Merge branch 'main' into aluo/onnx/sceloss

* main:
  [UnitTests][Contrib] Enable contrib tensorrt/coreml unit tests (apache#8902)
  [BUG] DataType Bug In SplitRel (apache#8899)
  Enable python debug runtime for exported network libraries (apache#8793)
  Set default value of p in LpPool as 2 (apache#8866)
  [Community] @Hzfengsy -> Committer (apache#8908)
  Trivial uTVM -> microTVM "spelling" fix to align with branding. (apache#8905)
  [Vulkan][Topi] Parametrizing additional topi tests, marking vulkan failures (apache#8904)
  Move to new style issue template system (apache#8898)
  [Onnx] Support Negative Log Loss (apache#8872)
  [ROCm][TVMC] Add ROCm to the TVMC driver (apache#8896)
  fix error report on Store (apache#8895)
  [Docker] Re-enabled automatic --tty flag when running bash. (apache#8861)

.github/ISSUE_TEMPLATE/bug-report.md

@@ -0,0 +1,27 @@
+---
+name: "\U0001F41B Bug report"
+about: To help the developer act on the issues, please include a description of your environment, preferably a minimum script to reproduce the problem.
+title: "[Bug] "
+labels: "type: bug"
+
+---
+
+Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking.  You are always welcomed to post on the forum first :smile_cat:
+
+Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed.
+
+### Expected behavior
+
+What you were expecting
+
+### Actual behavior
+
+What actually happened
+
+### Environment
+
+Any environment details, such as: Operating System, TVM version, etc
+
+### Steps to reproduce
+
+Preferably a minimal script to cause the issue to occur.

.github/ISSUE_TEMPLATE/ci-image.md

@@ -0,0 +1,29 @@
+---
+name: "\U0001F40B Update CI Docker Image"
+about: Provide information on CI Docker Images requiring updates
+title: "[CI Image] "
+
+---
+
+Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking.  You are always welcomed to post on the forum first :smile_cat:
+
+Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed.
+
+- [ ] S0. Reason: For example, a blocked PR or a feature issue
+
+- [ ] S1. Tag of nightly build: TAG. Docker hub: https://hub.docker.com/layers/tlcpackstaging/ci_cpu/...
+
+- [ ] S2. The nightly is built on TVM commit: TVM_COMMIT. Detailed info can be found here: https://ci.tlcpack.ai/blue/organizations/jenkins/docker-images-ci%2Fdaily-docker-image-rebuild/detail/daily-docker-image-rebuild/....
+
+- [ ] S3. Testing the nightly image on ci-docker-staging: https://ci.tlcpack.ai/blue/organizations/jenkins/tvm/detail/ci-docker-staging/...
+
+- [ ] S4. Retag TAG to VERSION:
+```
+docker pull tlcpackstaging/IMAGE_NAME:TAG
+docker tag tlcpackstaging/IMAGE_NAME:TAG tlcpack/IMAGE_NAME:VERSION
+docker push tlcpack/IMAGE_NAME:VERSION
+```
+
+- [ ] S5. Check if the new tag is really there: https://hub.docker.com/u/tlcpack
+
+- [ ] S6. Submit a PR updating the IMAGE_NAME version on Jenkins

.github/ISSUE_TEMPLATE/ci-problem.md

@@ -0,0 +1,22 @@
+---
+name: "\U0000274C CI Problem"
+about: To help the developers act on these problems, please give us as many details of the CI failure as possible.
+title: "[CI Problem] "
+
+---
+
+Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking.  You are always welcomed to post on the forum first :smile_cat:
+
+Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed.
+
+### Branch/PR Failing
+
+Please provide a link to the PR that has failed to run CI.
+
+### Jenkins Link
+
+Provide a link to the specific run that has failed.
+
+### Flakiness
+
+Have you seen this multiple times in this branch or in other branches?

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,5 @@
+blank_issues_enabled: false  # default: true
+contact_links:
+- name: 💬 Discourse
+  url: https://discuss.tvm.apache.org/
+  about: Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking.  You are always welcomed to post on the forum first 😺

.github/ISSUE_TEMPLATE.md → .github/ISSUE_TEMPLATE/feature-tracking.md

@@ -1,7 +1,14 @@
-Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking.  You are always welcomed to post on the forum first :)
+---
+name: "\U0001F527 Feature Tracking"
+about: List clear, small actionable items so we can track the progress of the change.
+title: "[Tracking Issue] "
+labels: type:rfc-tracking
 
-Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed.
+---
+
+Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking.  You are always welcomed to post on the forum first :smile_cat:
 
-For bug reports, to help the developer act on the issues, please include a description of your environment, preferably a minimum script to reproduce the problem.
+Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed.
 
