Interfaces and Buses / AMBA / AMBA_AXI
Block Name: AMBA_AXI

Code File Location: VisualSim/actor/arch/Buses/AMBA_AXI

Table of contents

    1. Block Descriptioon

    2. Operation

    3. Standard internal delay - Read

    4. Standard internal delay - Write

    3. List of models

    4. Model Parameters, Expected Data structure and ports

    5. How to Connect

    6. Statsitics and Analysis

    7. Enabling plot outputs

    8. Enabling Power

    9. Errors and solutions

Description

The AMBA_AXI Bus protocol is targeted at high-performance, high-frequency system designs and includes a number of features that make it suitable for a high-speed sub micron interconnect. The objectives of the AMBA_AXI Bus are to be suitable for high-bandwidth and low-latency designs and enable high-frequency operation without using complex bridges. It meets the interface requirements of a wide range of components and it is suitable for memory controllers with high initial access latency and provide flexibility in the implementation of interconnect architectures.

    It can able to establish more than one outstanding transaction between different master and different slave. AXI bus can be configured to establish maximum transaction between the masters and slaves using threshold values. Multiple Masters can send transaction to a single slave with the help of bus arbiter.

             

For detailed information on the AXI Bus, refer to $VS/doc/Chapter_8-Bus_Standards.pdf.

Figure 1. AXI_Read_Flow_Diagram

Figure 2. AXI_Write_Flow_Request

The AXI Read Request and Write Requests are plotted on a timing diagram. To activate this, add the Timing_Diagram block Hardware_Setup->Timing_Diagram ) to the model and list the AXI Bus name in the Timing Diagram block.

Operation

    Single Master with multiple requests:

    Master device can send single request at time or multiple requests at a time or consecutive requests. AXI master will send a request to bus arbiter, master can send request or data only when the arbiter grant that master.

    Multiple requests are handled by maintaining the Buffer at each master, the length of the buffer determined by the Master_Request_Threshold parameter.

           

    Multiple Master with Multiple requests:

    Multiple masters can receive multiple request or data from the master device. Each master will send a bus request to bus arbiter, but the arbiter will grants the request based on the availability and Arbitration algorithm.

    Multiple requests will be buffered in each master, but sometimes the overflow can occur because of buffer full. User can observe this from stats out port                 (DEBUG parameter = true).

            NOTE: First word output will be sent through the port for both read and write packet, the remaining data transfer will be performed internally.

        Read request operation:

    1.     Master sends request for bus grant (handshaking)

    2.     Arbiter grant the master based on availability and the Arbitration algorithm

    3.     Master transfer the request (Control/Address) to Slave.

    4.     Data will be returned from slave device and starts it transfer from slave to master.

    5.     Master sends out the first word of the data to master device.

    6.     That specific master can imitate the next transaction. This can be obtained by the Master_Request_Threshold parameter.

 

        Standard internal delay - Read
                Forward Transfer: (4 cycle)
                      1. 1 Cycle -  Master port to Address Channel
                      2. 1 Cycle -  Arbitration Grant
                      3. 1 Cycle -  Address channel transfer through crossbar
                      4. 1 Cycle -  Address channel to Slave port

                Pipeline Flow:
               

                Backward Transfer: (Return Data)
                        N +2 Cycle - N data burst (eg: 8 byte Burst and 32 byte request - 4 cycle data transfer) , 1 Cycle for slave port to channel and 1 Cycle for Channel to master port

                        1. Output will be issued after 3 cycle as first word out.

•   1 Cycle - Slave Port to Data Channel
•   1 Cycle - First Word transfer through Crossbar
•   1 Cycle - Channel to Master Port transfer
•   3 Cycle - Internal transfer for remaining words (Hidden Transfer)

                        2. Last Word will be completed internally after 6 cycle delay

•  1 Cycle - Slave Port to Data Channel
•  4 Cycle  - 4 data burst
•  1 Cycle -  Channel to Master Port transfer
  

                 Pipeline Flow:

                 

        Write request operation:

    1.     Master will request the bus arbiter

    2.     Arbiter grant the master based on availability and the Arbitration algorithm.

    3.     Data will be transferred from master to slave

    4.     Slave will send the first word of the data to slave device.

    5.     Slave can handle other transactions during the data is writing to the slave device. This can be achieved by Read_Threshold and Write_Threshold parameter.

