535

Page 1

Short Paper Proc. of Int. Conf. on Control, Communication and Power Engineering 2013

Design and Analysis of Xilinx Verified AMBA Bridge for SoC Systems Shaila S Math1, Veerabhadrayya Math2 1

BMS Institute of Technology, Department of Electronics and Communication Engineering, Bangalore, India Email: shaila.s.math@gmail.com 2 Srinivas School of Engineering, Department of Electronics and Communication Engineering, Mangalore, India Email: veerabhadrayya.math@ymail.com Abstract: ARM introduced the Advanced Microcontroller Bus Architecture (AMBA) 4.0 and its specifications define five buses/interfaces: Advanced eXtensible Interface Bus (AXI), Advanced High-performance Bus (AHB), Advanced System Bus (ASB), Advanced Peripheral Bus (APB) and Advanced Trace Bus (ATB). That means more and more existing Intellectual Property (IP) must be able to communicate with AMBA4.0 bus. This paper presents an IP core design of APB Bridge, to provide interface between AXI-Lite bus and APB bus operating at different frequencies. The maximum operating frequency of the module is 168.464MHz. Test cases are run to perform multiple read and write operations. Synthesis and Simulation is done using Xilinx ISE and Modelsim. Index Terms: AMBA, AXI, IP, Xilinx

Figure 1. System On-chip Interconnect showing AMBA/ CoreConnect/ Wishbone Buses

I. INTRODUCTION

The AMBA data bus width can be 32, 64, 128 or 256 byte, address bus width will be 32bits wide. The AMBA AXI4 [5] specification to interconnect different modules in a SoC was released in March 2010. The key features of AMBA4 including its all versions (AXI4 and AXI4-Lite) are discussed in detail in [5]. Few key features used in designing the modules are discussed here.

In recent years due to the miniaturization of semiconductor process technology and computation for survival in the current market conditions constant customization is required. The semiconductor process technology is changing at a faster pace during 1971 semiconductor process technology was 10µm [1], during 2010 the technology is reduced to 32nm and future is promising for a process technology with 10nm. Intel, Toshiba and Samsung have reported that the process technology would be further reduced to 10nm in the future. So with decreasing process technology and increasing consumer design constraints System-on-Chip (SoC) has evolved, where all the functional units of a system are modeled on a single chip. SoC buses are used to interconnect an Intellectual Property (IP) core to the surrounding interface. These are not real buses, but they reside in Field Programmable Gate Array (FPGA). Each bus/interface have different features, the designer chooses the bus, best suited for his application. Some challenges faced in SoC buses are latency, high bandwidth, arbitration schemes, number of masters and slaves connected on bus and performance of bus according to different application. Fig. 1 shows the exact picture where the SoC buses reside in a chip. Masters can be Central Processing Unit (CPU), Moving Picture Expert Group (MPEG), where as slaves can be memory, Arithmetic and Logical Unit (ALU) etc. Currently there are a few publicly available systemon-chip buses such as AMBA from ARM [2], Core Connect from IBM [3], Wishbone from Silicore [4] and others. The number of masters and slaves depends on application. © 2013 ACEEE DOI: 03.LSCS.2013.2.535

A. AMBA AXI4 Features of AXI4 includes the following: • Burst lengths up to 256 beats is supported • Quality of Service (QoS) signaling • Multiple region interface is supported • Write response requirements are updated • Update of AWCACHE and ARCACHE signaling details • For ordering requirements additional information is provided • Optional User signaling details • Locked transactions and write interleaving are removed B. AMBA AXI4-Lite Features of AXI4-Lite interface are: • All transactions burst length is 1 • All data access size is same as width of data bus • Supports 32-bit or 64-bit data bus width • All accesses are equivalent to AWCACHE or ARCACHE equal to b0000 • For exclusive accesses support is not provided. C. AMBA APB The APB bus is optimized for minimal power consumption and reduced interface complexity. The APB peripherals can 32


