Data Transfer Schemes

Page 1 of 6
Data between the P and I/O devices is transferred according to schemes that may fall
into one of the following categories:
1. Programmed data transfer, and
2. Direct memory access memory transfer.
Programmed data transfers are generally used when a relatively small amount of
data is transferred using relatively slow I/O devices, e.g. some A/D, D/A converters. In these
schemes, usually one byte or word of data is transferred at a time. When a large block of data is
to be transferred, DMA is used. DMA is used for transferring data from peripheral mass storage
devices like a hard disk or a high-speed line printer.
Programmed data transfer can be further classified as:
1. Synchronous transfer,
2. Asynchronous transfer (or handshaking),
3. Interrupt driven transfer.
All these schemes require both hardware and software for their implementation.
Within a C, more than one scheme can be used for interfacing various I/O devices.
Programmed Data Transfer -- Synchronous Mode
This is the simplest of all the data transfer schemes. Synchronous mode of data transfer is
used for I/O devices whose timing characteristics are precisely known, or fast enough to be
compatible in speed with the communicating MPU. Whenever data is to be obtained from the
device, or transferred to the device, the user program may issue a suitable instruction for the
device. At the end of the execution of this instruction, the transfer would have been completed.
Thus, if an output device is connected to the 8085 in the memory mapped mode, the instruction
MOV M, A may be used for transferring ACC contents to the device, assuming that the address
of the device is already stored in the HL register pair. If the device is connected in I/O mapped
mode, then the OUT instruction may be issued. The flowchart for this mode is shown in fig 1(a).
The synchronous mode can also be applied to slow I/O devices, if the timing
characteristics of these devices are precisely known. In this case the data transfer is initiated by
requesting the I/O device to get ready and hen the MPU waits for some predetermined time,
usually by generating a delay, and then executes the I/O instruction to complete the data
transfer. The flowchart is shown in fig 1(b).
R.BALA BHASKAR
Data Transfer Schemes
Page 2 of 6
Fig 1: Flowchart for the synchronous mode of data transfer:
a) Speed compatible with MPU b) I/O with known speed characteristics
Start
Execute I/O
Instructions
Stop
Start
Send get ready
signal to the I/O
device
Generate Delay
Execute I/O
instructions
Stop
Programmed Data Transfer - Asynchronous Mode
When the I/O device speed and p speed do not match, asynchronous mode may be used.
The key feature of this mode is that, the MPU initiates data transfer by requesting the device to
get ready and then goes on checking its status. The I/O instruction is executed only when the I/O
device is ready to accept or supply data. So each data transfer is preceded by a request ready
signal generated by the MPU and an acknowledgement signal issued by the I/O device. The
method is known as handshaking mode of data transfer, as it resembles the back-and-forth
movement of hands involved in handshaking, to coordinate the data transfer. The flowchart for
this mode of data transfer is shown in fig 2
The asynchronous mode is ideal for reconciling the timing differences between the MPU
and I/O devices. However, an important disadvantage is that a lot of precious MPU time is
wasted during the looping to check the I/O device status. The wastage of the MPU time may be
prohibitive or impractical in many situations.
R.BALA BHASKAR
Data Transfer Schemes
Page 3 of 6
Algorithm
begin
issue instruction to the device to get ready;
repeat
test device ready flag;
until device ready flag;
issue instructions to transfer data;
end
Fig 2: Flowchart for the asynchronous mode of data transfer
Start
Send get ready
signal to the I/O
device
Is the
device
ready?
Execute I/O
instructions
Stop
Yes
No
Programmed Data transfer -- Interrupt driven Mode
To overcome the drawbacks of the synchronous and asynchronous modes of data
transfer, another programmed data transfer called Interrupt driven mode has been designed. The
basic idea of this mode is that, the processor initiates the data by requesting the device to get
ready and then goes on executing its original program instead of wasting its time by
continuously monitoring the status of the I/O device. When the device is ready to accept or
supply data, it informs the processor through some special control line known as interrupt line.
