Repository navigation
Docs: TypedPython commits to GPU kernels and CPU SIMD (#185) - #186
Merged
Merged
Conversation
…el backend decision (#185) INTENT 1.5 promotes kernels from exploration to commitment. Design 5 now says: tile-first with SIMT fallback, our own IR shaped to map onto MLIR, emitters per target with MSL emitted directly, the MLIR wheel as an optional build-time helper, torchnative's device layer as the runtime and wgpu for the web only. The two investigations behind it are recorded under docs/investigations.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #185
Docs only.
docs/INTENT.md§1.5: GPU kernels and CPU SIMD move from exploration to commitment (maintainer, 2026-10-04).docs/design/typedpython.md§5 rewritten: AOT; tile-first with SIMT fallback; CPU SIMD in@compiled; our own IR shaped to map onto MLIR; MLIR wheel as an optional build-time helper (no 3.15t wheel yet); emitters SPIR-V / MSL (direct) / WGSL / PTX / C; runtime is torchnative's device layer, wgpu web only; landing as torch ops orkernelsbundle entries; graph-capture fusion deferred. §1, tier table, §6 steps, §7 risks and §8 item 4 updated to match.docs/investigations/typedpython-kernel-backend.mdandtypedpython-kernel-ir.md: the two investigations, with [V]/[R]/[I] tags. Time estimates in them are not measurements.