Professional Documents
Culture Documents
DHAKAD
Asst Professor
Asst.
Electronics & Communication Engg. Deptt.
(M-Tech-
Tech- Embedded System & VLSI Design)
Email--Id
Email Id--soveran_vlsi@rediffmail.com
Contact-- 09685396020 ,0751-
Contact ,0751-2387520(O)
RAM ROM
NITM , GWALIOR
bitline conditioning
wordlines
row decoderr bitlines
memory cells:
2n-k rows x
2m+k columns
n-k
k column
circuitry
n column
decoder
2m bits
NITM , GWALIOR
¾The memory cells in a SRAM are organized in rows and
columns.
Memory Cell = 2n-k Row Ҳ 2m+k Columns
¾I the
¾In th write
it operation
ti , the
th Word
W d line
li isi in
i active
ti state,
t t in
i
that cause each data bit to be stored in a selected cell in the
associated column.
NITM , GWALIOR
NITM , GWALIOR
The 6T SRAM cell has a differential read operation. This
means that both the stored value and its inverse are used in
evaluation to determine the stored value. Before the onset of
a read
d operation,
ti th Word
the W d line
li i held
is h ld low
l and
d the
th two
t
bitlines connected to the cell through transistors M5 and M6
are precharged high . Since the gates of M5 and M6 are held
low, these access transistors are off and the cross-coupled
latch is isolated from the bitlines.
NITM , GWALIOR
A_b bit_b
1.5
1.0
word bit
0.5
A
00
0.0
0 100 200 300 400 500 600
time (ps)
NITM , GWALIOR
Read Operation:-
¾If the value is a 1, stored at Q.
¾The read cycle is started by pre charging both the bit
lines to a logical 1, then asserting the word line WL,
enabling both the access transistors.
¾Th second
¾The d step
t occurs when
h the
th values
l stored
t d in
i Q and d
Q are transferred to the bit lines by leaving BL at its pre
charged
g value and discharging
g g BL through
g M1 and M5 to a
logical 0.
¾ On the BL side, the transistors M4 and M6 pull the bit line
toward VDD, a logical 1.1
¾If the content of the memory was a 0, the opposite would
happen and BL would be pulled toward 1 and BL toward 0.
NITM , GWALIOR
Ab
A_b
1.5 A
bit b
bit_b
1.0
0.5
word
0.0
0 100 200 300 400 500 600 700
time (ps)
NITM , GWALIOR
Write Operation:-
¾The start of a write cycle begins by applying the value to be
written
itt to
t the
th bit lines.
li
¾If we wish to write a 0, we would apply a 0 to the bit lines, i.e.
setting BL to 1 and BL to 0.
This is similar to applying a reset pulse to a SR-latch , which
causes the flip flop to change state.
¾A 1 is written by inverting the values of the bit lines. WL is
then asserted and the value that is to be stored is latched in.
¾Note that the reason this works is that the bit line input-
d i
drivers are designed
d i d to
t be
b much h stronger
t th
than th relatively
the l ti l
weak transistors in the cell itself, so that they can easily
override the previous state of the cross-coupled inverters.
NITM , GWALIOR
¾It is practically embedded in every application that
requires electronic use interface such as digital cameras,
cameras
cell phones, etc.
¾I t l CPU caches
¾Internal h , hard
h d disk
di k buffers,
b ff router
t b ff
buffers,
LCD screens and printers also normally employ static RAM
to hold the image displayed .
¾Small SRAM buffers are also found in CDROM and CDRW
drives; usually 256 kB.
NITM , GWALIOR
¾A Sense
S Amplifier
A lifi is
i an essential
i l
circuit in designing memory chips.
¾ The resulting signal,
signal in the event of a
Read operation, has a much lower
voltage swing. To compensate for that
swing a sense amplifier is used to
amplify voltage coming off Bit Line.
¾The voltage
¾Th lt coming
i outt off the
th sense
amplifier typically has a fully swing (0 -
2.5V)) voltage.
g
¾Sense amplifier also helps reduce the
delay times and power dissipation in
the overall SRAM chip.
NITM , GWALIOR
There are many versions of sense
amplifiers used in memory chips :-
¾The one that we will use in our
design is called a Cross-coupled
Sense Amplifier
p demonstrated on a
block diagram below.
