GPGPU
GPGPU is the use of graphics processing units to perform general-purpose computations, often by exploiting their parallel processing capabilities.
Fast Python: High performance techniques for large datasets
- How it is used
- Accelerates data analysis by distributing similar computations across many GPU processing units.
- How deep it goes
- Develops thread indexing, kernel launches and bounds checks, while deferring algorithms using shared block state.
- Examples
- Uses an illustrative CPU–GPU cost model, array doubling and a Mandelbrot generator.
- What you can do after
- Apply it via Numba to launch and validate an array-processing GPU kernel.
Read how Fast Python: High performance techniques for large datasets explains GPGPU →
Grokking GPU Programming
GPU programming is the practice of writing software that executes computational work on graphics processing units.
- How it is used
- Uses GPUs for general-purpose computation, especially workloads that repeat operations across large amounts of data.
- How deep it goes
- Motivates GPU programming with CUDA history, a matrix-multiplication comparison, and overhead caveats, but gives no implementation steps.
- Examples
- Compares 4,096×4,096 matrix multiplication on a multi-core Xeon and an A100, including tensor-core performance.
- What you can do after
- Understand why GPUs suit highly parallel work and how transfer and launch costs can reduce gains.
CSS in Depth, Second Edition
GPU acceleration is rendering computation performed by a graphics processing unit instead of primarily by the central processor.
- How it is used
- Explains why transforms and opacity can make animated interface rendering faster.
- How deep it goes
- Explains layers, repainting, compositing, and memory trade-offs, but stops short of implementation code.
- Examples
- Uses background-color changes, element resizing, and animated opacity or transforms as contrasting rendering cases.
- What you can do after
- Evaluate when layer promotion and GPU acceleration can improve animated rendering and when memory costs may outweigh benefits.
Inside Deep Learning: Math, Algorithms, Models
The use of graphics processing hardware to speed up computation.
- How it is used
- Accelerates deep-learning code through PyTorch’s GPU support.
- How deep it goes
- Connects GPU acceleration to PyTorch’s tensor API but does not explain hardware or execution internals.
- Examples
- Uses faster PyTorch code as its practical illustration.
- What you can do after
- Apply it via PyTorch GPU acceleration to run code faster.
Read how Inside Deep Learning: Math, Algorithms, Models explains GPGPU →
Retrieval Augmented Generation, The Foundational Ideas
GPU acceleration is the use of graphics processing units to execute computations faster or in greater parallelism than general-purpose processors.
- How it is used
- Uses GPU parallelism to speed large-scale FAISS similarity search and support practical semantic retrieval.
- How deep it goes
- Demonstrates GPU-backed IVF setup and relates GPU speedups to search scale without quantifying GPU-specific latency.
- Examples
- Demonstrates transferring and populating an IVF index on GPU, alongside a million-vector FAISS search example.
- What you can do after
- Apply GPU acceleration through FAISS by transferring, training, and populating a GPU-backed inverted-file index.
Read how Retrieval Augmented Generation, The Foundational Ideas explains GPGPU →
CUDA for LLMs
GPU programming is the practice of writing software whose computation executes on graphics processing units.
- How it is used
- Uses custom CUDA kernels when framework operations are missing or bottleneck a deep-learning workload.
- How deep it goes
- Traces host/device coordination and thread-to-memory indexing, then stops at correct reference kernels rather than production-level performance.
- Examples
- Demonstrates PyTorch model transfer and naive CUDA kernels for matrix multiplication, convolution, softmax, and max pooling.
- What you can do after
- Build correct CUDA reference kernels for neural-network operations while recognizing that naive implementations trail production libraries substantially.
Also covered in
What to know first
- language modelA language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- machine learningMachine learning is a field of computational methods in which systems learn patterns or relationships from data to make predictions, decisions, or other outputs.
- retrieval-augmented generationRetrieval-augmented generation is a technique that retrieves relevant information from external sources and supplies it as context to a generative model when producing an output.
- arrayAn array is an ordered collection of elements organized in one or more dimensions and accessible by position or index.
Related concepts
- GPUuses · A graphics processing unit (GPU) is a processor designed for highly parallel computation, especially graphics and other data-parallel workloads.
- Numbaused by · Numba is a Python compiler that uses just-in-time compilation to generate optimized machine code for numerical code, with support for CPU and GPU targets.
- parallelismuses · Parallelism is the execution of multiple independent tasks or computational operations at the same time rather than sequentially.
- CUDAproduced by · CUDA is NVIDIA’s parallel computing platform and programming model for performing general-purpose computations on NVIDIA GPUs.
- HNSWcontrasts with · HNSW is a graph-based approximate nearest-neighbor indexing and search method that organizes data points in a hierarchy of navigable proximity graphs.
Questions about GPGPU
What is GPGPU?
GPGPU is the use of graphics processing units to perform general-purpose computations, often by exploiting their parallel processing capabilities.
Which Manning books explain GPGPU?
Fast Python: High performance techniques for large datasets, Grokking GPU Programming, CSS in Depth, Second Edition, Inside Deep Learning: Math, Algorithms, Models, Retrieval Augmented Generation, The Foundational Ideas and CUDA for LLMs.
What should I know before learning GPGPU?
Manning books that teach GPGPU usually introduce language model, machine learning, retrieval-augmented generation and array first.
What is the difference between GPGPU and CPU programming?
GPGPU is the use of graphics processing units to perform general-purpose computations, often by exploiting their parallel processing capabilities. CPU programming is the practice of designing and implementing computations for execution on a central processing unit.
What is GPGPU also called?
GPGPU is also called GPU computing or GPU programming.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.





