You are on page 1of 39

• It Involves the Repetition or Rollback of

instructions , Segment of programs or entire


programs immediately after a fault is detected.
• Rollback operation requires that a program
restarts processing from the last check point ,
where all the information relevant to the
successful execution of the program beyond
the check point is stored .
• The Selection of correct number of checkpoints is important .

• If too many check points are used , the computation time increases
since the normal flow of the program is stopped at each check
point in order to save state information.

• The Recovery time will increase even if the check points are too far
apart since the program will have to be re-executed from the last
check point .

• The optimum check point interval is to be chosen to get the


optimum recovery time.
• If the fault is temporary , Rolling back the
program to a check point should allow
successful recovery .

• If the fault is permanent , the fault detection


mechanism will be activated and an alternative
recovery method should be attempted.
Transient fault detection
Distinguishing between Transient and
Permanent Faults
Permanent fault detection
Permanent fault detection Scheme
Coding Schemes
Alternating Logic
Alternating Logic Time Redundancy
DUALITY
Computing Dual Function(1)
Computing Dual Function(2)
EXAMPLE
Alternating Logic

 The Key Advantage of Alternating Logic is that any


combinational circuit with n input variables can be transformed
into a self dual circuit with no more than n+1 variables

 The dual of an n variable function f is given by


_ _ _ _
fd = f( X1 , X2 , , …….. , Xn ) Here Fsd is the
transformed
 The function Fsd is Given by
Self dual circuit of F
_
f = x f + x n+1 f d
sd n+1
RECOMPUTING WITH SHITED OPERANDS
(RESO)
RECOMPUTING WITH SHIFTED OPERANDS
(RESO)
RECOMPUTING WITH SHIFTED OPERANDS
(RESO)
RECOMPUTING WITH SHIFTED OPERANDS
(RESO)

Error
detection
mechanism
of ALU using
RESO
Technique
RECOMPUTING WITH SHIFTED OPERANDS
(RESO)

• The additional hardware required for this


technique is the three shifters , the storage
register to hold the results of the first
computation and the comparator also.
• Also, the ALU must be extended by 2 bits to
allow a two bit , arthematic shift to be
performed.
• Due to this extra bits required , we will go for
RESWO approach
RECOMPUTING WITH SWAPPED OPERANDS
(RESWO)
RECOMPUTING WITH SWAPPED OPERANDS
(RESWO)

• During the first computation, the operands are


manipulated in their standard form .
• During the second computation , the upper
and lower halves of the operands are swapped
such that a faulty bit slice operates on either
the lower or upper half of the operands during
the first computation and the upper or lower
half of the operands during the second
computation .
ERROR DETECTION FOR ALU USING
(RESWO)
At t0,
At t0+
C1AC
C1Cout
C2Cout
C2AC
RECOMPUTING WITH
DUPLICATION AND
COMPARISION
(REDWC)
RECOMPUTING WITH DUPLICATION AND
COMPARISION

• This method takes advantage of both Time


Redundancy and Hardware Redundancy.
• In both RESO and RESWO methods error
detection is performed with the help of time
redundancy
• But In this method,error detection accomplished
resembles that of duplication with comparision .
• Then Time redundancy is then used to complete
the calculation and obtain the final result
RECOMPUTING WITH DUPLICATION AND
COMPARISION
RECOMPUTING WITH DUPLICATION AND
COMPARISION

• For example, a 32 bit addition operation can


be realized by using each of two 16 bit adders
twice .
• During each calculation, error detection is
accomplished by comparing the results from
using the two adders .
RECOMPUTING WITH DUPLICATION
AND COMPARISION

32 bit
swapper
consists of
32
2 -1 MUXs
TIME REDUNDANCY FOR ERROR
CORRECTION

• Time redundancy can provide error correction


if the computations are repeated for 3 or
more times.

• Consider the logical and operation that was


illustrated in RESO method .
TIME REDUNDANCY FOR ERROR
CORRECTION
OPERATION 1:
without shifting

OPERATION 2:
1BIT LEFT SHIFT
W.R.T ORIGINAL DATA
TIME REDUNDANCY FOR ERROR
CORRECTION
OPERATION 3:
3BITS LEFT SHIFT
W.R.T. ORIGINAL DATA

By performing a majority vote on the 3 results


Error due to faulty bit slice can be corrected.
TIME REDUNDANCY FOR ERROR
CORRECTION

• Each of the three operations used operands


that were displaced from each other by atleast
one bit position , a different bit in each result
will be affected by the fault of bit slice .
• The results due to faulty bit slice can corrected
by performing a majority vote on the three
results obtained for each bit position
CONTRIBUTION
• D.V.R.Reddy
• Ramesh
• Satesh
• jitender
• Gufran
• Dubey
• Jignesh
• thakur
References
• P.k.lala
• Barry W. Johnson

You might also like