¾During a read sequence, Bit Line
and Bit Line are directed into X and
X inputs. Once SE has been set to
logic 1, the amplifier turns on, and
gives Y and Y as its outputs.
NITM , GWALIOR
¾Cache memory is basically a cost-effective method of
improving system performance.
¾Cache memory is a relatively small, high-speed memory
that stores the most recently used instructions or data.
¾Cache memory
e o can
ca also
al o use
e dynamic
d a ic RAM (DRAM).
(DRAM)
¾Cache memory stored information to the microprocessor
much faster than if only high-capacity DRAM is used.
used
¾Cache memory used to store data or instructions likely to
be used soon by the CPU. Its purpose is to speed up
operation by bridging the performance gap between the
CPU and the main memory.
NITM , GWALIOR
Two types of cache level are used in cache memory:
memory:-
L1 :- It usually integrated into the processor chip and has a
veryy limited storage
g capacity.
p y
¾It gives an extremely short access time, and therefore
provides the highest performance
¾This cache usually runs at the same clock frequency as
the CPU
L2 :-It is separate memory chip or set of chips external to
the processor and usually has a larger storage capacity
th
than L1 cache.
h
¾This is connected to CPU through an internal bus
Some higher-level caches (L3. L4, .), but L1 and L2 are the
most common.
NITM , GWALIOR
There are some very important requirements for a memory
when it is to be embedded as on-chip cache:
¾It has to be reliable and stable. This is of course true for
all memories, but is specially important for cache due to
the more extreme performance requirements and area
limitations.
¾Memory provide high performance gap between main
memory and the CPU.
¾Another important
p requirement
q is low power
p
consumption.
NITM , GWALIOR
Low power design is important from three different
reasons-
Technology driven forces
Minimum feature size,
Minimize parasitic capacitance
Higher operating speed
Design driven forces
Power consumption
p in digital
g circuits
Power consumption in analog circuits
Market driven forces
The growing demand for long life portable
equipment.
NITM , GWALIOR
There are various types of applications of low power-
¾Battery-powered portable systems, for example laptops,
CDs, ,DVDs
¾Electronic packet communication products such as;
cordless and cellular telephones, PDAs (Personal Digital
Assistants), pagers.
b GH processors for
¾Sub-GHz
¾S f high-performance
hi h f workstations
k t ti
and computers.
¾ Other applications such as WLANs (Wireless Local Area
Network) and electronic goals (calculators, hearing aids,
watches, etc.).
NITM , GWALIOR
¾The supply voltage must be reduced.
¾The threshold voltage (VT) must be reduced proportionally
with
ith the
th supply
l voltage
lt so th
thatt a sufficient
ffi i t gate
t overdrive
d i isi
maintained.
¾Reduction in the threshold voltage causes increase in
leakage current.
NITM , GWALIOR
¾During an idle phase,
phase the word lines are deselected (WL =
‘0’) and the bit lines are precharged (BL = ‘1’ and BL = ‘1’).
¾The memory cell data,data either transistors N4,
N4 P1,
P1 N2 (for bit
= ‘1’) or N3, P2, N1 (for bit = ‘0’) will be leaking .
¾The transistors in the off state in bold for bit = ‘0’. In this
case N3,N1 and P2 are off and will be leaking. The leakage
current in the memory cell would be as shown in equation:
ImemcellIdle = IDsub(N1) + IDsub(N3) + IDsub(P2)
where, IDsub is the sub threshold leakage current of
the MOSFET , which is given by the equation :-
IDsub = Is e VGS/(nKT)/q [1-VDS/eKT/q ]
where, Is and n are imperial parameters with n ≥ 1.
NITM , GWALIOR
¾ The sub threshold leakage in the whole memory core
is given by equation .
ImemcoreIdle = Nrows. Cools . ImemcellIdle
where, Nrows and Ncols are the number of rows and
columns respectively in the memory core.
¾ Thus to reduce the leakage of a memory cell we have
t concentrate
to t t on two
t componentst off leakage
l k :-
1. one is the leakage inside the cell .
2. Second is leakage to bit lines.
NITM , GWALIOR
Techniques
T h i are used
d to reduce
d the
h leakage
l k current is:-
i
Dual VT :-
¾This technique requires no additional control circuitry
and can substantially reduce the leakage current when
compared to low VT devices.
devices
¾No data are discarded and no additional caches
misses are incurred. However , high
high- transistors have
slower switching speed and lower current drive.
