Area Efficient VHDL implementation of AHB arbiter IP

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 08 | Aug -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 1483
Area Efficient VHDL implementation of AHB arbiter IP
Tamanna Sheikh1, Dr. Pranay Joshi 2, Dr. Sunil joshi3
1 M.Tech. Student, Department of Electronics and Communication, CTAE, Udaipur,
Rajasthan, India
2Faculty, Department of Electronics and Communication, CTAE, Udaipur, Rajasthan, India
3Professor & Head, Department of Electronics and Communication, CTAE, Udaipur, Rajasthan, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In SoC a processor needs to interact with
other processors,memories or input/output devicescto
complete the task. In SoC data integrity become most
vital challenge and communication planning under
different processors with shared bus system needs
bottleneck free communication.To justify the
multiprocessor environment a central bus controller is
required called as arbiter. The propose work develop
the IP of AHB arbiter for AMBA bus solution in VHDL.
Proposed arbiter can access handshaking signals of 16
master and split capable slave and give grant to
respective master according to priority defined by
round robin algorithm. The design is implemented and
tested in Xilinx. Further area, speed and power is
calculated as a performance evaluation parameters.
Key Words: AMBA, Arbiter, Round-Robin, IP, SOC,
VHDL.
1.INTRODUCTION
Advanced Microcontroller Bus Architecture (AMBA)
[1] is not properly speak a bus, but a family of buses,
defined by ARM. The main buses of this Family are
Advanced eXtensible Interface (AXI), Advanced High-
Performance Bus (AHB) and Advanced Peripheral
Bus (APB). APB [2] is a bus for interconnecting low-
rate IPs, i.e. having little data to be transferred, for
example keyboard-like IPs or UART (Universal
Asynchronous Receiver Transmitter asynchronous,
managing serial links). APB allows a very simple and
very small read / write operations, with only one
master per bus. In particular, this bus allows only
unit operations, that is to say transfer a single data
word by transfer request. As soon as the IP becomes
more bandwidth-intensive, APB and its specifications
are becoming a limiting factor. In particular, the
burst mode, which allows several words to be
transferred after a single query operation, is not
present in APB.AHB is a bus that allows reads /
writes of different sizes and supporting the burst
mode, as well as several masters. The different
masters in competition for access to the bus, the
latter has an arbitrator to distribute access. It is
possible to perform operations on 4, 8 or 16
successive words, but also on unspecified burst sizes,
which will continue as long as the requesting IP
requires it, or the bus arbitrator decides to give
Access to another IP. This bus also has error signals
allowing slaves to notify the master who initiated the
transaction that not succeed. These two types of
buses are relatively conventional, each having its
interest depending on the type of IPs to be linked.In a
system is constantly increasing. Depending on the
case, access to a single bus, central, for all IPs, can
constitute a limiting bottleneck.
2.AMBA BUS
The AMBA 2.0 bus is a built-in bus in SOC design. The
AMBA standard is documented, free access and is
used for System-on-chip processors (example:
smartphones). The specification contains BUS types:
AHB (Advance High-performance Bus), ASB
(Advance System Bus) and APB (Advance Peripheral
Bus). The arbitrator / bus controller is configurable
in order to perform various arbitration modes,
"Round-robin" or "priority-fix" mode. When a query
from a master is supported by the BUS arbiter, the
request is transmitted to all the slaves and the
arbitrator selects the slave concerned. The AHB bus
is multiplexed and therefore can be implemented in
FPGAs, as can be moreover see it in the following
figure:

