Start Discovering Solved Questions and Your Course Assignments
TextBooks Included
Active Tutors
Asked Questions
Answered Questions
vector reduction instructionswhen operations on vector are being decreased to scalar items as a result then these are vector reduction instructions
vector-scalar instructionsin this category when combination of vector and scalar are fetched and stored in vector register these instructions are
vector-vector instructionsin this category vector operands are fetched from vector register and accumulated in another vector register these
vector processing a vector is an ordered set of similar type of scalar data items the scalar tem may be a logical value an integer or a floating
throughputthroughput of a pipeline may be defined as number of results which have been achieved per unit time it can be referred ast n m n-1 c e
q describe target processor arrangementshaving seen how to describe one or more target processor arrangements we need to initiate mechanisms for
q what do you mean by data distributiondata distribution directives tell compiler how program data is to be distributed among memory areas connected
efficiencythe effectiveness of pipeline can be measured the same as the ratio of busy time to total time counting the idle time let c be
q example of processor arrangementshpf processors p 10this initiates a group of 10 abstract processors assigning them combined name phpf processors q
q what do you mean by processor arrangementsit is a very common event in data parallel programming to combine many processors to execute specific
in 1993 high performance fortran forum which is a group of academicians and many leading software and hardware vendors in field of parallel
q show programming based on data parallelismin data parallel programming model the focal point is on data distribution every processor works with a
q addition of array elements using two processorsin this example we have to find sum of all elements of an array a of size n we will divide n
q illustrate programming based on message passingsince we know programming model based on message passing employs high level programming languages
q what are the types of parallel programmingthere are various parallel programming models in general use a few of them are data parallel
conventionally software has been written for serial computation in that programs are written for computers which have a single central processing
q explain the properties of hypercubeproperties of hypercube hypercube is both edge and node symmetric the labels of any two neighbouring nodes vary
q programming languages array operationsin programming languages array operations are written in compact form which frequently makes programs more
speedupfirst we take the speedup factor which is we see how much speed up performance we achieve by pipelining first we take ideal case for measuring
q example of arrays pointersgeneral form of declaration of array in fortran 90 istype dimensionbound attr nameeg the
arithmetic pipelinesthe technique of pipelining can be applied to various complex and low arithmetic operations to speed up processing time pipelines
instruction buffersfor taking the complete advantage of pipelining pipelines must be filled continuously so instruction fetch rate must be matched
bernstein conditions for detection of parallelismfor execution of a number of instructions or a block of instructions in parallel it must be made
resource dependencethe parallelism between instructions can also be affected because of the shared resources if two instructions are occupying the
control dependencesegments or instructions in a program can include control structures so dependency among statements is able to be in control