFIP international conference on Hardware/software codesign and system synthesis, October 2003. [10] Min-Chi Tsai, “Smart Memory Controller Design for Video Applications,” Master thesis: National Chiao Tung University, July 2006.