-For feature proposals, list clear, small actionable items so we can track the progress of the change.
+### This issue is to track progress for FEATURE NAME
+- [ ] P1. Title of this piece of the feature (PR link if available)

CONTRIBUTORS.md

@@ -35,6 +35,7 @@ We do encourage everyone to work anything they are interested in.
 - [Tianqi Chen](https://github.com/tqchen) (PMC): @tqchen - topi, compiler, relay, docs
 - [Wei Chen](https://github.com/wweic): @wweic - runtime, relay, vm
 - [Zhi Chen](https://github.com/zhiics) (PMC): @zhiics - relay, quantization, pass manager
+- [Siyuan Feng](https://github.com/Hzfengsy): @Hzfengsy - tir
 - [Josh Fromm](https://github.com/jwfromm): @jwfromm - frontends, quantization, topi
 - [Yuwei Hu](https://github.com/Huyuwei): @Huyuwei - topi, frontends
 - [Nick Hynes](https://github.com/nhynes): @nhynes: - sgx, rust

cmake/config.cmake

@@ -108,7 +108,7 @@ set(USE_GRAPH_EXECUTOR_CUDA_GRAPH OFF)
 # Whether to enable the profiler for the graph executor and vm
 set(USE_PROFILER ON)
 
-# Whether enable uTVM standalone runtime
+# Whether enable microTVM standalone runtime
 set(USE_MICRO_STANDALONE_RUNTIME OFF)
 
 # Whether build with LLVM support

docker/bash.sh

@@ -38,7 +38,7 @@ set -euo pipefail
 
 function show_usage() {
     cat <<EOF
-Usage: docker/bash.sh [-i|--interactive] [--net=host]
+Usage: docker/bash.sh [-i|--interactive] [--net=host] [-t|--tty]
          [--mount MOUNT_DIR] [--repo-mount-point REPO_MOUNT_POINT]
          [--dry-run]
          <DOCKER_IMAGE_NAME> [--] [COMMAND]
@@ -95,7 +95,7 @@ DOCKER_IMAGE_NAME
 COMMAND
 
     The command to be run inside the docker container.  If this is set
-    to "bash", both the --interactive and --net=host flags are set.
+    to "bash", the --interactive, --tty and --net=host flags are set.
     If no command is specified, defaults to "bash".  If the command
     contains dash-prefixed arguments, the command should be preceded
     by -- to indicate arguments that are not intended for bash.sh.
@@ -235,6 +235,7 @@ fi
 
 if [[ ${COMMAND[@]+"${COMMAND[@]}"} = bash ]]; then
     INTERACTIVE=true
+    TTY=true
     USE_NET_HOST=true
 fi
 

docs/dev/debugger.rst

@@ -123,12 +123,12 @@ Example of loading the parameters
 How to use Debugger?
 ***************************************
 
-1. In ``config.cmake`` set the ``USE_GRAPH_EXECUTOR_DEBUG`` flag to ``ON``
+1. In ``config.cmake`` set the ``USE_PROFILER`` flag to ``ON``
 
    ::
 
        # Whether enable additional graph debug functions
-       set(USE_GRAPH_EXECUTOR_DEBUG ON)
+       set(USE_PROFILER ON)
 
 2. Do 'make' tvm, so that it will make the ``libtvm_runtime.so``
 
@@ -148,6 +148,22 @@ How to use Debugger?
     m.run()
     tvm_out = m.get_output(0, tvm.nd.empty(out_shape, dtype)).numpy()
 
+4. If network previously was exported to external libray using ``lib.export_library("network.so")``
+     like shared object file/dynamic linked library, the initialization
+     of debug runtime will be slightly different
+
+::
+
+    lib = tvm.runtime.load_module("network.so")
+    m = graph_executor.create(lib["get_graph_json"](), lib, dev, dump_root="/tmp/tvmdbg")
+    # set inputs
+    m.set_input('data', tvm.nd.array(data.astype(dtype)))
+    m.set_input(**params)
+    # execute
+    m.run()
+    tvm_out = m.get_output(0, tvm.nd.empty(out_shape, dtype)).numpy()
+
+
 The outputs are dumped to a temporary folder in ``/tmp`` folder or the
 folder specified while creating the runtime.
 