Short Paper Proc. of Int. Conf. on Control, Communication and Power Engineering 2013 TABLE 1: SIGNAL D ESCRIPTIONS

be integrated easily to any design flow, with the following advantages: • High-frequency operation can be easily achieved • Use of a single clock edge to simplify the static timing analysis • No special considerations are required for automatic test insertion • Many Application Specific Integrated Circuit (ASIC) libraries have a better selection of rising edge registers • Easy integration with cycle-based simulators. D. Handshake Mechanisms of AXI-Lite and APB All five channels of AMBA use the same VALID/READY handshake to transfer data and control information. Both master and slave are controlled by this two way mechanism. When the data or control information is available on the source it generates VALID signal. The destination generates the READY signal to indicate that it accepts the data or control information. When both the VALID and READY signals are HIGH transfers will occur. On both master and slave interfaces there must be no combinatorial paths between input and output signals. Table 1 gives the information of signals used in the complete design of the module. Fig. 2 to Fig. 4 show examples of the handshake sequence. In Fig. 2, the source drives the VALID signal HIGH when it gets data or control information. The destination drives READY signal HIGH if it is ready to accept the data or information sent by the source. The arrow shows when the transfer occurs. It is not permitted to wait until READY is asserted before asserting VALID. Once VALID is asserted it must remain asserted until the handshake occurs.

Signal

Input/ Output

Aclk

Global

Input

AResetn AWADDR WSTRB

Global Master Master

Input Input Input

AWPROT

Master

Input

WDATA ARADDR

Master Master

Input Input

ARPROT

Master

Input

RDATA

Master

Input

AWVALID

Master

Output

AWREADY

Slave

Output

WVALID ARVALID RVALID WREADY BRESP RRESP

Master Slave Slave Slave Slave Slave

Output Output Output Output Output Output

BVALID

Slave

Output

BREADY RREADY

Master Master

Output Output

ARREADY

Slave

Output

PCLK

Global

Input

PRESETn

Global APB bridge APB bridge APB bridge APB bridge

Input

Global clock signal. Global reset signal Write address Write burst type Write protection type Write data Read address Read Protection type Read data Write address valid Write address ready Write valid Read address valid Read valid Write ready Write response. Read response. Write response valid. Response ready. Master ready. Read address ready. Global clock signal Global reset signal

Input

APB address bus

Input

Protection type

PPROT PSELx

PENABLE

PWRITE

Figure 2. VALID before READY handshake

© 2013 ACEEE DOI: 03.LSCS.2013.2.535

AMBA B RIDGE

Source: master/ slave/APB Bridge

PADDR

PWDATA

In Fig. 3, before the data or control information is valid, destination drives READY HIGH. The data or control information is accepted in a single cycle, as soon as it becomes valid. The arrow shows when the transfer occurs. It is permitted to wait for VALID to be asserted before the corresponding READY is asserted. If READY is asserted, it is permitted to de assert READY before VALID is asserted. In Fig. 4, both the source and destination drives READY and VALID signal at the same cycle. Here the transfer can be data or control information. In this case the transfer occurs immediately. The arrow shows when the transfer occurs. In this paper interfacing bridge of AXI4-Lite and APB [6] is designed and verified. Here we used Verilog HDL [7] (Hardware Description Language) for designing the RTL (Register Transfer Level) code. Synthesis and Simulation is done using Xilinx ISE and Modelsim [8]. The rest of the paper is organized as follows: Section 2 discusses related work. Sec

OF

PSTRB PREADY PRDATA PSLVERR

Input

Input

APB bridge APB bridge APB bridge Slave interface Slave interface Slave interface

Description

Select line for Slave Enable signal indicates the second and subsequent cycles of an APB transfer

Input

Direction

Input

Write data

Input

Write strobes size

Output

Response ready

Input

Read data

Output

Transfer failure

Figure 3. READY before VALID handshake

33


Short Paper Proc. of Int. Conf. on Control, Communication and Power Engineering 2013 Lite of the AMBA on the simulator Riviera was discussed in [24]. IP solutions for AMBA interconnect was discussed in [25]. III. PROPOSED WORK In this work an IP core of APB bridge is designed to provide interface between AXI-Lite bus and APB bus operating at different frequencies. The maximum operating frequency of the module is 168.464MHz. 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 AXI4Lite to APB Interfacing module block diagram is shown in Fig. 5. The test layer shown in the Fig. 5 tests the module for multiple read and write operations.

