IT1910: Advanced Architectures - Virtualization

Virtualization

• The American Heritage Dictionary defines virtual as:
  • Existing or resulting in essence or effect, but not in actual fact or form
  • Created, simulated, or carried on by means of a computer or computer network
• Virtualization of a computer allows a single computer to appear as multiple computers, each with its own operating system and hardware resources
• Multiple computers share the same CPU and I/O facilities, with each simulated machine known as a virtual computer or virtual machine

Parallel Computing

• Preemptive multitasking refers to the ability of the operating system to interrupt a task and switch to another task
• Concurrency is the perception of having tasks that run at the same time, while parallelism is the actual running of tasks at the same time
• Parallelism is a subset of concurrency
• Two problems that can occur when multiple threads access shared memory are:
  • Data Race: a writer thread modifies the memory while a reader thread is reading from it
  • Race Condition: multiple threads access the same shared data and try to change it simultaneously
• Solutions to these problems include:
  • Synchronization: ensuring that resources are used by only one thread at a time
  • Atomic Operations: using special instructions to ensure that shared data is always in a valid state
  • Immutable Data: marking shared data as immutable, so that threads can only read from it

Multiprocessor and Multicore Systems

• Multicore microprocessor: a microprocessor that contains multiple processors (cores) in a single integrated circuit
• Task-level parallelism (process-level parallelism): a method of utilizing multiple processors by running independent programs simultaneously
• Flynn's Taxonomy: a categorization of parallel hardware based on the number of instruction streams and data streams
  • SISD (Single Instruction stream, Single Data stream): a non-parallel computer
  • SIMD (Single Instruction stream, Multiple Data streams): a parallel computer, best suited for specialized problems with regularity
  • MISD (Multiple Instruction streams, Single Data stream): a parallel computer, used in multiple frequency filters and cryptography
  • MIMD (Multiple Instruction streams, Multiple Data streams): a parallel computer, currently the most common type, used in supercomputers and multi-core PCs

Multithreading

• MIMD relies on multiple processes or threads to keep multiple processors busy
• Hardware multithreading allows multiple threads to share the functional units of a single processor
• Types of multithreading:
  • Fine-grained multithreading: switches between threads on each instruction, resulting in interleaved execution of multiple threads
  • Coarse-grained multithreading: switches threads only on expensive stalls, such as last-level cache misses
  • Simultaneous multithreading (SMT): a variation of hardware multithreading that uses the resources of a multiple-issue, dynamically scheduled pipelined processor to exploit thread-level parallelism and instruction-level parallelism

Vector/Array Processing

• Vector architecture: a set of vector registers that collect data elements from memory, operate on them sequentially, and write the results back to memory
• Instruction set architectures (scalar architectures): a single vector instruction that specifies a great deal of work, equivalent to executing an entire loop
• Benefits of vector architecture: independence of computation within a vector, allowing for efficient processing of large datasets

High-Performance Computing (HPC)

• HPC refers to the use of aggregated computing power to handle computation and data-intensive tasks
• Clustering: a group of loosely coupled computers that work together as a unit, with each computer (node) having its own CPU, memory, and I/O facility
• Characteristics of clustering:
  • Increasing computing power by combining individual systems
  • Creating fault-tolerant systems
  • Creating high-availability systems
  • Load-balancing systems with large workloads
• Models of clustering:
  • Shared-nothing model
  • Shared disk model

Cloud Computing

• Cloud computing: a conceptual expansion of client-server computing, where many functions of an organization's data center are moved to services on the Internet
• Benefits of cloud computing:
  • Simplified data center
  • Cost reduction
  • Strong support for user collaboration
  • Scalability
  • Access to files from anywhere the Internet is available
• Risks of cloud computing:
  • Security concerns
  • Dependence on the cloud service provider
  • Outages or connectivity issues
  • Data leakage or theft

Graphics Processing Unit (GPU) Architecture

• GPU: a processor optimized for 2D and 3D graphics, video, visual computing, and display
• GPU computing: computing via a parallel programming language and API, without using the traditional graphics API and graphics pipeline model
• APIs used in rendering and computing:
  • Direct3D
  • OpenGL
  • OpenCL
  • Vulkan### Multithreading
• A gentle introduction to multithreading is available online as a web log post.

Vector Processor

• A vector processor is a type of microprocessor that is designed to perform vector calculations.
• It is a central processing unit (CPU) that is capable of executing the same instruction on multiple pieces of data simultaneously.

High-Performance Computing (HPC)

• High-performance computing (HPC) refers to the use of high-performance computers and algorithms to solve complex problems.
• HPC is used to solve complex problems in fields such as science, engineering, and finance.

Computer Systems Architecture

• Computer systems architecture is a field of study that deals with the design and organization of computer systems.
• The book "Computer Systems Architecture" by A. Yadin, published in 2016, is a comprehensive resource on this topic.

Miscellaneous

• STI (name of the institution) provides a feedback form for students at [email protected].