Ridge point
A ridge point is the arithmetic intensity at which a roofline model transitions from memory-bandwidth-limited to compute-throughput-limited performance.
Grokking GPU Programming
A ridge point is the arithmetic-intensity threshold that separates memory-bound from compute-bound performance for a given bandwidth and compute rate.
- How it is used
- Classifies whether a kernel is memory-bound or compute-bound by comparing arithmetic intensity with a hardware-specific threshold.
- How deep it goes
- Defines the threshold, works through its calculation, and notes that latency, occupancy, or instruction throughput can also limit performance.
- Examples
- Uses a 100-operation kernel and stride-8 access to compare ideal and effective ridge points.
- What you can do after
- Diagnose whether memory traffic or arithmetic is the first suspect by comparing intensity with the effective ridge point and profiling bandwidth.
GPU Programming with Triton
A ridge point is the arithmetic-intensity threshold separating memory-bound and compute-bound performance on a roofline model.
- How it is used
- Classifies performance as memory-bound or compute-bound by comparing arithmetic intensity with a roofline threshold.
- How deep it goes
- Defines the threshold through arithmetic intensity and gives a numerical comparison, without deriving the H100 ridge point.
- Examples
- Compares SwiGLU elementwise intensity of about 32 FLOPs per byte with the H100 Tensor Core ridge point of roughly 590.
- What you can do after
- Evaluate whether a workload is memory-bound by comparing its arithmetic intensity with the ridge point.
Related concepts
- arithmetic intensitycontrasts with · Arithmetic intensity is the ratio of computational operations performed to bytes of data transferred by a computation.
- compute-boundcontrasts with · A computation is compute-bound when its performance is limited primarily by processing capacity rather than memory transfer, input/output, or other resources.
- memory boundcontrasts with · A workload is memory-bound when its performance is limited primarily by memory access or data-transfer capacity rather than computational throughput.
- effective memory bandwidthuses · Effective memory bandwidth is the achieved rate at which a computation transfers data through memory, reflecting its memory-access efficiency rather than the memory system's peak rate.
Questions about ridge point
What is ridge point?
A ridge point is the arithmetic intensity at which a roofline model transitions from memory-bandwidth-limited to compute-throughput-limited performance.
Which Manning books explain ridge point?
Grokking GPU Programming and GPU Programming with Triton.
What is the difference between ridge point and arithmetic intensity?
A ridge point is the arithmetic intensity at which a roofline model transitions from memory-bandwidth-limited to compute-throughput-limited performance. Arithmetic intensity is the ratio of computational operations performed to bytes of data transferred by a computation.
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.