ABC-MTCMOS :-
¾It can reduce the leakage current significantly using a
simple circuit while in the sleep mode.
NITM , GWALIOR
¾I order
¾In d to reduce
d undesirable
d i bl leakage
l k current in
i the
h
sleep mode, the back gate bias is automatically
controlled to increase the threshold voltage.
g
NITM , GWALIOR
DVS (Dynamic Voltage Scaling):-
In this method to reduce the leakage power of SRAM cells,
in active mode.
¾When cells are not intended to be accessed for a time
period,
e iod they
the are
a e placed
laced in
i a sleep
lee mode.
ode
¾In a sleep mode the leakage power is significantly
reduced due to the decreases in both leakage current
and supply voltage.
NITM , GWALIOR
We have to used Two PMOS transistor P1,P2, to control the
supply voltage of the memory cell based on the
operating.
p g
1. Active Mode 2. Sleep Mode
¾ If cell is active mode , P1 supplies a standard supply
voltage, and P2 supplies a standby voltage.
¾ If cell is Sleep mode, P1 and P2 are controlled by
complementary supply voltage control signals.
NITM , GWALIOR
Embedded
E b dd d memory-
¾Easy to implement in generic CMOS process.
¾Easy to design as logic circuit.
¾Easy to test by finite-state machine.
Compliable design-
ed ce
¾Fixed cell size
s e to a
allow
ow us ded
dedicating
cat g in peripheral
pe p e a c circuit
cu t
design
¾Synchronous interface since 0.35µm
0 35µm generation simplifies
the design
¾A larger number of instances required
NITM , GWALIOR
NITM , GWALIOR
¾In the 5T SRAM cell differs fundamentally from the cell
used in 2PMOS & 3 NMOS Transistors.
¾This requires
q an additional metal wire and also destabilizes
all cells on the bit line during write.
NITM , GWALIOR
Read Operation-The operation scheme when reading a 5T
cell is very similar to the 6T SRAM.
¾Before the onset of a read operation, the word line is held
low and the bitline is precharged.
¾The bitline is not precharged to VCC, So another value is
carefully chosen according to stability and performance
requirements.
¾If reading
di a ’0’,
’0’ BL will
ill now be
b pulled
ll d down
d th
through
h the
th
transistor combination. If instead a ’1’ is to be read, the
situation is slightly different from the 6T case.
NITM , GWALIOR
Write Operation-Writing in the 5T SRAM cell differs from
the 6T cell mainly by the fact that it is done from only one
bitline.
¾In the 5T cell the value to be written is held on the bitline,
and the word line is asserted.
¾The 6T cell was sized so that a ’1’
1 could not be written by
a high voltage on the bitline, the 5T cell has to be sized
differently.
NITM , GWALIOR
¾The difference between the 5T SRAM and the 6T SRAM is
how the sensing of the stored value is done.
¾The 6T cell has two bit lines and the stored value is sensed
differentially.
NITM , GWALIOR
NITM , GWALIOR
NITM , GWALIOR
NITM , GWALIOR
Table 5.1: Leakage power and performance of 6T cell
NITM , GWALIOR
Table 5.2: Comparison of leakage power reduction
techniques
Conventional 2.030 -
G t d VDD
Gated-VDD 0 033
0.033 98 3
98.3
NITM , GWALIOR
Table 5.3:
5 3: Leakage power and performance of 5T cell
Metrics Standard 6T cell
Read time (WL high up to 100mV difference in bit lines) 365ps
6T 5T
Conventional 2.030 1.790 11.8
DVS 0.230
. 0.170
. 7 26.0
Gated-VDD 0.033 0.029 12.1
NITM , GWALIOR
NITM , GWALIOR
NITM , GWALIOR
NITM , GWALIOR
NITM , GWALIOR
NITM , GWALIOR
NITM , GWALIOR
¾ Various circuit level techniques have been applied to 6T
and designed 5T SRAM cell for leakage power reduction
and compared. Out of all the techniques discussed DVS has
f
found
d to
t be
b the
th best
b t as it reduces
d l k
leakage comparable
bl to
t
Gated VDD as well as retain the cell information.
