This is an automated email from the ASF dual-hosted git repository.
andrewzhaoluo pushed a change to branch ci-docker-staging
in repository https://gitbox.apache.org/repos/asf/tvm.git.
discard c8be9ec jostle ci
discard 55d3a42 Merge branch 'ci-docker-staging' of github.com:apache/tvm
into ci-docker-staging
discard 2a8325f turn on ethosu build for qemu
discard 8d8cf31 upgrade to latest images
discard 9849fab fix missing - with _
discard 4d7109b swtich to proper repo for docker
discard 6784d6b formatting
discard ee78852 jenkinsfile and one test
discard 249da38 upgrade to latest images
discard 3ee8610 jostle
discard 0f698d3 fix missing - with _
discard cc84556 swtich to proper repo for docker
discard 0bd0c0f formatting
discard 51369c5 jenkinsfile and one test
add 670de9b [CI] Fix pip cache config bug (#9933)
add 79c59fe dynamic to static use infer_type_local (#9869)
add 89ae603 [USMP] Hill Climb allocator (#9704)
add 4419241 [TEST] Remove `llvm -device=arm_cpu` and `cuda -libs=cudnn`
from (#9905)
add e7c8141 [Relay/Frontend][TFLite] Change the output shape calculation
based on keep_dim option in fully connected (#9840)
add 6f6fc68 [TIR] Encode conditional accesses info into block read/write
regions (#9880)
add b3c6625 [Int8] Support cublas on e2e int8 models (also tried cudnn
but doesn't work) (#9898)
add 1b1cfb3 remove clang compile warnings (#9942)
add 84ee90c [ONNX] Fix onnx convtranspose error (#9938)
add 6eb4ed8 [Fix] relay onnx frontend bug when [A, B, M, N] * [1, B, N,
K] (#9911)
add be0677d [Caffe Frontend] supporting group > 1 cases for Deconv op
(#8260)
add 3c8de42 [Caffe Frontend] extending Eltwise to handle multiple inputs
(#8136)
add 596333b [MetaSchedule] Schedule Rule: Auto Inline (#9943)
add 24bccd2 [microNPU] Remove remaining UnsupportedLayout checks (#9791)
add 133bb9c [microNPU][2c] Add performance modelling to cascader (#9778)
add 77c66f0 [MetaSchedule] random compute location (#9940)
add 1e5373f [MetaSchedule] PostProcessor: Verify GPU Code (#9945)
add 211291f [CUDNN] Refactor descriptor initialization, remove
`cudnn.conv.output_shape_from_cudnn` (#9948)
add 364e2db [microNPU] Add support for scalar values (#9794)
add 31de5bc [HotFix] Skip the flaky MetaSchedule Auto-Unroll test (#9956)
add 4f29562 Enable NPU and CMSIS in ci_qemu (#9957)
new 10b74f6 jenkinsfile and one test
new 9eb9f7c formatting
new feb159d swtich to proper repo for docker
new 8b31457 fix missing - with _
new fb53dc3 jostle
new 5f57cc7 upgrade to latest images
new a5ee1bf jenkinsfile and one test
new 02bec52 formatting
new c69bfb9 swtich to proper repo for docker
new 7f3476f fix missing - with _
new 77ed92d upgrade to latest images
new dee6b35 jostle ci
new 07f9030 update with official images
This update added new revisions after undoing existing revisions.
That is to say, some revisions that were in the old version of the
branch are not in the new version. This situation occurs
when a user --force pushes a change and generates a repository
containing something like this:
* -- * -- B -- O -- O -- O (c8be9ec)
\
N -- N -- N refs/heads/ci-docker-staging (07f9030)
You should already have received notification emails for all of the O
revisions, and so the following emails describe only the N revisions
from the common base, B.
Any revisions marked "omit" are not gone; other references still
refer to them. Any revisions marked "discard" are gone forever.
The 13 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "add" were already present in the repository and have only
been added to this reference.
