Scribd
Upload
67 views24 pages

Sri.S.A.Hariprasad SR - Lecturer R.V.C.E Bangalore

The document discusses various techniques to improve computer performance such as cache memory, pipelining, superscalar execution, and increasing processor clock frequency. It also covers the differences between RISC and CISC processors as well as multiprocessor and multicomputer systems. Performance is measured using SPEC ratings which calculates the execution time on the system being tested relative to a reference computer.

Uploaded by

pranav_47
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
67 views24 pages

Sri.S.A.Hariprasad SR - Lecturer R.V.C.E Bangalore

The document discusses various techniques to improve computer performance such as cache memory, pipelining, superscalar execution, and increasing processor clock frequency. It also covers the differences between RISC and CISC processors as well as multiprocessor and multicomputer systems. Performance is measured using SPEC ratings which calculates the execution time on the system being tested relative to a reference computer.

Uploaded by

pranav_47
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 24

|  

Sri.S.A.Hariprasad
Sr.Lecturer
R.V.C.E Bangalore
PERFORMANCE EQUATION

N ×S
T= R
N - Number of instruction (m/c)
S ± Average steps required to execute one machine instruction
R- Clock frequency
PERFORMANCE IMPROVEMENT
] Cache memory
]Pipelining
]Superscalar execution
CACHE MEMORY
] Cache can be built within the processor ± L1 cache
]Cache can be placed between main memory and CPU ± L2 Cache
]Cache size being small, currently required instructions and data for
execution are stored in cache
]Entire instructions and data connected with program cannot be stored
at a time, if program is big
]New set of programs (instructions and data) can be loaded by
swapping the previous contents back into memory
] Cache Hit ± If wanted item by CPU (instruction or data) is
present in cache.
]Cache miss ± If wanted item by CPU is not present in cache
]To improve the performance hit ratio to be increased
PIPELINING

]Normal sequence of execution in sequential machines will


have the 4 phases ± fetching, decoding, executing, writing.

]Till the completion of instruction execution other instruction


will not be fetched

]Only one unit will be active at a time


] Pipelining: Improves the throughput of the system

]Throughput: No of instructions executed per second or per


. unit time

]Perfecting : Fetching of an instruction prior to it execution and


placing them in instruction queue when some other instruction is
under execution
1st CLK CYCLE F D E W

2nd CLK CYCLE F D E W

3rd CLK CYCLE F D E W

4th CLK CYCLE F D E W


PIPELINING CONCEPT OF EXECUTION
Eg: ADD R1, R2, R3,(Assuming 4 clk cycle required for execution)
MOVE R4, R5 ( Assuming 4 clk cycle required for execution)

T1 T2 T3 T4 T5 T6 T7 T8

ADD R1,R2,R3 MOVE R1,R2,R3


F D E W
ADD R1,R2,R3

F D E E
MOV R4,R5

Clk cycles 1 2 3 4 5
PERFORMANCE MEASURMENT

] SPEC: Organization called system performance evaluation


corporation
]Developed in the year 1995 and modified in the year 2000
]SPEC 95 uses SUN SPARC station 10/40 as standard
]SPEC 2000 uses ultra SPAC 10 with a 300 Mhz processor
Running time on the reference computer
SPEC rating = Running time on the computer under test

1/ 2
r

SPEC rating =   J
 1
MULTIPROCESSOR AND
MULTICOMPUTERS
MULTIPROCESSOR SYSTEM: System which
uses more than one processor unit.

Advantage: Different application tasks can be executed


in parallel
OR
Single large task can be divided into small task and are
distributed to all processor
]All the processor shares the memory
]Increased performance is achieved at the cost of increased
complexity in design and debugging
]Different processor can communicate with each other using
interconnection network
]Cost will also increase
]Increase software overhead for the design of optimized and
sophisticated system software tool to integrate or coordinate
different processor
MULTICOMPUTER OR
DISTRIBUTED SYSTEM

] Interconnected complete computer (own local memory) to


achieve high computational power
]Information exchange is through communication network
RISC & CISC PROCESSOR
CISC: COMPLEX INSTRUCTION SET COMPUTER
]Complex instruction set
]Micro programmed control unit
]Frequent memory access
]Different length instruction formats
]More addressing modes
]Instruction execution cycles are different

]Doesn¶t support HLL

]S.>1, N small

]Less member of registers


RISC: Reduced instruction set computers
]Reduced instruction set
]Hard wired control unit
]Memory access is limited to load and store operation
]Fixed instruction length
]Less addressing modes
]Single ± cycle execution

]Support to HLL

]N >>1, S is small

]Supports pipelining
DISADVANTAGES OF RISC
]RISC programs are longer than CISC program

]More memory is required to store more program

]Hence traffic between the memory and CPU will increase


Computer Architecture : It is an abstraction of a physical
computing system as seen by a machine language programmer or a
compiler writer

Computer Organization: It refers to the physical arrangement


and interconnection of the components in a computing system and the
characteristics of those components
EXPECTED QUESTIONS
]Explain the various functional units of computer?
]Explain the steps involved in the execution of an instruction?
]Explain how performance of a system can be improved?
]Comment on
(a) Performance (b) Performance execution (c) SPEC rating
]Explain the difference between RISC and CISC machines?
]Write a short note on computer software?
]Explain the difference between multiprocessor and multi computers?
(a) Given the number of instruction in a program is 10, average
steps required is 4. Clock speed of processor is 4Mhz
calculate the time taken to execute the program
(b) If processor speed is increased to 8Mhz calculate the time
required to execute the same program

You might also like