ACA Assignment 

 
Name: Anjali Singh Roll No.: 10001012017 
Semester: 7th Class: CSE-2 
 
 
MULTI SCALAR ARITHMETIC PIPELINE 
 

Arithmetic Pipelines are mostly used in high-speed computers. They are 

used to implement​ floating-point operations, multiplication of fixed-point 
numbers, and similar computations encountered in scientific problems.  

To understand the concepts of arithmetic pipeline in a more 

convenient way, let us consider an example of a pipeline unit for 
floating-point addition and subtraction.  

The inputs to the floating-point adder pipeline are two normalized 

floating-point binary numbers defined as:  

X = A * 2​a​ = 0.9504 * 10​3  

Y = B * 2​b​ = 0.8200 * 10​2  

Where A and B are two fractions that represent the mantissa and a 
and b are the exponents.  

The combined operation of floating-point addition and subtraction is 

divided into four segments. Each ​segment contains the corresponding 
sub operation to be performed in the given pipeline. The sub​ operations 
that are shown in the four segments are:  

1. Compare the exponents by subtraction.  

2. Align the mantissas.  
3. Add or subtract the mantissas.  
4. Normalize the result. 

A scalable engine having multiple datapaths, each of which is a unique 

multi-function floating point ​pipeline capable of performing a four 
component dot product on data in a single pass through the​ ​datapath, 

1
which allows matrix transformations to be computed in an efficient 
manner, with a high data​ ​throughput and without substantially 
increasing the cost and amount of hardware required to implement​ the 
pipeline. 

Fig 1. ​Multi-function floating point arithmetic pipeline 

2
SUPERSCALAR PIPELINE 

Superscalar architecture is a method of parallel computing used in many  

processors. In a superscalar computer, the central processing unit (CPU) 
manages ​ ​multiple instruction pipelines to execute several instructions 
concurrently during ​ ​a clock cycle.  
A superscalar architecture includes parallel execution units, which can 
execute ​ ​instructions simultaneously. This parallel architecture was first 
implemented in RISC ​ ​processors, which use short and simple instructions 
to perform calculations. Because ​ ​of their superscalar capabilities, RISC 
processors have typically performed better ​ ​than CISC processors 
running at the same megahertz. However, most CISC-based​ ​processors 
(such as the Intel Pentium) now include some RISC architecture as well,  
which enables them to execute instructions in parallel. Nearly all 
processors ​ ​developed after 1998 are superscalar. In a superscalar design, 
the processor or the ​ ​instruction compiler is able to determine whether an 
instruction can be carried out ​ ​independently of other sequential 
instructions, or whether it has a dependency on ​ ​another instruction and 
must be executed in sequence with it. The processor then ​ ​uses multiple 
execution units to simultaneously carry out two or more independent  
instructions at a time. Superscalar design is sometimes called “second 
generation​ ​RISC.”  
 

Advantages of Superscalar Architecture :  

• ​In a Superscalar Processor, the detrimental effect on performance of 
various ​ ​hazards becomes even more pronounced.  
• ​The compiler can avoid many hazards through judicious selection 
and ​ ​ordering of instructions.  
• ​The compiler should strive to interleave floating point and integer  
instructions. This would enable the dispatch unit to keep both the 
integer ​ ​and floating point units busy most of the time.  
 

Disadvantages of Superscalar Architecture :  

Due to this type of architecture, problems in scheduling can occur. 
 

3