    6.     Acknowledgment for the write is completed can be obtained from slave device (based on the A_Task_Flag field). This response will be forwarded to the master.

    7.     Master will sends out the ACK immediately.

Standard internal delay - Write

Forward Transfer:

               N +3 Cycle - N data burst (eg: 8 byte Burst and 32 byte request - 4 cycle data transfer) , 1 Cycle for master port to channel, 1 Cycle for  arbitration and 1 Cycle for channel to slave port

1. Output will be issued after 4 cycle as first word out.

• 1 Cycle    -  Master port to Address Channel
• 1 Cycle    -  Arbitration Grant
• 1 Cycle    -  Address and Data channel transfer
• 1 Cycle    - 1st Data word to output port
• N-1 cycle - Remaining Data word transfer (Hidden Transfer)

      2. Last Word will be completed internally after 7 cycle delay

•  1 Cycle    -  Master port to Address Channel
•  1 Cycle    -  Arbitration Grant
•  1 Cycle    -  Address and Data channel transfer
•  4 Cycle    -  4th Data word
  

Pipeline Flow:

     

      Backward Transfer: (Return response, 3 Cycle)

•  1 Cycle - Slave Port to channel transfer
•  1 Cycle  - Response channel transfer
•  1 Cycle -  Channel to Master Port transfer
  

       Pipeline Flow:

      

For flows like Sequential Read, Sequential Write, Read followed by Write and Write followed by Read, delaywill be vary

This includes buffering delay due to Slave busy. Arbitration to each request will be based on the availability of the slave and the arbitration algorithm

selected.

If the slave is busy then the packet will be waiting in the master. Once the request is granted,  the sequence of delays will be same as the standard delays.

List of models

              VS_AR/Demo/Bus_Std/AHB_AXI:

