You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@tvm.apache.org by gi...@apache.org on 2024/03/17 05:02:52 UTC

(tvm) branch nightly updated (feb104393c -> 1c73491669)

This is an automated email from the ASF dual-hosted git repository.

github-bot pushed a change to branch nightly
in repository https://gitbox.apache.org/repos/asf/tvm.git


    from feb104393c [TIR][CUDA] Add native FP8 support to codegen (#16548)
     add e57ab7a9dc [Fix] Introduce TVM_DEBUG_WITH_ABI_CHANGE to warn ABI changes in debug mode (#16728)
     add 174f46ea1d Fix cpp_rtvm cmake build on Windows (#16724)
     add b8f64c21c5 [Builtin] Sliding window and sink support for PagedKVCache (#16729)
     add 1c73491669 [Dlight] Fix GeMV shared memory estimation (#16731)

No new revisions were added by this update.

Summary of changes:
 CMakeLists.txt                                     |  12 +
 apps/cpp_rtvm/CMakeLists.txt                       |   4 +-
 apps/cpp_rtvm/main.cc                              |   2 +-
 cmake/config.cmake                                 |   3 +
 cmake/modules/LibInfo.cmake                        |   1 +
 include/tvm/runtime/container/map.h                |  20 +-
 python/tvm/dlight/gpu/gemv.py                      |  18 +-
 src/runtime/relax_vm/kv_state.cc                   |  11 +-
 src/runtime/relax_vm/kv_state.h                    |  40 +-
 src/runtime/relax_vm/paged_kv_cache.cc             | 626 +++++++++++++--------
 src/support/libinfo.cc                             |   5 +
 tests/cpp/container_test.cc                        |   4 +-
 ..._builtin_paged_attention_kv_cache_flashinfer.py | 110 ++--
 ...runtime_builtin_paged_attention_kv_cache_tir.py | 557 +++++++++++-------
 14 files changed, 853 insertions(+), 560 deletions(-)