Figure 4. VALID with READY handshake

tion 3 of this paper, discusses proposed work. In Section 4, simulation parameters and results are discussed. Future scope and concluding remarks are given in Section 5. II. RELATED WORK Milica Miti and et al gives the survey of various buses On-Chip Buses [9]. Computer architecture [10] defines the physical structure of the CA. Types of computer architecture are single shared bus, hierarchical bus, token ring bus, crossbar bus or custom networks. Protocols are defined for communication channels/buses. Each bus/interface have different features, the designer chooses the bus, best suited for his application. Various on-chip bridging techniques in short are discussed in this section. AMBA’s design methodology and set of specifications that will aid designers in making detailed comparisons with other buses was discussed in detail in [11]. Designing AHB/PCI bridge was discussed in [12]. Here function of AHB/PCI bridges to map various control signals and address spaces from one bus into those of another was explained in detail. On-chip communication architectures for reconfigurable System-on-Chip [13] surveys existing solutions for SoC and analyses the suitability for SoC application. In [14] multichannel on-chip-bus architectures for system-on-chips were discussed. An on-chip bus bridge between heterogeneous buses with different clock frequencies was discussed in [15]. Design and implementation of AXI bus protocol-based on MPSoC architecture was discussed in [16] which uses Altera Stratix II EP2S180 Field Programmable Gate Array (FPGA). An IP core of APB bridge, which translates the AXI4.0-lite transactions into APB 4.0 transactions was discussed in detail in [17] and [18]. In [19] the analysis was carried out by simulating a multicore platform with a shared-link AXI backbone running a video phone application. The performance was evaluated in terms of bandwidth utilization, average transaction latency and system task completion time. Design of AHB2APB bridge simulated using Xilinx and Modelsim was discussed in [20]. AMBA to SoCWire network-on-chip bridge [21] presents an efficient connection between a SoCWire network and a LEON processor bus systems. Direct memory access enables the AHB2SOCW bridge to operate efficiently. Design and analysis of AMBA AXI4 system for various read and write operations using VCS tool [22] by Synopsis was discussed in [23]. The system level modelling of the Advanced AHBŠ 2013 ACEEE DOI: 03.LSCS.2013.2.535

Figure 5. Block diagram of AXI4-LITE and APB interface

A. Signal Description Fig. 6 shows the master and slave component signal connections. The Interfacing module uses AMBA AXI-Lite signals and APB signals as described in the AMBA protocol specification. The read and write operation signals B. Finite State Machines (FSM) Fig. 7 shows FSM [26] for operational activities of the module for read and write operation. When reset signal goes high the module gets address and data information from test layer. Either it is read or write operation, the module will wait for VALID signal. For read operation the data is read and on reading the last byte RLAST signal is raised. Acknowledgement signal is received on end of the operation to master module. If it is a multiple read operation, than the process is continued or else new address and data information is received from the test layer. For write operation the data is written to the destination location and on writing the last byte WLAST signal is raised. Acknowledgement signal is received on end of the operation to master module. If it is a multiple write operation, than the process is continued or else new address and data information is received from the test layer. 34


Short Paper Proc. of Int. Conf. on Control, Communication and Power Engineering 2013

Figure 7. Operational activity of the module

Figure 6. Signals used to design interface between AMBA AXI4 and APB module

1) Write address operation Fig.8 shows the FSM for write address operation. In state1 AWVALID is set to zero. When Pending_tran_write_addr_reg ==1 then it transits to state S2. In state2 AWVALID is set to one. When pending_trans_write_addr_reg =1(more addresses) it will remain in the same state if there are more addresses to be transmitted or if AWREADY ==0.

Figure 8. FSM for write address operation

RVALID==1 and RLAST==0 then it will go to state S2. If RVALID==0 then it will remain in the same state S1. Further state transitions will be done according to the state diagram (Fig. 11).

