Gpu core
A CUDA core is a relatively simple arithmetic and logic execution unit in an NVIDIA GPU that processes instructions for individual threads.
Build a DeepSeek Model (From Scratch)
A CUDA core is a general-purpose arithmetic processing unit in an NVIDIA GPU that supports a broad range of computations.
- How it is used
- Uses CUDA cores to recover accumulation precision when low-precision Tensor Core computation risks numerical error.
- How deep it goes
- Explains CUDA Core and Tensor Core roles, WGMMA bursts, periodic promotion, and FP32 accumulation, but stops short of implementation code.
- Examples
- Uses k≥4096 matrix products, WGMMA accumulation bursts, periodic promotion, and potential errors approaching 2%.
- What you can do after
- Evaluate how periodic CUDA Core promotion can improve accumulation precision in low-precision matrix multiplication.
Read how Build a DeepSeek Model (From Scratch) explains gpu core →
CUDA for LLMs
A GPU core is a simple GPU execution component composed of arithmetic and logic units for parallel numerical operations.
- How it is used
- Explains how GPU cores support high-throughput parallel work in deep-learning workloads.
- How deep it goes
- Defines ALUs and CUDA-thread scheduling, while deferring deeper streaming-multiprocessor architecture.
- Examples
- Uses million-element ReLU work and transformer attention and projection GEMMs as parallel-work examples.
- What you can do after
- Evaluate whether general-purpose CUDA cores or specialized tensor cores suit a workload’s size and operation type.
Also covered in
- Fast Python: High performance techniques for large datasets
- GPU Programming with TritonIn early access (MEAP)
What to know first
- threadA thread is a distinct sequence of execution with its own control flow and state, managed or scheduled by a processor, runtime, or operating system.
- bfloat16bfloat16 is a 16-bit floating-point format with one sign bit, eight exponent bits, and seven fraction bits, providing a wide numerical range with reduced precision.
- inner productAn inner product is a scalar-valued operation on two vectors that generalizes the dot product, which for real coordinate vectors is the sum of products of corresponding components.
- matrix-matrix multiplicationMatrix multiplication is an operation on compatible matrices in which each entry of the product is the dot product of a row of the first matrix and a column of the second.
Related concepts
- streaming multiprocessorpart of · A streaming multiprocessor is a GPU execution unit that schedules and executes groups of threads using its processing and on-chip memory resources.
- Tensor Corecontrasts with · A Tensor Core is a specialized NVIDIA GPU execution unit designed to accelerate high-throughput matrix and tensor arithmetic.
- parallelismused by · Parallelism is the execution of multiple independent tasks or computational operations at the same time rather than sequentially.
Questions about gpu core
What is gpu core?
A CUDA core is a relatively simple arithmetic and logic execution unit in an NVIDIA GPU that processes instructions for individual threads.
Which Manning books explain gpu core?
Build a DeepSeek Model (From Scratch) and CUDA for LLMs.
What should I know before learning gpu core?
Manning books that teach gpu core usually introduce thread, bfloat16, inner product and matrix-matrix multiplication first.
What is the difference between gpu core and Tensor Core?
A CUDA core is a relatively simple arithmetic and logic execution unit in an NVIDIA GPU that processes instructions for individual threads. A Tensor Core is a specialized NVIDIA GPU execution unit designed to accelerate high-throughput matrix and tensor arithmetic.
What is gpu core part of?
Gpu core is part of streaming multiprocessor. A streaming multiprocessor is a GPU execution unit that schedules and executes groups of threads using its processing and on-chip memory resources.
What are the parts of gpu core?
Gpu core includes execution unit.
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.

