Design of Power Efficient Posit Multiplier Using Compressor Based Adder
Design of Power Efficient Posit Multiplier Using Compressor Based Adder
https://doi.org/10.22214/ijraset.2023.51597
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue V May 2023- Available at www.ijraset.com
Abstract: Posit number system has been used in many applications, especially the deep learning. Because of how well its non-
uniform number distribution aligns with deep learning's data distribution, deep learning's training process can be sped up. The
hardware multiplier is typically built with the widest mantissa bit-width available due to the flexibility of posit numbers' bit-
width. Such multiplier designs consume a lot of power since the mantissa bit-width is not necessarily the maximum value. This is
especially true when the mantissa bit-width is tiny. The mantissa multiplier is still built to have the widest bit-width feasible, but
it is broken into numerous smaller multipliers. At run-time, just the necessary tiny multipliers are turned on. The regime bit-
width, which can be used to determine the mantissa bit-width, controls those smaller multipliers. This design technique is applied
to 8-bit, 16-bit, and 32-bit posit formats.
Keywords: posit number system component; formatting; style; styling; insert (key words)
I. INTRODUCTION
A new datatype called Posit was created to completely supplant IEEE Standard 754 floating-point numbers. (floats). Plots do not
require interval arithmetic or variable size, in contrast to previous iterations of universal number (unum) arithmetic. operands;
similar to fractions, they round if a response is not precise. Larger dynamic range, better accuracy, better closure, bitwise identical
results across systems, simpler hardware, and easier exception management are just a few of the compelling benefits they offer over
floats. Posits never go over or under their limits, and "Not-a-Number '' (NaN) denotes an action rather than a bit sequence. An IEEE
float FPU requires more hardware than a posit processing unit. Furthermore, its uneven data distribution blends well with the
uneven data distribution of some uses, including deep learning. In deep learning algorithms, the 8-bit or 16-bit posit formats are
frequently employed. In some scientific computation uses, the 64-bit floating-point format is replaced by the 32-bit posit format. A
posit number is defined as Posit (nb,es), where nb is the overall bit width and es is the exponent bit width. Sign (s), regime (rg),
exponent (exp), and mantissa are its four constituents. (frac). The bit-width of the component varies. For various values, the regime
bit-width changes.
value = (−1)s × useedrg × 2exp × (1 + frac)
where used = 2^2res
In contrast to the traditional floating-point format, each component of the posit format, as shown in Fig. 1, has a dynamic bit breadth
(aside from the single-bit sign). The format is always used for the sign and the protocol. Only when the sign and regime do not take
up all of the bit positions do the leftover exponent and fraction part become visible. Since nb is the overall bit-width of the posit
format and es is the exponent bit-width, the mantissa's bit-width (which includes the implicit bit) can range from 1 bit to (nb es)-bit.
The mantissa does not always require a (nb es)-bit multiplier because the actual bit-width of the mantissa is not always the
maximum number. Power or energy is lost when using a (nb es)-bit multiplier for a tiny bit-width mantissa.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 2768
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue V May 2023- Available at www.ijraset.com
In this article, we demonstrate how a 16-bit posit multiplier, in which the mantissa (fraction) multiplier is divided into several
smaller multipliers, can be successfully implemented. At runtime, only the necessary factors are applied. So, to increase power
efficiency, we effectively create a low-power posit multiplier architecture. To increase the power efficiency of the related
component or device, this architecture can also be used in multipliers other than posit multipliers.
A ripple carry adder is a digital circuit that produces the arithmetic sum of two binary numbers. It can be constructed with full
adders connected in cascaded with the carry output from each full adder connected to the carry input of the next full adder in the
chain. The interconnection of four full adder (FA) circuits to provide a 4-bit ripple carry adder. But here, it will give the output for
whatever the input bit sequences with some delay. As per the digital circuits if the circuit gives output with delay won’t be
preferable. This can be overcome by a carry look-ahead adder circuit.
The 4-bit-carry-skip adder consists of a n-bit-carry-ripple-chain, a n-input AND-gate and one multiplexer. It is shown in figure 4.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 2769
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue V May 2023- Available at www.ijraset.com
Each propagate bit, that is provided by the carry-ripple-chain is connected to the n-input AND-gate. The resulting bit is used as the
select bit of a multiplexer that switches either the last carry-bit or the carry-in to the carry-out signal This greatly reduces the latency
of the adder through its critical path, since the carry bit for each block can now "skip" over blocks with a group propagate signal set
to logic 1.