Fig-1: AHB controller principle diagram (AHBCTRL)
The bus arbiter allows one master at a time to
communicate to the slaves. Since the processors are
master and the slave memory controller are on the
AHB bus, our component will have to monitor the
actions of the components (AHB master) that access
the memory controller. The bus arbiter allows one
master at a time to communicate to the slaves. Since
the processors are master and the slave memory
controller are on the AHB bus, our component will
have to monitor the actions of the components (AHB
master) that access the memory controller.
3.LITERATURE REVIEW
The author present a self-motivated arbitration
scheme according to data length transfer form
master [4]. They have combined the length and
priority to process the arbiter, this scheme work well
with priority, but dynamic behaviour claimed, will
create bottleneck under different clock structure [5].
Author present similar type of work by multilayer
based design approach to justify the self-motivated
arbitration, for dynamic behaviour justification of
arbitration. But it cost additional hardware to create
different layer for arbiter and maximum allowable
latency is already defined as 16 clock for any type of
transfer, hence multilayer approach for AHB2.0 is not
suggestible, when concern of area is there. In [8],
authors explore the performance under different
beat and burst mode of operation in AHB arbiter, and
its impact on power and throughput.[8] Most of work
carried in 4 master and single slave To show the
arbitration, our proposed work takes 16 master [9]
with burst and beat capabilities and slave with split
capabilities.
4.ARBITER DESIGN
 To design round robin arbiter in VHDL.
 The arbiter is capable to handle 16 bus
request and gives centralized command as a
grant to respective master in priority list. To
avoid bottleneck, protocols claims maximum
latency of 16 clk.
 Split and error handling capable features in
arbiter according to protocol.
 Operation and control according to burst
mode.
4.1 Priority block:
The priority logic block is implemented through FSM
approach. The priority scheme follows the Round
Robin theorem of priority .The bus request have
highest priority will get Grant First and rest of
request will wait for their priority .In1 signal is the
input for this block that is nothing but the Bus_req,
Out1,Out2…….Out16 these are the Grant signal as
output, these signal is further OR and sent to the
output port,as Shown in figure 2.
The proposed arbiter consist of priority shift and
priority logic with round robin algorithm. There will
be separate controller and counter to track the
operation according the operation initiated by
master in AHB system. There will be some or gates
and an encoder to combine the different grants from
arbiter. The flows follows the sequence initiated by
different master as a bus request to access the bus,
arbiter takes the responsibility for the monitoring of
handshaking signals from different slave and
controls from master. According to priority arbiter
generates the grant to respective master. The
following are the objective or scope of work.
Objective :

Contro-
ller
or gate
or gate
or gate
priority logic
block 16
priority logic
block 2
priority
logic block
1
encoder
for
generate
bus master
no
counter
interface
for burst
transfer
bus_req16
bus_req1
grant1
grant2
grant16
bus
master
no
enable1
enable2
enable16
data_done
out1
out2
out16
or
gate
bus_req
2
Priority shift
Fig 2: Block diagram of AHB arbiter

Fig 3: State diagram for priority logic
rst M1
M2
M3
M4
M5
M6
M7
M8M9
M10
M11
M12
M13
M14
M15
M16
in3=0
in1=0
in2=0
in1=1
in4=0
in5=0
in6=0
in7=0
in8=0
in9=0
in10=0
in11=0
in12=0
in13=0
in14=0
in15=0
in16=0
error=1
in2=1
in3=1
in4=1
in5=1
in6=1
in7=1
in8=1in9=1
in10=1
in11=1
in12=1
in13=1
in15=1
in14=1
in16=1
in1=1in16=1
in15=1
in14=1
in13=1
in2=1
in3=1
in4=1
in5=1
in6=1
in7=1
in8=1
in10=1
in11=1
in12=1

4.2 Counter and Mux
This block contains 16:1 Mux, and Counter. The
counter is used here to control the all operation of
arbiter with respect to the master and slave input. All
Bus_req are mapped with Mux and with the help of
select line Mux will select the appropriate master
Busreq ,because the undefined burst is depend upon
the Master bus_req.As counter is used to generate
the start of transfer and end of transfer Depends
upon the burst.
Fig-4: RTL schematic for top entity
Synthesis Report:
Device utilization summary:
Selected Device : 4vlx15sf363-12
Number of Slices: 1273 out of 6144 20%
Number of Slice Flip Flops: 830 out of 12288 6%
Number of 4 input LUTs: 2297 out of 12288 18%
Number of IOs: 64
Number of bonded IOBs: 64 out of 240 26%
IOB Flip Flops: 4
Number of GCLKs: 1 out of 32 3%
Timing Summary:
Speed Grade: -12
Minimum period: 3.130ns (Maximum Frequency:
319.519MHz)
Minimum input arrival time before clock: 6.164ns
Maximum output required time after clock: 5.388ns
4. Simulation
Fig-5: Simulation of integrated Arbiter
The above figure depicts the functionality of
arbitration, where three request occurs
simultaneously and grant1 gets first access, once the
operation of Master 1 is over then grant is shifted to
request 2,and so on accordingly.
Fig-6: simulation of controller
The above figure depicts the simulation of controller,
it is finite state machine which take inputs from
other interface and gives control signal to counter for
synchronous operation. It works in two state “rst”
and “arbiter-operation”. Initially it will be in “rst”
state, once start signal is there then it will shift to
“arbiter-operation” state and gives outcomes as
operation start signal high.

