Sony Rama et al./ Elixir Comp. Sci. & Engg. 41 (2011) 5579-5582 5579
Introduction
EMBEDDED memories occupy more than 60% chip area of
most today’s system-on-chip (SOC) designs. Keeping the
memory cores at a reasonable yield level is thus vital for SOC
products. Built-in self-repair (BISR) techniques have been
shown to improve the memory yield from 5% to 20%, such that
the net SOC yield increase can range from 2% to 10% [1].
Therefore BISR techniques are gaining popular for enhancing
the yield of embedded memories in SOCs [2]–[14]. Effective
redundancy allocation methods are needed for BISR designs
while the targeted repairable memories have 2-D (i.e., spare
rows and spare columns) or more complicated redundancy
organizations. However, the redundancy allocation for a
memory with 2-D redundancy is very difficult since it is an NP-
hard problem [15].
Many redundancy analysis algorithms for performing
redundancy allocation at automatic test equipment (ATE) have
been proposed, e.g., [15]–[18]. However, these algorithms are
not adopted to be realized in built-in circuits. Thus, built-in
redundancy- analysis (BIRA) algorithms which can cost-
effectively be realized with built-in circuits are required for
BISR schemes. In [3], Kawagoe et al. presented a
comprehensive real-time exhaustive search test and analysis
(CRESTA) scheme for bitoriented memories. The CRESTA
uses an exhaustive search approach to allocate redundancies of a
RAM such that it can achieve the optimal repair rate (i.e., the
ratio of the number of repaired memories to the number of
defective memories).The hardware cost of CRESTA increases
drastically with the number of spare elements. To reduce
hardware cost of BIRA design, some BIRA schemes use
heuristic redundancy analysis algorithms
Most of the heuristic BIRA algorithms are developed based
on a 2-D local bitmap, and some are only for bit-oriented RAMs.
However, if the user wants to extend the algorithms to repair
word-oriented memories, each bit of 2-D local bitmap needs to
be replaced by a word.
This results that the area cost for implementing the 2-D
local bitmap is large. Also, the area cost is heavily affected by
the word length. To reduce area cost of BIRA designs, effective
BIRA algorithms based on a low-cost local bitmap are needed,
especially for word-oriented memories. In [21], a BIRA
algorithm for word-oriented RAMs with 2-D redundancy using
1-D local bitmap was proposed to reduce the area cost of the
BIRA circuit. The BIRA algorithm is used for a RAM with
global spare rows and local spare columns. In comparison with
the work [21], main contributions of this paper are as follows.
1) An effective BIRA algorithm used for the RAM with local
spare rows and local spare columns is proposed.
2) A simulation flow for selecting the size of local bitmap is
proposed to optimize the repair rate and minimize the area cost
of the BIRA scheme.
3) Effective design techniques for implementing the key
components of the proposed BIRA scheme are proposed.
4) More complete simulation and analysis results are shown.
Furthermore, comparison results of different BIRA schemes are
shown.
Proposed BIRA Algorithm
Tele: +91-9963893478
E-mail addresses: sony.smile24@gmail.com,
© 2011 Elixir All rights reserved
A low-cost built-in redundancy-analysis scheme for word-oriented RAMs with
2-D redundancy
Sony Rama
1
, Kishore Kumar
1
and P.Rajesh Kumar
2
1
Department of ECE, Shri Vishnu Engineering College for Women, Bhimavaram, A.P, India
2
Dept of ECE, AUCE, Visakhapatnam.
ABSTRACT
Built-in self-repair (BISR) techniques are widely used for repairing embedded random
access memories (RAMs). One key component of a BISR module is the built-in redundancy-
analysis (BIRA) design. This paper presents an effective BIRA scheme which executes the
2-D redundancy allocation based on a 1-D local bitmap. Two BIRA algorithms for
supporting two different redundancy organizations are also proposed. Simulation results
show that the proposed BIRA scheme can provide high repair rate (i.e., the ratio of the
number of repaired memories to the number of defective memories) for the RAMs with
different fault distributions. Experimental results show that the hardware overhead of the
BIRA design is only about 2.9% for an 8192 64-bit RAM with two spare rows and two spare
columns. Also, the ratio of the BIRA analysis time to the test time is only about 0.02% if the
March-CW test is performed. Furthermore, a simulation flow is proposed to determine the
size of the 1-D local bitmap such that the BIRA algorithm can provide the best repair rate
using the smallest-size 1-D local bitmap.
© 2011 Elixir All rights reserved.
ARTICLE INFO
Article history:
Received: 9 September 2011;
Received in revised form:
15 November 2011;
Accepted: 24 November 2011;
Keywords
Built-in redundancy-analysis (BIRA),
Built-in-self-repair (BISR),
Local bitmap,
RAMs,
Redundancy.
Elixir Comp. Sci. & Engg. 41 (2011) 5579-5582
Computer Science and Engineering
Available online at www.elixirpublishers.com (Elixir International Journal)