In response to this, the processor completes the execution current instruction. Then, instead of
executing the next instruction, it saves the PC (and other registers as determined by the
R.BALA BHASKAR
Data Transfer Schemes
Page 4 of 6
architecture of the MPU) in stack and branches to a predetermined location, which is the starting
address of a subroutine called interrupt service subroutine (ISS). The ISS saves the processor
status on the stack, completes the data transfer with the I/O responsible for interrupt, restores the
processor status and then returns to the original program that the MPU was executing prior to
the interrupt. Fig 3 shows the flowchart of this mode of data transfer.
In this mode, the time needed by the I/O device to get ready after receiving the get
ready is utilized by the MPU. However, the MPU incurs some overhead in storing and restoring
the processor status to and from the stack.
Fig 3 Data transfer in interrupt driven mode
Start
Send get ready
signal to the I/O
Fetch and execute
next instruction
Is the
interrupt
line high?
Save PC in stack and jump
to ISS starting address
Start ISS
Push processor status on stack
Execute I/O instructions
Restore processor status
Return
DMA mode of data transfer
For any MPU, there are instructions for data transfer from memory or I/O device to the
MPU registers and vice versa. But there does not exist any instruction for data transfer between
R.BALA BHASKAR
Data Transfer Schemes
Page 5 of 6
memory and I/O devices directly. So, using any one of the programmed I/O modes, data transfer
between memory and I/O devices can be done in two steps using one of the MPU registers
(usually the accumulator) as via-media. This makes the programmed I/O data transfer modes
inherently slow. If a large block of data is to be transferred between memory and a fast I/O
device, the overhead incurred may be prohibitive. The direct memory access (DMA) mode has
been designed to overcome this problem. The main idea underlying this method is that, the MPU
is dissociated from the data transfer process by tristating its system bus. A direct link is
established between the memory and I/O device, and an external circuit known as DMA
controller controls data transfer. Here, the data transfer rate is only limited by the minimum
speed of either of the two devices. However, to facilitate DMA mode of data transfer, the MPU
must be equipped with the following features:
1. An input control line, through which the I/O device, via the DMA controller, requests
the MPU for DMA data transfer.
2. An output control line, through which the MPU informs the DMA grant.
3. The MPU must be able to tristate the address, data and necessary control lines
throughout the DMA data transfer duration.
Almost all the MPUs provide these features. The DMA controller must also perform the
following functions:
1. Interface the MPU buses with the I/O device.
2. Generate DMA request signal.
3. In response to the DMA grant signal from the MPU, it must control the address
bus and the control lines needed fro data transfer.
4. It must hold the information about the number of bytes to be transferred, along with
starting address of the data in memory, so that it can sequentially generate the RAM address one
by one and can withdraw the DMA request when the last byte of data transfer is over.
The sequence of events the take place is shown by the flow diagram of fig 4. Two
different types of DMA transfer are possible. In the first case, once initiated, the data transfer
process does not stop until the complete block is transferred. Therefore, this is known as the
block DMA mode. If, however, the MPU cannot be kept inactive for the long duration needed to
transfer the complete block of data, or there is significant delay between the consecutive data,
R.BALA BHASKAR
Data Transfer Schemes
Page 6 of 6
the other scheme known as the cycle stealing DMA mode can be used. In this case, one or two
bytes of data are transferred on being granted the DMA by the MPU, and then withdraw the
DMA request. After certain time the DMA controller interrupts the MPU, indicating the end of
the DMA.
Fig 4. Flowchart of the data transfer in DMA mode
Start
Initialize DMA request
Fetch and execute
the next instruction
Is DMA
request
active?
Generate DMA grant
signal and tristate buses
I/O device gets ready and
sends DMA request to the
DMA controller
DMA controller sends DMA
request to the MPU
Data transfer in DMA mode
continues till the assigned
block is transferred and then
the DMA controller
withdraws the DMA request
Signifies hardware control transfer
R.BALA BHASKAR