NITM , GWALIOR
¾ In this thesis various circuit level leakage g p power
reduction techniques have been analyzed with 6T and 5T
SRAM cell at 180nm technology. A large reduction in
leakage has been observed.
observed As memory cells being
discussed have to be used in cache memory their stability is
also very important. So stability analysis of both 6T and 5T
cells after
f applying
i leakage reduction
i techniques
i can be
analyzed.
q
¾ Device level techniques such as retrograde
g well;; Halo
doping and LDD (Light Doped Drain) implantation can be
employed for leakage reduction in individual MOSFETs
which eventually will reduce in large reduction.
reduction As leakage
will be more significant beyond 100nm technology so this
work should be extended to higher technologies such as
90
90nm, 70
70nm or beyond.
b d
NITM , GWALIOR
[1] Soveran
S Singh
Si h Dhakad,
Dh k d Shyam
Sh Akashe,
Ak h “CMOS VLSI
Design” National Conference in NEE ,Gwalior , Ist Oct. 2009.
[[2]] Soveran Singh g Dhakad, Shyam
y Akashe, “Cache
Memory cell for Leakage Power ” National Conference in
NEE , Gwalior ,26th June . 2010.
NITM , GWALIOR
[1[. H. Tran, “Demonstration
Demonstration of 5T SRAM and 6T Dual
Dual-Port
Port RAM Cell Arrays,
Arrays,” Symposium on VLSI Circuits, pp.
74-79, Jun. 2005.
[2]. V. De and S. Borkar, “Technology and design challenges for low power and high performance”, International
Symposium Low Power Electronics and Design, pp.167- 170, 2000.
[3]. S. Borkar, “Technology trends and design challenges for microprocessor design”,ESSIRC, pp. 10-18, Sep. 1995.
[4]. S. Narendra, S. Borkar, V. De, D. Antoniadis, and A. Chandrakasan, “Scaling of stack effect and its application for
leakage
g reduction,” in Proc. IEEE/ACM International Symposium
y on Low Power Electronics and Design, g pp. 175–
182, Aug.2006.
[6]. J.
[6] J M.
M Rabaey,
Rabaey A.A Chandrakasan,
Chandrakasan and B.B Nikolic,
Nikolic “Digital
Digital Integrated Circuits: A Design Perspective
Perspective”, Prentice Hall
series in electronics and VLSI, Prentice Hall,second edition, 2006.
[7].I. Carlson, S. Anderson, S. Natarajan and A. Alvandpour, “ A high density, low leakage, 5T SRAM for embedded
caches”, Proceedings of the 30th Solid State Circuits Conference, ESSCIRC, pp. 220-230, December 2002.
[8].M. Mamidipaka, K. Khouri, N.Dutt, and M. Abadir, “Analytical models for leakagepower estimation of memory
array structures”, International Conference on Hardware/Software and Co-design and System Synthesis pp. 149-167,
2001.
[9].J. T. Koa and A. P. Chandrakasan, “Dual threshold voltage techniques for low-power digital circuits”, in IEEE
Journal of solid state Circuits, Vol. 37, No.10, pp.1119-1218,March 2006.
NITM , GWALIOR
[[
[[10].B.
] Amelifard,, F. Fallah,, M. Pedram,, “Reducing
g the sub-threshold and ggate-tunneling
g leakage
g of SRAM cells
using dual-vt and dual-tox assessment”, in IEEE Proceedings of Design, Automation and Test, Vol. 2, pp. 5-7, 2001.
[11].C. H. Kim and K. Roy, “A leakage tolerant cache memory for low voltage microprocessors,” Proceedings of the
1998 International Symposium on Low-Power Electronics and Design, pp. 271-280, 2000.
[12].M. Powell, S. Yang, B. Falsafi, K. Roy, and T. Vijaykumar, “Gated-VDD: A circuit technique to reduce leakage
in deep-submicron cache memories”, Proceedings IEEE/ACM International Symposium on Low Power Electronics
and Design, 2002, pp. 98–100.
[13].S. Yang, M. Powell, B. Falsafi, K. Roy, and T. Vijaykumar, “An integratedcircuit/architecture approach to
reducing leakage in deep-submicron high-performance I-caches”, in Proc. IEEE/ACM International Symposium on
High-Performance Computer Architecture, 2005, pp. 157–162.
