Lei Wang bdf436d4c3 [Feature] Support TMA store in T.tma_copy() (#1981) ...

* [Feature] Support TMA store in T.tma_copy() with user-managed synchronization

T.tma_copy() previously only supported TMA loads (global -> shared) with
user-managed barrier synchronization. This extends it to also support TMA
stores (shared -> global), where:

- The barrier parameter is now optional (not needed for stores)
- For stores, only tma_store + tma_store_arrive is emitted (no wait)
- Users call T.tma_store_wait() explicitly to synchronize

* Refine tma_store_wait to accept count parameter and clean up builtin signatures

- Wire tma_store_wait count parameter through the full stack (Python -> C++
  builtin registration -> CUDA/CuTeDSL codegen)
- Replace vague *args with proper signatures for fence_proxy_async() and
  tma_store_arrive() (both take no arguments)
- Improve docstrings for internal TMA APIs (create_tma_descriptor, tma_load,
  tma_load_2sm)

2026-03-27 15:29:48 +08:00

..

amd

[AMD] Fix gfx950 ci and add 16x16x32_bf16/fp16 instructions support (#1878)

2026-02-25 19:46:46 +08:00

analysis

[Refactor] Improve cp.async lowering and add async_copy op (#1887)

2026-03-06 17:35:16 +08:00

arith

[Refactor] Unify @jit and @lazy_jit into a single @jit decorator (#1632)

2026-01-09 15:41:39 +08:00

autotune

[Enhancement] Add eager-mode support for tilelang.autotune (#1906)

2026-03-12 20:16:55 +08:00

cache

test: reduce CI runtime for slow Python suites (#1932)

2026-03-18 13:08:30 +08:00

carver

Refactor CUDA version checks for compute 9.0 (#1893)

2026-03-03 18:28:11 +08:00

components

[Bugfix] Open Rocm ci test and fix some bugs. (#1443)

2026-01-06 02:18:29 +08:00

cpu

[Feature] Add T.gemm support for CPU target (#1904)

2026-03-10 18:02:40 +08:00

cuda

Unified packed x2 intrinsics with multi-dtype support and bug fixes (#1978)

2026-03-26 23:45:26 +08:00

debug

test: reduce CI runtime for slow Python suites (#1932)

2026-03-18 13:08:30 +08:00

fastmath

[Enhancement] Update examples and tests for improved type handling functionality (#1448)

2025-12-17 14:41:25 +08:00

issue

[Feature] Add Producer-Consumer Warp Specialization and T.tma_copy() API (#1909)

2026-03-18 13:08:10 +08:00

jit

[Bugfix] Fix double buffer versioning when TMA is used without warp specialization (#1962)

2026-03-24 14:21:57 +08:00

kernel

[Refactor] Separate gemm into explicit wgmma_gemm and tcgen05_gemm functions (#1949)

2026-03-20 14:54:43 +08:00

language

[Feature] Support TMA store in T.tma_copy() (#1981)

2026-03-27 15:29:48 +08:00

layout

[Feature] Introduce higher-dimensional gemm layout support (#1798)

2026-02-25 15:50:58 +08:00

math

test: reduce CI runtime for slow Python suites (#1932)

2026-03-18 13:08:30 +08:00

metal

[Codegen] Metal codegen on Linux (#1857)

2026-02-19 11:19:17 +08:00

profiler

test: reduce CI runtime for slow Python suites (#1932)

2026-03-18 13:08:30 +08:00

runtime

[Runtime] Improve TMA descriptor diagnostics (#1931)

2026-03-13 16:15:02 +08:00

target

[Refactor] Improve cp.async lowering and add async_copy op (#1887)

2026-03-06 17:35:16 +08:00

tilelibrary

test: reduce CI runtime for slow Python suites (#1932)

2026-03-18 13:08:30 +08:00

transform

Fix wrapped pre-loop TMA prefixes in producer-consumer WS (#1975)

2026-03-26 12:21:46 +08:00

utils

[Lint] Phaseout Yapf format and embrace ruff format (#1417)

2025-12-12 17:22:18 +08:00

webgpu

[Enhancement] Update examples and tests for improved type handling functionality (#1448)

2025-12-17 14:41:25 +08:00