undefined | unit 5 data level parallelism in vector simd and gpu architectures

Back to Study material

ACA

UNIT 5

Data-Level Parallelism in Vector, SIMD and GPU Architectures

Q1) Mention The Pros And Cons Of Emulation?

A1)

Emulation is the process in which a target CPU and its corresponding hardware would be emulated exactly the same way.

Emulation is a relatively old concept and is not widely used to emulate full scale OS usage.

It is considered to be the best platform for embedded/os development.

Emulation is possible for any hardware and it does not affect the underlying OS ( host ).

Although there are many positives of emulation there are few downside of it as well, Emulation can be extremely slow.

Complete thorough hardware support cannot be possible in emulation.

Emulation is also very resource hungry and requires a lot of ram to function smoothly.

Q2) What Is Software Interrupts?

A2)

These types if interrupts can occur only during the execution of an instruction. They can be used by a programmer to cause interrupts if need be. The primary purpose of such interrupts is to switch from user mode to supervisor mode.

Q3) What Is Vertical Microcode?

A3)

Vertical microcode can be considered to be a segment of code or operators that have been clubbed together into fields. In this field every micro operation is given a unique value.

Q4) Explain About Designing Strategy Of A Control Unit Coded On Vertical Code?

A4)

A no operation NOP can be included in each field if necessary.

The remaining micro operations can be distributed among the other operation field bits.

Also micro operations that modify the same registers could be grouped together in the same field.

Q5) Can You Explain The Two Hardware Methods To Establish Priority?

A5)

Two different ways of establishing hardware priority are Daisy Chaining and parallel priority.

Daisy chaining is a form of a hardware implementation of the polling procedure.

Parallel priority is quicker of the two and uses a priority encoder to establish priorities.

In parallel priority interrupt a register is used for which the bits are separated by the interrupt signals from every device.

The parallel priority interrupt may also contain a mask register which is primarily used to control the status of every request regarding interrupts.

Q6) Described Some Of The Common Rules Of Assembly Language?

A6)

Some of the common rules of assembly level language are as follows:

In assembly language the label field can be either empty or may specify a symbolic address.

Instruction fields can specify pseudo or machine instructions.

Comment fields can be left empty or can be commented with.

Up to 4 characters are only allowed in the case of symbolic addresses.

The symbolic addresses field are terminated by a comma whereas the comment field begins with a forward slash.

Q7) List The Components Of The Ven Neumann Architecture?

A7)

The main components of the Von Neumann architecture were as follows:

It consisted of a main memory which would be used to store all the data and instructions.

It would consist of an arithmetic logical unit also known as the ALU. This part was to be able to work with binary data.

It also comprised of a control unit which would be responsible for the interpretation of instructions and their execution.

The control unit would also be controlled by the control unit itself.

Q8) Explain About The Major Difficulties Of Pipeline Conflicts In Processors Supporting Pipe Lining?

A8)

The following are the main reasons for pipe line conflicts in the processor:

When the same resource is accessed at the same time by two different segments it results in resource conflicts. The only way to resolve this problem is to use separate data memories.

In case an instruction's execution depends on the result of a previous instruction and that result is unavailable it leads to data dependency conflicts.

Instructions that change the count of the PC can cause a lot of problems. This is prevalent particularly in the case of Branch instructions. A method to resolve this issue is known as delayed load where certain instruction are made to execute in a delayed manner to avoid conflicts.

Q9) Described The Different Types Of Fields That Are Part Of An Instruction?

A9)

An instruction can be considered to be a command that has been issued to a computer to perform a particular operation. The instruction format contains various field in them such as:

Operation Code Field: Also known as the op code field, this field is used to specify the operation to be performed for the instruction.

Address Field: This field as its name specifies is used to designate the various addresses such as register address and memory address.

Mode field: This field specifies as to how effective address is derives or how an operand is to perform.

For ex: ADD R0, R1. In this case the ADD is the operand whereas the R1, ans R0 are the address fields.

List The Different Types Of Micro-operations?

Answer: The micro-operations in computers are classified into the following categories:

Register transfer micro-operations: These type of micro operations are used to transfer from one register to another binary information.
Arithmetic micro-operations: These micro-operations are used to perform on numeric data stored in the registers some arithmetic operations.
Logic micro-operations: These micro operations are used to perform bit style operations / manipulations on non-numeric data.
Shift micro operations: As their name suggests they are used to perform shift operations in data store in registers.

Sign Up