                             1. AXI_Bus.xml

                             2. AXI_Bus123.xml

                             3. AHB_AXI_Bus.xml

                             4. Multi_AXI_HWDRAM.xml

                             5. multi_inp_demo.xml

                             6. Simple_AXI_Bus.xml

                             7. Simple_AXI_Bus_FixedArbiter_A_Priority_Variable.xml

                             8. Simple_AXI_FixedArbiter_Modified_Diff_A_Priority.xml

                             9. Simple_AXI_Model_V8.xml

                             10. Simple_AXI_RR.xml

                             11. Single_AXI.xml

                                         12. Three_level_Bus_models.xml

Block Parameters

General Parameters

Parameter	Explanation	Example
Architecture_Name	It refers the Architecture_Setup name. Please keep both the names as same. It can be linked to top level parameter. Giving different names other than Architecture_Setup block's name will throw an error.	"Architecture_1"
Bus_Name	Unique name for this Bus. Different from all architecture blocks and global model memories.	"AXI_Top"
Key Parameters for Configuration
AXI_Speed_Mhz	Speed of the bus in terms of Megahertz. It can be linked to the top level parameter.	1000.0
AXI_Cycle_Time	Cycle time for all the Master ports are defined using the AXI_Speed_Mhz parameter.	1.0E-06 / AXI_Speed_Mhz
Bus_Width	It defines the maximum bytes that the AXI bus can transfer within a single cycle. User can define the value based on the requirement. It can be linked to top level parameter.	8
Read_Threshold	It defines the Maximum outstanding Read Request that the AXI bus can transfer from any master to one slave. If the Read request for a particular slave reached the threshold then the bus cannot send any data to that slave.	2
Write_Threshold	It defines the Maximum outstanding Write Request that the AXI bus can transfer from any master to one slave. If the Write request for a particular slave reached the threshold then the bus cannot send any data to that slave.	2
Master_Request_Threshold	It defines Threshold value for each master. If the Request (Read/Write) coming to the specific master has reached the threshold then the AXI will drop the incoming request. User can increase or decrease the threshold based on the Read buffer requirement.	{2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2}
Number_Masters	It defines the number of actively configured masters that are connected to the AXI.	16
Number_Slaves	It defines the number of actively configured slaves that are connected to the AXI.	8
Threshold_Trans_T_Bytes_F	It defines the Mode of Transaction in the AXI. If it set to true, then the bus will perform the transfer in terms of request threshold eg: Threshold values can be 2,3,4,.etc. If the parameter is set to false, then the AXI will perform the transfer based on the byte threshold.     Following are the example for byte based threshold 1.     Read_Threashold = 32    2.     Write_Threashold = 32 3.     Master_Request_Threashold = {32,16,32,32,32,8,64,32,2,2,2,2,2,2,2,2}	true
Arbiter_FIX_1_RR_2_CUSTOM_3	It defines the Scheduling algorithm in the AXI between masters to slaves.. if value 1 is given then Fixed_Priority algorith will be used. if value 2 is given, then Round Robin algorithm will be used. if Value 3 is given then users custom algorithm can be used, which can be implemented by right click on the block -> open block -> right click on the script under "Custom Request Arbiter" ->open block -> (user can impklement thier algorithm here). OR the Custom_File with algorithm can be used.	1
Slave_Speeds_Mhz	It defines the speed of the each slave in this AXI bus. This can be used to define the clock cycle and internal timing of each slave. Each index is for the slave ports in order.	{AXI_Speed_Mhz, AXI_Speed_Mhz, AXI_Speed_Mhz, AXI_Speed_Mhz,AXI_Speed_Mhz, AXI_Speed_Mhz, AXI_Speed_Mhz, AXI_Speed_Mhz}
Extra_Cycles_for_RdReq_ WrReq_RdData_WrData	It defines the overhead delays for each operation in the AXI bus. Each points to the corresponding operations specified in the parameter. User can put number of delay cycles based on the requirement. Array with 0 values will not provide any delay to that operation.	{0, 0, 0, 0}
Devices_Attached_to_Slave_by_Port	This is a array of arrays. Each array contains the list of devices that are accessed via a Slave Port. Index 0 of the array is for Slave Port 1, Index 1 is for Slave Port 2 and so on. There must be one array with one value for each Slave. The names are strings. AXI bus uses this information for routing the input request reaching the master. If user forgot to give the slave name then the bus will throw an error saying "destination is not found".	{{"Cache"}, {"DRAM"}, {"DRAM3"}, {"Device_4"}}
Master_Throttle_Enable	It defines the option for choosing flow control between the AXI master and the master device. If the user set true for specific master then there will be a flow control maintained. For detail information check the Master throttle enable example	{false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false}
Slave_Throttle_Enable	It defines the option for choosing flow control between the AXI slave and the slave device. Similar to Master_Throttle_Enable	{false, false, false, false, false, false, false, false}
DEBUG	This will help the user to see how the transaction is are executing on the AXI bus. User can perform debugging by setting the value true and connecting a text display to AXI's "stats_out" port.	false
Custom_Arbiter_File	Name of File containing custom arbitration in Virtual Machine script.	"Custom_Script.txt"
Custom_Arbiter_Path	Path to the custom arbitration script file.	"C:/VisualSim/Scripts for Windows  /VisualSim/Scripts for UNIX"
Fixed_Priority_Array	This is a 2D array. Each index of this Parameter corresponds to each slave. Since we can have a max of 16 masters on VisualSim AXI Bus Module, each of the array at slave index will also have a max length of 16. This 16 length array has to be configured in such a way that the max priority master port number should be at the starting of the array. Example: If one of the  array index is defined as  {5,7,3,4,1,6,2,8,9,10,11,12,13,14,15,16}, then this means that priority order for masters to send to that slave will be Master 5, then Master 7, then Master 3, then Master 4, then Master 1 etc.....	{{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}, {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16},......., {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}}
Ports_to_Plot	This will enable the AXI to generate ploting information for specific master and slave. User can observe the Timing diagram for the internal operations. This can be achieved by connecting a DS_xTime_yData_Plotter to the plot out port in AXI.	{0,0}

   

Master Throttle Enable

      

        Only one outstanding request will reach the master, when the master becomes free after transferring the data to slave then the next request will reach the master.