Xilinx Power Summary of ARBITER:
Fig-7: Power report of AHB arbiter
Table 2: Comparison Table:
Previous
design [9]
Proposed design
Number of
Slices
1566 1273
Number of
Slice Flip Flops
533 830
Number of 4
input LUTs
2752 2297
Number of
bonded IOBs
64 64
IOB Flip Flops 4 4
Number of
GCLKs
2 1
POWER N.A 198.21mW
SPEED N. A. 319.519MHz
No. of Master
and Slave
16,1 16,1
The above table explains the comparative outcomes
of proposed method. In earlier approach there was
an interface block to connect the enable to all priority
blocks, which is with more conditional statements,
hence it has created the more latches with references
of 4 input Look up tables. The earlier approach has
used priority storage block along with finite state
machine based controller, which cost additional
hardware and complex handshaking. The proposed
method removes priority storage and fsm instead of
that priority shift is introduced with simple shift
controller by data_done coming from counter to
control the occurrence of shift. The proposed method
has removed the interface block and directly
connected the enable from priority shift to priority
logic blocks. This approach has reduces the hardware
from earlier work. Hence it is known for FPGA design
approach there are dedicated flip-flops are available
for design inference ,hence we have designed our
code in such a way which more flip-flops rather than
latches(4 input LUT),hence our Flip-flops counts are
more than earlier work,. Flip-flops based design
ensure more stability and predictability then latches.
Overall methods give better hardware utilization
than earlier approach.
5.CONCLUSION
The IP module has been designed as an Arbiter of the
AMBA AHB bus [2], so which can communicate with
the processor for it to configure, it also has the ability
to observe transfers between the processor and a
peripheral, and is capable of controlling the
communication that occur in such transfers. The
design consider the system model with bus request
of 16 master and one slave. Additional feature of split
control is also considered in proposed IP. The design
has been developed using VHDL code and
synthesized and simulation using Xilinx ISE. The
designed is performed in Virtex device of Xilinx and
claimed 5.5 % improvement in area occupancy in
devices. The speed and power is calculated as
319.519MHz and 198.21mW respectively. The
advantage of this design is that we have taken care of
latch formation, with less latch & maximum flip-flop
have enhanced our area efficiency.
REFERENCES
[1] https://www.arm.com/products/system-
ip/amba-specifications
[2] Bacciarelli, L., Lucia, G., Saponara, S., Fanucci,
L. and Forliti, M., 2006, June. Design, testing
and prototyping of a software programmable
I2C/SPI IP on AMBA bus. In Research in
Microelectronics and Electronics 2006, Ph.
D.(pp. 373-376). IEEE.
[3] Conti, M., Caldari, M., Vece, G.B., Orcioni, S.
and Turchetti, C., 2004, June. Performance
analysis of different arbitration algorithms of
the AMBA AHB bus. In Proceedings of the
41st annual Design Automation Conference
(pp. 618-621). ACM.
[4] Shete, P.S. and Oza, S., 2014. Design of an
AMBA AHB Reconfigurable Arbiter for On-
chip Bus Architecture. International Journal
of Application or Innovation in Engineering &
Management, 3(5), pp.245-252.
[5] Hwang, S.Y., Kang, D.S., Park, H.J. and Jhang,
K.S., 2010. Implementation of a self-

motivated arbitration scheme for the
multilayer AHB busmatrix. IEEE transactions
on very large scale integration (VLSI)
systems, 18(5), pp.818-830.
[6] Conti, M., Caldari, M., Vece, G.B., Orcioni, S.
and Turchetti, C., 2004, June. Performance
analysis of different arbitration algorithms of
the AMBA AHB bus. In Proceedings of the
41st annual Design Automation Conference
(pp. 618-621). ACM.
[7] Jamadagni, T. and Kumar, J.M.,
Implementation of a Special Arbitration
Scheme for AHB of AMBA.
[8] Mitić, M. and Stojčev, M., 2006. An overview
of on-chip buses. Facta universitatis-series:
Electronics and Energetics, 19(3), pp.405-
428.
[9] Gautam, P.K. and Upadhyay, N., A Novel
Arbitration Technique of AMBA AHB.

Area Efficient VHDL implementation of AHB arbiter IP

More Related Content

What's hot (19)

Similar to Area Efficient VHDL implementation of AHB arbiter IP (20)

More from IRJET Journal (20)

Recently uploaded (20)

Area Efficient VHDL implementation of AHB arbiter IP