2) Write data operation Fig.9 shows the FSM for write data operation. In state1 if WVALID==1 and WLAST==0 then it will go to state S2. If WVALID==0 then it will remain in the same state S1. WREADY=0 in state S1. Further state transitions will be done according to the state diagram (Fig. 9). 3) Read address operation Fig.10 shows the FSM for read address operation. In state1 ARVALID is set to zero. When Pending_tran_read_addr_reg ==1 then it transits to state S2. In state2 ARVALID is set to one. When Pending_tran_write_addr_reg==1 then it will come to this state and remains in the same state until AWREADY is high && Pending_tran_write_addr_reg==1. 1) Read data operation Fig.11 shows the FSM for read data operation. In state1 if Š 2013 ACEEE DOI: 03.LSCS.2013.2.535

IV. RESULT The timing diagram shown in Fig. 12 illustrates the AXI4Lite to APB Interfacing module operation for various read and write transfers. Synthesis and Simulation is done using Xilinx and Modelsim. It shows that when both read and write requests are active, read is given more priority. For write operation address locations passed to module are 31, 33 and 35; for read operations 32, 34 and 36. After synthesis summary reports generated by the tool is shown in Fig. 13. 35


Short Paper Proc. of Int. Conf. on Control, Communication and Power Engineering 2013

Figure 9. FSM for write data operation

Figure 12. Simulation results of multiple read and write data opera tion

Figure 10. FSM for read address operation

Figure 13. Area report of simulation

CONCLUSIONS

AND FUTURE SCOPE

A. Future Scope The AMBA AXI4 has limitations with respect to the burst data and beats of information to be transferred. The burst must not cross the 4k boundary. Bursts longer than 16 beats are only supported for the INCR burst type. Both WRAP and FIXED burst types remain constrained to a maximum burst

Figure 11. FSM for read data operation

Š 2013 ACEEE DOI: 03.LSCS.2013.2.535

36


Short Paper Proc. of Int. Conf. on Control, Communication and Power Engineering 2013 length of 16 beats. These are the drawbacks of AMBA AXI4 system which need to be overcome.

[14] Sanghun Lee, Chanho Lee, Hyuk-Jae Lee, “A new multi-channel on-chip-bus architecture for system-on-chips,” in Proceedings of the IEEE International SOC Conference, pp.20-27, September 2004. [15] Sangik Choi and Shinwook Kang, Mobile SamsungElectronics Co.,Ltd, “Implementation of an On-Chip Bus Bridge between Heterogeneous Buses with Different Clock Frequencies”. IEEE, IDEAS’05, pp. 530-534, July 2005 [16] Fu-ming Xiao, Dong-sheng Li ; Gao-ming Du ; Yu-kun Song ; Duo-li Zhang ; Ming-lun Gao “Design of AXI bus based MPSoC on FPGA”3rd IEEE international conference on Anticounterfeiting, Security, and Identification in Communication, ASID 2009, pp. 560-564, August 2009 [17] Jaehoon Song, Juhee Han ; Dooyoung Kim ; Hyunbean Yi ; Sungju Park “Design Reuse of on/off-Chip Bus Bridge for Efficient Test Access to AMBA-based SoC”, 16th Asian Test Symposium, pp. 193 – 198, October 2007. [18] Chenghai Ma, Zhijun Liu ; Xiaoyue Ma “Design and implementation of APB bridge based on AMBA 4.0”, International Conference on Consumer Electronics, Communications and Networks (CECNet), pp.193 – 196, April 2011 [19] N.Y.C. Chang, Y.Z. Liao, T.S. Chang, “Analysis of shared-link AXI” IET Computer Digit. Tech., vol. 3, no. 4, pp. 373–383, July 2009 [20] Vani.R.M and M.Roopa “Design of AMBA Based AHB2APB Bridge” IJCSNS International Journal of Computer Science and Network Security, vol.10 no.11, pp. 14-18, November 2010 [21] H. Michel, F. Bubenhagen, B.Fiethe, H.Michalik,B.Osterloh, W.Sullivan, A.Wishart, J.Ilstad, S.A.Habinc, “AMBA to SoCWire network on Chip bridge as a backbone for a Dynamic Reconfigurable Processing unit” NASA/ESA Conference on Adaptive Hardware and Systems (AHS), pp.227 – 233 June 2011, [22] Synopsys, VCS / VCSi User Guide Version 10.3, , Available at, www.synopsys.com, March 2008 [23] Shaila S Math, Manjula R.B,S S Manvi, P. Kaunds, “Data transactions on system-on-chip bus using AXI4 protocol” in Proceedings of the IEEE International Conference on Recent Advancements in Electrical, Electronics and Control Engineering (ICONRAEeCE), pp. 423 – 427, Dec 2011 [24] Richa Sinha, Akhilesh Kumar and Archana Kumari Sinha “Verification analysis of AHB-Lite protocol with coverage” International Journal of Advances in Engineering & Technology, Vol. 2, Issue 1, pp. 121-128, Jan 2012 [25] “DesignWare IP solutions for AMBA interconnect” Synopsis Inc. Available at http://www.synopsys.com/products/ designware/amba_solutions.html,2009 [26] 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.

