Processors, Parallel Processing & Virtual Machines (Cambridge (CIE) A Level Computer Science): Flashcards

RISC has lower power usage and lower cost. CISC is higher on both.

RISC is usually used in smartphones and tablets. CISC is usually used in laptops and desktop computers.

CISC

RISC

False.

It will not work. It will not necessarily work on another RISC processor either, because they may have different instruction sets.

Using microcode and dedicated hardware instructions, which manages interrupts efficiently.

Using software-based interrupt handlers, managed through general-purpose registers and simple instructions.

CISC may have hardware support to save and restore the processor state automatically. RISC typically handles it in software, giving more control but requiring more instructions.

CISC can handle multiple tasks in a single instruction, including context saving.

CISC, because of its built-in support. RISC is slower, but more flexible and consistent across all instructions.

False.

CISC is generally faster due to built-in support. RISC is slower but more flexible.

It is more efficient but less flexible.

It is simpler and more predictable, and gives the operating system greater control.

Respond to interrupts quickly and resume execution correctly, using different design philosophies.

Pipelining is the process of carrying out multiple instructions concurrently, with each instruction at a different stage of the fetch-decode-execute cycle.

False.

Each instruction is at a different stage of the fetch-decode-execute cycle.

One instruction can be fetched while the previous one is being decoded and the one before that is being executed.

The pipeline is flushed.

Instruction C is being fetched, instruction B is being decoded, and instruction A is being executed.

Instruction fetch (IF), instruction decode (ID), operand fetch (OF), instruction execute (IE), and write back result (WB).

One clock cycle.

No. No two instructions can execute their same stage at the same clock cycle.

In the second clock cycle, while the first instruction has moved on to its second subtask.

Each instruction takes one clock cycle, instructions are fixed-length, the instruction set is limited, and there are many general-purpose registers.

They make decoding easier.

It reduces complexity and stage duration.

Registers are fast, temporary storage inside the CPU.

RISC processors rely heavily on registers for operand storage rather than accessing RAM.

Register usage reduces memory bottlenecks, allowing pipelining to run smoothly.

In registers between stages.

Increased throughput, better use of CPU components, reduced idle time, and faster execution of the instruction stream.

Pipelining means the fetch, decode and execute units are all in use simultaneously.

False.

Individual instructions do not run faster. It is the instruction stream as a whole that executes faster.

A computer architecture is the design and structure of a computer system, describing how it fetches, processes and stores data and instructions.

How components like the CPU, memory and input/output devices work together to execute programs.

Whether it has a single or multiple instruction stream, and a single or multiple data stream.

Single Instruction, Single Data: one processor executes one instruction on one data stream at a time.

SISD is a traditional serial, non-parallel architecture.

In basic, single-core processors, for example a classic desktop CPU running one task at a time.

Single Instruction, Multiple Data: one instruction is applied to multiple pieces of data at once.

They perform the same operation in parallel.

False.

All processing units perform the same operation in parallel.

SIMD is ideal for graphics or scientific computation.

In GPU operations and image processing, where it is useful for parallel processing.

Multiple Instruction, Single Data: multiple processors execute different instructions on the same data.

Very uncommon.

Specialised systems for fault tolerance, where each unit checks the same input differently, such as redundant systems in safety-critical environments.

Multiple Instruction, Multiple Data: multiple processors execute different instructions on different data sets.

Full concurrent processing of independent tasks, making it flexible and scalable for parallel programs.

In most modern multi-core processors, distributed systems and cloud computing.

False.

MIMD is used by most modern multi-core processors. MISD is very uncommon in practice.

SIMD applies one instruction to many data items at once. MIMD runs different instructions on different data, fully independently.

SIMD, for example in GPUs.

A system made up of thousands of processors working simultaneously to solve a single large problem.

Each processor executes part of a program, and the results are combined to produce the final output.

False.

They work simultaneously to solve a single large problem, each executing part of a program.

Scientific research, weather simulation, cryptography and AI.

Each processor may have its own memory, or access shared memory resources.

Processors in a massively parallel system are linked by high-speed interconnects.

High-speed interconnects, which are fast communication pathways used to share results.

Data parallelism is where many data items are processed at once, often using SIMD or MIMD.

Task parallelism is where different processors perform different tasks on different data sets.

Specialised software, written to distribute the work efficiently across the processors.

Massively parallel systems are tightly coupled, so the processors depend on one another.

Processors depend on one another and work collaboratively as a single system.

Massively parallel systems are thousands of processors forming a single tightly integrated system. Clusters are multiple independent systems networked together.

False.

A cluster is multiple independent systems networked together, often loosely coupled. It is the massively parallel computer that is tightly integrated.

Massively parallel processors communicate continuously via shared architecture. Clusters communicate via a network, often loosely coupled.

A massively parallel system acts like one machine with distributed processing. A cluster is a group of co-operating systems.

Massively parallel systems give higher performance and are used for supercomputing tasks. Clusters are more general-purpose or batch processing systems.

SIMD, applying one instruction to many data points at once, and MIMD, running different instructions on different data.

A virtual machine is an entire operating system running inside another operating system.

Navigate its GUI and install software on it.

It helps access software that is only designed to run on specific operating systems.

A hypervisor is VM management software that monitors all activity happening inside the virtual machine.

Because some software is written to run only on MacOS, so the virtual machine lets them install and use it.

Most software needs updating to work on the latest operating system versions, so a user on the newest release may run a VM of a previous release to use an application that has not had a forwards-compatibility update.

They create isolated test environments that leave the host operating system unaffected.

False.

Virtual machines create isolated test environments that leave the host operating system unaffected.

Monitor how their software affects system performance, and test on a clean-slate system with no other applications running.

VM software can create machines that act like they have older hardware, so developers can build software that runs on older hardware too.

A developer can test against various operating systems, such as MacOS, Linux and Windows, for greater compatibility.

Intermediate code, which is generated through compilation.

VMs share the same system hardware as the host operating system.

False.

Virtual machines share the same system hardware as the host operating system.

It can exhaust the host system of CPU, hard disk and memory.

VM software can set maximum limits on system resources.

To allocate only 1GB of memory and 20% of the CPU.

Up to 8GB of memory and 50% of the CPU.

Operating systems are commonly free to download, but require an activation payment to access all features.