        To establish the flow control, user wants to add a field names "Event_Name". The Master device should send first request and wait for the virtual event from the AXI master.

        User can implement throttle control to specific master or slave and keep the remaining connections as same.

            Following are the implementation example(in script block) for Independent Flow control.

                       

                                    Event = "Master_Flow_Ctrl"

                                    First_Word = true

            LABEL:BEGIN

                                    -----

                                    -----

                                    port_token.Event_Name = Event

                                    -----

                                    -----

                                    QUEUE("Flow_Control_Queue",port_token,0,put)

                                    if(!First_Word){

                                                EXIT

                                    }

                                    First_Word = false

            LABEL: Loop

                                    Data_ Structure =QUEUE("Flow_Control_Queue",pop)

                                    SEND(output,Data_Structure)

                                    TIMEQ(Event)

                                    len = QUEUE("Flow_Control_Queue",length)

                                    if(len== 0)

                                                First_word = true

                                                EXIT

                                    }

                                    GTO(Loop)

Expected Data Structure and Ports

Data Structure Field	Value (Data Type)	Explanation
A_Bytes                              (necessary)	100	This is the total bytes to be transfered for this transaction.  All bursts of this transaction will have this value.
A_Command                       (necessary)	"Read" or "Write"	This determines the operation.
A_Priority                           (necessary)	2	This is priority of the transaction.  The transactions waiting for service are reordered based on this value.
A_Source                            (necessary)	"Processor"	This is unique name of the Source.  When the transaction returns from the Destination, the Source and Destination names are flipped.  So, the Source becomes the Destination and Destination becomes Source.
A_Destination                     (necessary)	“DRAM”	Final Destination

Port Name	Type	Explanation
input, input2, input3, input4, input5, input6, input7, input8, input9, input10, input11, input12, input13, input15, input16	Multiport or input/output. Connection must be made in order- first input to this port and then output away from this port.	Each port is for a Slave. Transaction Data Structure of type Processor_DS.
output, output2, output3, output4, output5, output6, output7, output8	Multiport or input/output. Connection must be made in order- first input to this port and then output away from this port.	Each port is for a Slave. Transaction Data Structure of type Processor_DS.

Input request combination

Description	A_Command	A_Bytes	A_Bytes_Remaining (optional)	A_Bytes_Sent (optional)
100 Byte Read at Slave. Bus Width = 4	Read	100	96	4
100 Byte Read Return at Master	Write	100	0	100
100 Byte Write at Slave	Write	100	0	100

How to connect

     The following image shows the right way to connect the AXI in the Architecture. Source like Processor and DMA can be connected directly to the AXI master       port, but if the Master device is designed with simple library block(as logic) then user has to include Device Interface block to provide the interface with master.     Similarly for Slave. Slave like RAM, Integrated cache and display can be  connected directly to the slave port.

The Bridge block plays important role in connecting the different buses in the architecture. Two buses can not be connected directly, user has to use Bridge block.

   

Statistics and analysis

   The statistics are placed directly in the Architecture_Setup block and will be available at the end of the simulation from the statistics output of that block.