The carry-select adder generally consists of ripple-carry adders and a multiplexer. Adding two n-bit numbers with a carry-select
adder is done with two adders (Therefore two ripple-carry adders), in order to perform the calculation twice, one time with the
assumption of the carry-in being zero and the other assuming it will be one. After the two results are calculated, the correct sum, as
well as the correct carry-out, is then selected with the multiplexer once the correct carry-in is known. It is shown in figure 5.
A carry select adder is an arithmetic combinational logic circuit which adds two N-bit binary numbers and outputs their N-bit binary
sum and a 1-bit carry.
A carry look-ahead adder definition is it is the faster circuit in performing binary addition by using the concepts of Carry Generate
and Carry Propagate. It is shown in figure 6.
A CLA is termed as the successor of a ripple carry adder. A CLA circuit minimizes the propagation delay time through the
implementation of complex circuitry.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 2770
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue V May 2023- Available at www.ijraset.com
The sum and carry can be recombined in a normal addition to form the correct result. This process may seem more complicated
and pointless, but the power of this technique is that any amount, number of additions can be added together in this manner. It is
only the final recombination of the final carry and sum that requires a carry propagating addition. 3:2 compressor is also known as
full adder. It adds three one bit binary numbers, a sum and a carry. The full adder is usually a component in a cascade of adders. The
carry input for the full adder circuit is from the carry output from the cascade circuit. Carry output from full adder is fed to another
full adder. It is used for low power consumption.
Time consumption consuming under proposed method is 14.19 ns. The power consumption is 20.00 watts.
IV. APPLICATION
Audio and video processing: Posit multipliers can be used in audio and video processing systems to perform multiplication
operations on signals such as sound or video data. This can help improve the accuracy and fidelity of the processed signals. Machine
learning: Machine learning algorithms often involve large numbers of multiplication operations, which can be computationally
expensive. Using posit arithmetic and posit multipliers can help reduce the computational cost of these operations while maintaining
accuracy. Cryptography: Posit arithmetic can be used in some cryptographic algorithms, such as elliptic curve cryptography. Using
posit multipliers can help improve the efficiency and security of these algorithms. Scientific computing: Posit arithmetic and posit
multipliers can be used in scientific computing applications where high precision and accuracy are required. This can include
simulations, modeling, and other numerical analysis tasks.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 2771
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue V May 2023- Available at www.ijraset.com
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 2772
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 11 Issue V May 2023- Available at www.ijraset.com
Performance Comparision
REFERENCES
[1] J. Johnson, “Rethinking floating point for deep learning,” CoRR, Nov. 2018.
[2] R. Chaurasiya et al., “Parameterized posit arithmetic hardware generator,” in Proc. IEEE 36th Int. Conf. Comput. Design (ICCD), Orlando, FL, USA, Oct.
2018.
[3] M. K. Jaiswal and H.-K. So, “Architecture generator for type-3 unum posit adder/subtractor,” in Proc. IEEE Int. Symp. Circuits Syst. (ISCAS), Florence,
Italy,May 2018.
[4] Podobas and S. Matsuoka, “Hardware implementation of POSITs and their application in FPGAs,” in Proc. IEEE Int. Parallel Distrib. Process. Symp.
Workshops (IPDPSW), Vancouver, BC, Canada, May 2018.
[5] Z. Carmichael, S. H. F. Langroudi, C. Khazanov, J. Lillie, J. L. Gustafson, and D. Kudithipudi, “Deep positron: A deep neural network using the posit number
system,” Dec. 2018.
[6] M. Klöwer, P. D. Düben, and T. N. Palmer, “Posits as an alternative to floats for weather and climate models,” in Proc. Conf. Next Gener. Arithmetic, Mar.
2019.
[7] H. Zhang, J. He, and S.-B. Ko, “Efficient posit multiply-accumulate unit generator for deep learning applications,” in Proc. IEEE Int. Symp. Circuits Syst.
(ISCAS), Sapporo, Japan, May 2019.
[8] Z. Carmichael, H. F. Langroudi, C. Khazanov, J. Lillie, J. L. Gustafson, and D. Kudithipudi, “Performance-efficiency trade-off of low-precision numerical
formats in deep neural networks,” in Proc. Conf. Next Gener. Arithmetic, Mar. 2019.
[9] Uguen, Yohann, Luc Forget, and Florent de Dinechin. "Evaluating the hardware cost of the posit number system." 2019 29th International Conference on
Field Programmable Logic and Applications (FPL). IEEE, 2019.
[10] Murillo, R., Del Barrio, A. A., Botella, G., Kim, M. S., Kim, H., & Bagherzadeh, N. (2021). PLAM: A posit logarithm-approximate multiplier. IEEE
Transactions on Emerging Topics in Computing.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 2773