Summary of changes:
Jenkinsfile | 4 +-
docker/Dockerfile.ci_arm | 2 +-
docker/Dockerfile.ci_cpu | 2 +-
docker/Dockerfile.ci_gpu | 2 +-
docker/Dockerfile.ci_i386 | 2 +-
docker/Dockerfile.ci_lint | 2 +-
docker/Dockerfile.ci_qemu | 2 +-
docker/Dockerfile.ci_wasm | 2 +-
include/tvm/meta_schedule/schedule_rule.h | 2 -
include/tvm/tir/schedule/schedule.h | 8 +
include/tvm/tir/stmt.h | 7 +
include/tvm/tir/usmp/algo/greedy.h | 85 +++
python/tvm/contrib/cudnn.py | 62 --
python/tvm/contrib/ethosu/cascader/__init__.py | 3 +
.../ethosu/cascader/block_config.py} | 32 +-
.../tvm/contrib/ethosu/cascader/device_config.py | 661 +++++++++++++++++++++
python/tvm/contrib/ethosu/cascader/graph.py | 19 +-
python/tvm/contrib/ethosu/cascader/parts.py | 13 +-
python/tvm/meta_schedule/postproc/__init__.py | 1 +
.../verify_gpu_code.py} | 15 +-
python/tvm/meta_schedule/schedule_rule/__init__.py | 2 +
.../tvm/meta_schedule/schedule_rule/auto_inline.py | 67 +++
.../random_compute_location.py} | 13 +-
python/tvm/meta_schedule/testing/schedule_rule.py | 47 ++
.../tvm/meta_schedule/testing/space_generation.py | 29 +-
.../tvm/relay/backend/contrib/ethosu/legalize.py | 20 +-
.../relay/backend/contrib/ethosu/te/convolution.py | 55 +-
.../tvm/relay/backend/contrib/ethosu/te/inline.py | 5 +-
.../relay/backend/contrib/ethosu/tir/compiler.py | 11 +-
.../relay/backend/contrib/ethosu/tir/scheduler.py | 5 +-
python/tvm/relay/frontend/caffe.py | 88 ++-
python/tvm/relay/frontend/onnx.py | 72 ++-
python/tvm/relay/frontend/tflite.py | 8 +
python/tvm/testing/utils.py | 2 -
python/tvm/tir/schedule/schedule.py | 26 +
python/tvm/topi/cuda/dense.py | 5 +-
src/contrib/ethosu/cascader/block_config.cc | 60 ++
src/contrib/ethosu/cascader/block_config.h | 93 +++
src/contrib/ethosu/cascader/common.h | 26 +
src/contrib/ethosu/cascader/graph.cc | 13 +-
src/contrib/ethosu/cascader/graph.h | 18 +-
src/contrib/ethosu/cascader/parts/ethosu.cc | 142 +++--
src/contrib/ethosu/cascader/parts/ethosu.h | 34 +-
src/contrib/ethosu/cascader/parts/inline.cc | 4 +-
src/contrib/ethosu/cascader/parts/inline.h | 2 +-
src/meta_schedule/postproc/verify_gpu_code.cc | 130 ++++
src/meta_schedule/schedule_rule/auto_inline.cc | 174 ++++++
.../schedule_rule/random_compute_location.cc | 123 ++++
src/relay/ir/indexed_graph.cc | 2 +
src/relay/qnn/op/convolution.cc | 54 +-
src/relay/transforms/dynamic_to_static.cc | 14 +-
src/runtime/contrib/cudnn/conv_forward.cc | 228 +------
src/runtime/contrib/cudnn/cudnn_utils.cc | 95 +++
src/runtime/contrib/cudnn/cudnn_utils.h | 9 +-
src/tir/analysis/block_access_region_detector.cc | 29 +-
src/tir/analysis/device_constraint_utils.cc | 16 +-
src/tir/schedule/analysis.h | 78 +++
src/tir/schedule/analysis/analysis.cc | 285 +++++++++
src/tir/schedule/concrete_schedule.cc | 9 +
src/tir/schedule/concrete_schedule.h | 2 +
src/tir/schedule/primitive.h | 11 +
src/tir/schedule/primitive/block_annotate.cc | 6 +-
src/tir/schedule/primitive/sampling.cc | 64 ++
src/tir/schedule/schedule.cc | 2 +
src/tir/schedule/traced_schedule.cc | 14 +
src/tir/schedule/traced_schedule.h | 1 +
src/tir/schedule/utils.h | 40 ++
src/tir/transforms/compact_buffer_region.cc | 10 +-
