liveBook · Manning

Block-based language

As explained in GPU Programming with Triton

A block-based language expresses computation over whole data blocks rather than individual scalar threads.

A block-based language expresses computation over whole tensor tiles instead of individual scalar GPU threads. The compiler maps these tile operations onto hardware, combining Python-level productivity with performance close to hand-tuned CUDA. section 1.1.1

The model changes GPU implementation in several practical ways:

  • Operate on blocked, multidimensional arrays rather than interleaved scalar-thread instructions. section 1.1.1
  • Let the compiler map tile operations to GPU hardware and generate target-specific low-level representations. section 1.1.1
  • Use one kernel source across NVIDIA, AMD through ROCm, and Intel through the XPU backend. section 1.1.1

How GPU Programming with Triton covers it

How deep it goes
Defines whole-tile operations, compiler mapping, and portability, while deferring formal treatment of threads, warps, and grids.
Examples
Uses blocked tensor operations and multidimensional numerical arrays as its concrete computational domain.
What you can do after
Build it from scratch: write tile-oriented kernels whose compiler mapping targets NVIDIA, AMD, and Intel GPUs.
How it connects
Connects tensors, GPU threads, CUDA, compiler lowering, and vendor backends.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of GPU Programming with Triton. The section links point to the parts of the book it draws on.