[Docs] State the compile boundary in the manifest guide as the entry derives it by lcy-seso · Pull Request #74 · tile-ai/TileOPs.github.io

[Docs] State the compile boundary in the manifest guide as the entry derives it

lcy-seso

committed

73d899d

[Docs] State a target kernel's request_tune and config, register_device_facts, and what a non-CUDA backend's tests and reports cover

lcy-seso

committed

e89df26

[Docs] Point the dev image at ghcr.io/tile-ai/tileops-dev

lcy-seso

committed

f775414

[Docs] Pass a device facts reader the whole device, and scope the per-device tuning and benchmark statements

lcy-seso

committed

6b3d6b1

[Docs] Put one rule in each cell of the backend contract, and state reads, config and timing as the code does

lcy-seso

committed

2478640

[Docs] State the three markers a backend's test run reads, and that its target serves the device

lcy-seso

committed

46d657a

[Docs] Align the op, dispatch and manifest guides with the call device and target tuning on main

lcy-seso

committed

cd9c557

[Docs] Scope the compile boundary, value checks and CUPTI timing statements to where main holds them

lcy-seso

committed

d7243c8

[Docs] Say up front that the timing page covers CUDA

lcy-seso

committed

f647a50

Navigation Menu

Repository navigation

[Docs] State the compile boundary in the manifest guide as the entry derives it - #74

[Docs] State the compile boundary in the manifest guide as the entry derives it#74
lcy-seso merged 9 commits into
tile-ai:mainfrom
lcy-seso:docs/manifest-compile-boundary

Commits on Oct 17, 2026

[Docs] State the compile boundary in the manifest guide as the entry derives it

[Docs] State a target kernel's request_tune and config, register_device_facts, and what a non-CUDA backend's tests and reports cover

[Docs] Point the dev image at ghcr.io/tile-ai/tileops-dev

[Docs] Pass a device facts reader the whole device, and scope the per-device tuning and benchmark statements

[Docs] Put one rule in each cell of the backend contract, and state reads, config and timing as the code does

[Docs] State the three markers a backend's test run reads, and that its target serves the device

[Docs] Align the op, dispatch and manifest guides with the call device and target tuning on main

[Docs] Scope the compile boundary, value checks and CUPTI timing statements to where main holds them

[Docs] Say up front that the timing page covers CUDA