[14].S.
[14] S Mutoh,
Mutoh T.T Douseki,
Douseki Y.
Y Matsuya,
Matsuya T.T Aoki,
Aoki S.
S Shigematsu,
Shigematsu and J.
J Yamada,
Yamada “1-V
1-V power supply high-speed digital
circuit technology with Multi-threshold-voltage CMOS,” IEEE Journal Solid-State Circuits, vol. 34, pp. 1007-1025,
Aug. 2000.
[15].K. Nii, H. Makino, Y. Tujihashi, C. Morishima, Y. Hayakawa, H. Nunogami, T.Arakawa, and H. Hamano, “A low
power SRAM using Auto-Backgate-Controlled MT-CMOS”, in Proceedings IEEE/ACM International Symposium on
Low Power Electronic Devices, 2007, pp. 296–300.
[16].H. Makino et al., “An Auto-Backgate-Controlled MTCMOS Circuit”, submitted to Symposium on VLSI
Circuits, June 2000.
NITM , GWALIOR
[17].N. S. Kim, K. Flautner, D. Blaauw and T. Mudge, “Circuit and Micro-architectural techniques for reducing
cache leakage power”, IEEE Transaction on VLSI systems Vol. 12, No. 4, pp. 168-198, Feb. 2008.
[18].K. Flautner, N. S. Kim, S. Martin, D. Blaauw, and T. Mudge, “Drowsy caches: Simple techniques for reducing
leakage power
power”, in Proc.
Proc IEEE/ACM International Symposium on Computer Architecture,
Architecture 2005,
2005 pp.
pp 142–157.
142 157
[19].X. Chen and H. Bajwa, “Energy-efficient dual-port cache architecture with improved performances,” IEE
Journal of Electronic Letters, Vol. 43, No. 1, pp. 15-18, Jan.2007.
[20].H. Tran, “Demonstration of 5T SRAM and 6T Dual-Port RAM Cell Arrays,” Symposium on VLSI Circuits, pp.
69-71, Jun. 1994.
[21].S. Kim, N. Vijaykrishnan, M. Kandemir and M. J. Irwin, “Optimizing leakage energy consumption in cache
bitlines” Journal of Design Automation for Embedded Systems,
bitlines Systems Mar.
Mar 2005.
2005
[22].A. Karandikar and K. K. Parhi, “Low power SRAM design using hierarchical divided bitline approach”, in
Proceedings International Conference on Computer Design: VLSI in computers and Processors, pp. 82-100, 2000.
[24]. E. Seevinck, F. J. List and J. Lohsttoh, “Static-Noise Margin Analysis of MOS SRAM Cells,”IEEE JSSC,
VOL. SC-22,, NOS,, pp
pp.848-854,, Oct.1998.
NITM , GWALIOR
[25]. J. Lohstroh, E. Seevinck and J. de Groot, “Worst-Case Static Noise Margin Criteria for Logic Circuits and
Their Mathematical Equivalence,”IEEE JSSC, VOL. SC-18, NO. 6, pp. 801-807, Dec.1999.
[26]. A. Alvandpour, D. Somasekhar, R. Krishnamurthy, V. De, S. Borkar and C. Svensson, “Bitline leakage
equalization for sub - lOOnm caches,
caches ” European Solid
Solid-state
state Circuits
Circuits, 2003
2003, ESSCIRC’03
ESSCIRC 03. Conference on on,
pp. 420-425, Sept. 2004.
[27] Soveran Singh Dhakad, Shyam Akashe, Sanjay Sharma “Cache Leakage : A Leakage aware cache
simulator” International Journals of Computing and Applications, vol 5 no.2 (july-Dec-2010).
[28] Soveran Singh Dhakad, Shyam Akashe, Sanjay Sharma “Dynamic Zero compression for Cache
Energy Reduction ” International Journals of power engineering ,vol 2 no.2 (july-Dec-2010).
[30] Soveran Singh Dhakad, Shyam Akashe, “Cache Memory cell for Leakage Power” National
Conference in NEE, Gwalior, 26th June. 2010.
NITM , GWALIOR
THANK YOU
AND
HAVE A NICE DAY
NITM , GWALIOR