src/tir/transforms/ir_utils.cc | 62 +-
src/tir/transforms/ir_utils.h | 18 +-
src/tir/transforms/unify_thread_binding.cc | 2 +-
src/tir/usmp/algo/greedy.cc | 159 +++--
src/tir/usmp/algo/hill_climb.cc | 339 +++++++++++
tests/python/contrib/test_cudnn.py | 10 +-
tests/python/contrib/test_ethosu/cascader/infra.py | 66 ++
.../cascader/test_ethosu_block_config.py | 262 ++++++++
.../cascader/test_ethosu_conv2d_matcher.py | 71 +--
.../cascader/test_ethosu_inline_matcher.py | 3 +-
.../test_ethosu/cascader/test_ethosu_part.py | 28 +-
.../cascader/test_ethosu_part_performance.py | 200 +++++++
.../contrib/test_ethosu/cascader/test_graph.py | 5 +-
tests/python/contrib/test_ethosu/test_codegen.py | 5 +-
tests/python/contrib/test_ethosu/test_compiler.py | 3 +-
tests/python/frontend/caffe/test_forward.py | 69 +++
tests/python/frontend/onnx/test_forward.py | 10 +
tests/python/frontend/pytorch/qnn_test.py | 15 +-
tests/python/frontend/tflite/test_forward.py | 1 +
tests/python/relay/test_any.py | 2 +-
.../test_meta_schedule_postproc_verify_gpu_code.py | 231 +++++++
...test_meta_schedule_schedule_rule_auto_inline.py | 302 ++++++++++
...hedule_schedule_rule_random_compute_location.py | 93 +++
.../test_tir_analysis_get_block_access_region.py | 66 ++
.../python/unittest/test_tir_schedule_sampling.py | 77 +++
.../test_tir_transform_compact_buffer_region.py | 1 +
tests/python/unittest/test_tir_usmp_algo.py | 21 +-
.../unittest/test_tir_usmp_algo_hill_climb.py | 397 +++++++++++++
tests/scripts/task_config_build_qemu.sh | 1 +
97 files changed, 4967 insertions(+), 721 deletions(-)
create mode 100644 include/tvm/tir/usmp/algo/greedy.h
copy python/tvm/{target/virtual_device.py =>
contrib/ethosu/cascader/block_config.py} (53%)
create mode 100644 python/tvm/contrib/ethosu/cascader/device_config.py
copy python/tvm/meta_schedule/{measure_callback/update_cost_model.py =>
postproc/verify_gpu_code.py} (70%)
create mode 100644 python/tvm/meta_schedule/schedule_rule/auto_inline.py
copy python/tvm/meta_schedule/{measure_callback/add_to_database.py =>
schedule_rule/random_compute_location.py} (70%)
create mode 100644 python/tvm/meta_schedule/testing/schedule_rule.py
copy apps/dso_plugin_module/Makefile =>
python/tvm/meta_schedule/testing/space_generation.py (53%)
create mode 100644 src/contrib/ethosu/cascader/block_config.cc
create mode 100644 src/contrib/ethosu/cascader/block_config.h
create mode 100644 src/meta_schedule/postproc/verify_gpu_code.cc
create mode 100644 src/meta_schedule/schedule_rule/auto_inline.cc
create mode 100644 src/meta_schedule/schedule_rule/random_compute_location.cc
create mode 100644 src/tir/usmp/algo/hill_climb.cc
create mode 100644
tests/python/contrib/test_ethosu/cascader/test_ethosu_block_config.py
create mode 100644
tests/python/contrib/test_ethosu/cascader/test_ethosu_part_performance.py
create mode 100644
tests/python/unittest/test_meta_schedule_postproc_verify_gpu_code.py
create mode 100644
tests/python/unittest/test_meta_schedule_schedule_rule_auto_inline.py
create mode 100644
tests/python/unittest/test_meta_schedule_schedule_rule_random_compute_location.py
create mode 100644 tests/python/unittest/test_tir_usmp_algo_hill_climb.py
