taichi-dev/taichi

[Lang] Make `@ti.func` callable from Python-scope, or `@ti.kfunc` if we're afraid of breaking API

Open

#1,379 opened on Jul 2, 2020

View on GitHub
 (1 comment) (1 reaction) (0 assignees)C++ (28,195 stars) (2,384 forks)batch import
feature requestgood first issuepython

Description

Concisely describe the proposed feature It would be convenient to unify these two concepts, both for API clarity and reducing user learning obstacle. This issue will be just a begining of this refactor.

Describe the solution you'd like (if any) We may first add @ti.kfunc as a experimental one, then replace @ti.func when it's mature and battle-tested.

  1. When @ti.func called from Python-scope will be translated into a minimal kernel that executes it, and exit.
  2. Add @ti.kfunc that create both a @ti.func and @ti.kernel, when called from corresponding scope, call func/kernel.
def kfunc(foo):
  func_foo = ti.func(foo)
  kernel_foo = ti.kernel(foo)
  @functools.wraps(foo)
  def wrapped(*args, **kwargs):
    if impl.inside_kernel():
      func_foo(*args, **kwargs)
    else:
      kernel_foo(*args, **kwargs)

Additional comments @yuanming-hu @k-ye WDYT? Inputs are welcome!

Contributor guide

Tech stack
pythoncpp
Domain
apiperformancedeveloper experience
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Basic Taichi knowledgePython decoratorsGPU programming concepts
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
60
Research direction
Investigate the current implementation of `@ti.func` and `@ti.kernel` decorators in the Taichi codebase. Understand how the `impl.inside kernel()` function works to determine execution context. Look at the provided code snippet and consider how to integrate it into the existing decorator chain. Check for any existing tests or issues related to this feature.