B. Conclusion The APB bridge using Verilog HDL to provide interface between AXI-Lite bus and APB bus operating at different frequencies has been designed. APB bridge designed has got following features: APB master interface and 32-bit slave interfaces are provided; Clock domains PCLK and ACLK are completely independent; The designed module supports 16 APB peripherals; READY signal of APB is translated to wait states on AXI; and AXI gives SLVERR signal if error occurs on read/write response. REFERENCES [1] Chifeng Wang; Wen-Hsiang Hu; Bagherzadeh, N. “A Wireless Network-on-Chip Design for Multicore Platforms” in Proceedings of the 19 th IEEE Euromicro International conference on Parallel, Distributed and Network-Based Processing (PDP), February 2011, pp.409 – 41 [2] “AMBA specifications V2.0”, ARM. Available at http:// www.arm.com, 1999 [3] “The CoreConnect Bus Architecture”, IBM Microelectronics. Available at http://www.ibm.com/chips/products/coreconnect [4] “Wishbone system-on-chip (soc) interconnection architecture for portable IP cores, revision: B.3,” OPENCORES. Available at http://cdn.opencores.org/downloads/wbspec_b3.pdf, 2002 [5] “AMBA AXI Protocol Specification” (Rev 2.0), ARM. Available at http://www.arm.com, 2010 [6] IP AXI to APB Bridge (v1.00a) DS788, LogiCORE, Product Specification, June 2011 [7] Samir Palnitkar, Verilog HDL: A Guide to Digital Design and synthesis, 2nd ed, Prentice Hall PTR Publication, 2003 [8] Xilinx ISE Synthesis and Verification Design Guide Available at www.xilinx.com, 2008 [9] Milica Miti´c and Mile Stoj¡cev., “ An Overview of On-Chip Buses”, FACTA UNIVERSITATIS, 2006, pp. 405-428 [10] J. Hennessey and D. Patterson, Computer Architecture—A Quantitative Approach, 2nd ed., Morgan Kaufmann, San Mateo, Calif., 1996. [11] Flynn, D. Adv. RISC Machines Ltd., Cambridge, “AMBA: enabling reusable on-chip designs”, IEEE Micro, Publication, pp. 20-27, Jul/Aug 1997. [12] Wang Zhonghai,Ye Yizheng,Wang Jinxing, and Yu Mingyan, “Designing AHB/PCI Bridge,”in Proceedings of 4 th IEEE International Conference on ASIC, pp.578-580, October 2001. [13] A.S. Lee, N.W. Bergmann “On-chip communication architectures for reconfigurable System-on-Chip” in Proceedings of 2003 IEEE International Conference on FieldProgrammable Technology (FPT), pp.332-335, December 2003.

© 2013 ACEEE DOI: 03.LSCS.2013.2.535

37


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.