Statistics

DISPLAY AT TIME          ------ Part 1 ------ {DS_NAME                      = "AXI_Level_2_Detected_OverFlows"} DISPLAY AT TIME          ------ Part 2 ------ {AXI_Level_2_Master_1_Wr_Data_Bytes      = 256, AXI_Level_2_Master_1_Wr_Data_MBps      = 51.2, AXI_Level_2_Master_2_Rd_Data_Bytes      = 64, AXI_Level_2_Master_2_Rd_Data_MBps      = 12.8, AXI_Level_2_Slave_2_Wr_Data_Bytes      = 256, AXI_Level_2_Slave_2_Wr_Data_MBps      = 51.2, AXI_Level_2_Slave_3_Rd_Data_Bytes      = 64, AXI_Level_2_Slave_3_Rd_Data_MBps      = 12.8, DS_NAME                       = "AXI_Level_2_Rd_Wr_MBps"} DISPLAY AT TIME          ------ Part 3 ------ {AXI_Level_2_Slave_3_Rd_Threshold_Usage      = 1.0, AXI_Level_2_Slave_Transactions      = 2, DS_NAME                       = "AXI_Level_2_Slave_3_Rd_Threshold"} DISPLAY AT TIME          ------ Part 4  ------ {AXI_Level_2_Slave_2_Wr_Threshold_Usage      = 1.0, AXI_Level_2_Slave_Transactions      = 8, DS_NAME                       = "AXI_Level_2_Slave_2_Wr_Threshold"} DISPLAY AT TIME          ------ Part 5 ------ {BLOCK                        = "AHB_AXI_Bus.AMBA_AXI4.Master_1.Master", DS_NAME                       = "Master_Transactions_1", Number_Entered                = 8, Number_Exited                 = 8, Occupancy_Max                 = 1.0, Occupancy_Mean                = 0.5, Occupancy_Min                 = 0.0, Occupancy_StDev               = 0.5} DISPLAY AT TIME          ------ Part 6 ------ {BLOCK                        = "AHB_AXI_Bus.AMBA_AXI4.Slave_2.AXI_Level_2_Slave_Out_Queue", DS_NAME                       = "Slave_Transactions_2", Number_Entered                = 8, Number_Exited                 = 8, Occupancy_Max                 = 1.0, Occupancy_Mean                = 0.5, Occupancy_Min                 = 0.0, Occupancy_StDev               = 0.5, Total_Delay_Max               = 3.01E-8, Total_Delay_Mean              = 2.6755555555556E-8, Total_Delay_Min               = 0.0, Total_Delay_StDev             = 9.4595173838734E-9}

Part 1 Shows the number of transactions that were dropped for each Master. Part 2 First section shows the number of read/write data in bytes for each Master. Second shows the MBps for read/write for each Master. Last section shows the number of read/write data in bytes for each Slave. Last shows the MBps for read/write for each Slave. Part 3 Shows the read threshold usage and the Threshold value set in the parameters of the block.  This is shown separately for each Slave. Part 4 Shows the write threshold usage and the Threshold value set in the parameters of the block.  This is shown separately for each Slave. Part 5- One for each master Shows the statistics for each Master Part 6- One for each Slave Shows the statistics for each Slave

Debug

DISPLAY AT TIME          ------ 19.00 ns ------ AXI_Level_2 Master_2 Rd_Request  Src: uProcessor  Des: DDRDRAM3  Size: 32 Master_2 Increment   Rd_Requests: 1  Wr_Requests: 0  Master_Req_Threshold: 2 DISPLAY AT TIME          ------ 30.00 ns ------ AXI_Level_2 Arbiter  Slave_3  Increment Rd_Requests: 1  Rd_Threshold: 8 Arbiter  Master_2 Decrement Rd_Requests: 0 DISPLAY AT TIME          ------ 30.00 ns ------ AXI_Level_2 Slave_3  Rd_Transaction  Src: uProcessor  Des: DDRDRAM3  Size: 32 DISPLAY AT TIME          ------ 60.00 ns ------ AXI_Level_2 Master_2 Rd_Transaction  Src: DDRDRAM3  Des: uProcessor  Size: 32 DISPLAY AT TIME          ------ 78.10 ns ------ AXI_Level_2 Master_2 Slave_3  Decrement Rd_Requests: 0 Rd_Threshold: 8 DISPLAY AT TIME          ------ 2.91100 us ------ AXI_Level_2 Master_1 Wr_Request  Src: DMA  Des: DDRDRAM2  Size: 32 Master_1 Increment   Wr_Requests: 1  Rd_Requests: 0  Master_Req_Threshold: 4 DISPLAY AT TIME          ------ 2.92800 us ------ AXI_Level_2 Slave_2  Wr_Transaction  Src: DMA  Des: DDRDRAM2  Size: 32 DISPLAY AT TIME          ------ 2.94600 us ------ AXI_Level_2 Master_1 Wr_Transaction  Src: DDRDRAM2  Des: DMA  Size: 32 DISPLAY AT TIME          ------ 2.94610 us ------ AXI_Level_2 Arbiter  Slave_2  Increment Wr_Requests: 1  Wr_Threshold: 8 Arbiter  Master_1 Decrement Wr_Requests: 0 DISPLAY AT TIME          ------ 2.94610 us ------ AXI_Level_2 Slave_2  Decrement Wr_Requests: 0  Wr_Threshold: 8

19;00ns Displays the current time Diaplays the name of the AXI Bus Master Port Number- Master_2 Request Type- Rd and Wr Source- Src: uProcessor Destination- Des: DDRDRAM3 Request Size- 32 Increment states whether it is incrementing the Threshold counting or Decrement which is decreasing the count. Provides the number of outstanding Requests for this Master and the Destination Slave. 30.00 ns Increment the counter for the Read/Write request and the read threshold. Decrement the counter for the Request

Enabling plots

     The Plots can be enables by the following methods

1. Single master and Single Slave analysis plot
1. Configuration parameter "Ports_to_Plot" - enter master number and slave number as follows: {2,3}  ( Transaction between second master and third slave will be plotted).
2. Connect DS_xTime_yData_Plotter(Full Library -> Result -> Plotter-> DS_xTime_yData_Plotter) to the bottom of the AXI bus port "plot_out" and configure the plotter as follows:
1. Field_Trace_Name    -     Plot_Name
2. Field_Y_Value           -     Plot_Value
   
3. Field_Color                -     Plot_Color
   
4. Field_Offset               -     Plot_Offset
3. 
   
2. Timing Diagram:
1. Add  the Timing Diagram block the model (Hardware Setup -> TimingDiagram)
2. Configure the block: modify the AXI bus name.
3. 
   

Enabling power

        To enable power analysis, please configure the AXI bus parameter "Power_Managera_Name" with the power block name (eg: "Manager_1")
        Following that, configure the power table with the following device names:
             Bus_Name                                                        eg: AXI_bus
             Bus_Name   +  "Rd_Address_Channel"               eg: AXI_bus_Rd_Address_Channel
             Bus_Name   +  "Wr_Address_Channel"               eg: AXI_bus_Wr_Address_Channel
             "STR_"+Bus_Name + "_Rd_Data_Channel"         eg: STR_AXI_bus_Rd_Data_Channel
             "STR_" + Bus_Name + "_Wr_Data_Channel"       eg: STR_AXI_bus_Wr_Data_Channel

1.              Bus_Name + "_Master_RD_Wr"                         eg: AXI_bus_Master_RD_Wr

         The power values for each state can be configured according to the vendor data sheet or custom requirement.

Errors and solutions

       

        1. VisualSim.kernel.util.InternalErrorException: Cannot create receivers

            in .<Unnamed Object>.AMBA_AXI

            Because:

            Cannot create a receiver without an executive director.

                  Make sure the Digital simulator is avaialble in the model.
   

        2. Error                :  Issue with RegEx execution

            Exception        :  VisualSim.kernel.util.IllegalActionException: Error invoking function public static java.lang.String    

            VisualSim.data.expr.UtilityFunctions.throwMyException(java.lang.String) throws VisualSim.kernel.util.IllegalActionException

            Because:

            User RegEx Exception:

            AXI_Top_Master_2 did not find Slave named DRAM

            on port number: 2 from Source: Processor2

            Check Device_Attached_to_Slave_N parmeters

           Make sure the "Device_Attached_to_Slave_by_Port" parameter contains the connected slave names.

       Make sure the destination name is correct.

   
    

---

Created with the Personal Edition of HelpNDoc: Free Qt Help documentation generator