python/tvm/driver/tvmc/runner.py

@@ -50,7 +50,7 @@ def add_run_parser(subparsers):
     #      like 'webgpu', etc (@leandron)
     parser.add_argument(
         "--device",
-        choices=["cpu", "cuda", "cl", "metal", "vulkan"],
+        choices=["cpu", "cuda", "cl", "metal", "vulkan", "rocm"],
         default="cpu",
         help="target device to run the compiled module. Defaults to 'cpu'",
     )
@@ -394,6 +394,8 @@ def run_module(
         dev = session.metal()
     elif device == "vulkan":
         dev = session.vulkan()
+    elif device == "rocm":
+        dev = session.rocm()
     else:
         assert device == "cpu"
         dev = session.cpu()

python/tvm/relay/frontend/onnx.py

@@ -37,21 +37,9 @@
 from .. import random as _random
 from .. import ty as _ty
 from .. import vision as _vision
-from .common import (
-    AttrCvt,
-    Renamer,
-    fold_constant,
-    get_name,
-    get_relay_op,
-    gru_cell,
-    infer_channels,
-    infer_shape,
-    infer_type,
-    infer_value,
-    lstm_cell,
-    new_var,
-    unbind,
-)
+from .common import (AttrCvt, Renamer, fold_constant, get_name, get_relay_op,
+                     gru_cell, infer_channels, infer_shape, infer_type,
+                     infer_value, lstm_cell, new_var, unbind)
 
 __all__ = ["from_onnx"]
 
@@ -909,8 +897,9 @@ def _impl_v1(cls, inputs, attr, params):
         else:
             attr["layout"] = onnx_default_layout(dims=(len(input_shape) - 2), op_name="LpPool")
 
-        p = _expr.const(attr["p"], dtype)
-        reci_p = _expr.const(1.0 / attr["p"], dtype)
+        p_value = attr.get("p", 2)
+        p = _expr.const(p_value, dtype)
+        reci_p = _expr.const(1.0 / p_value, dtype)
         data = _op.power(data, p)
 
         out = AttrCvt(

python/tvm/rpc/client.py

@@ -217,6 +217,10 @@ def metal(self, dev_id=0):
         """Construct Metal device."""
         return self.device(8, dev_id)
 
+    def rocm(self, dev_id=0):
+        """Construct ROCm device."""
+        return self.device(10, dev_id)
+
     def ext_dev(self, dev_id=0):
         """Construct extension device."""
         return self.device(12, dev_id)

python/tvm/script/parser.py

@@ -536,7 +536,7 @@ def transform_SubscriptAssign(self, node):
             if len(indexes) != 1:
                 self.report_error(
                     f"Store is only allowed with one index, but {len(indexes)} were provided.",
-                    tvm.ir.Span.union([x.span for x in indexes]),
+                    node.params[1].span,
                 )
             # Store
             return tvm.tir.Store(

python/tvm/testing/__init__.py

@@ -24,6 +24,7 @@
 from .utils import known_failing_targets, requires_cuda, requires_cudagraph
 from .utils import requires_gpu, requires_llvm, requires_rocm, requires_rpc
 from .utils import requires_tensorcore, requires_metal, requires_micro, requires_opencl
+from .utils import requires_package
 from .utils import identity_after, terminate_self
 
 from ._ffi_api import nop, echo, device_test, run_check_signal, object_use_count

python/tvm/testing/utils.py

@@ -774,7 +774,51 @@ def requires_rpc(*args):
     return _compose(args, _requires_rpc)
 
 
+def requires_package(*packages):
+    """Mark a test as requiring python packages to run.
+
+    If the packages listed are not available, tests marked with
+    `requires_package` will appear in the pytest results as being skipped.
+    This is equivalent to using ``foo = pytest.importorskip('foo')`` inside
+    the test body.
+
+    Parameters
+    ----------
+    packages : List[str]
+
+        The python packages that should be available for the test to
+        run.
+
+    Returns
+    -------
+    mark: pytest mark
+
+        The pytest mark to be applied to unit tests that require this
+
+    """
+
+    def has_package(package):
+        try:
+            __import__(package)
+            return True
+        except ImportError:
+            return False
+
+    marks = [
+        pytest.mark.skipif(not has_package(package), reason=f"Cannot import '{package}'")
+        for package in packages
+    ]
+
+    def wrapper(func):
+        for mark in marks:
+            func = mark(func)
+        return func
+
+    return wrapper
+
+
 def parametrize_targets(*args):
+
     """Parametrize a test over a specific set of targets.
 
     Use this decorator when you want your test to be run over a

src/relay/op/tensor/transform.cc

@@ -2829,15 +2829,18 @@ bool SplitRel(const Array<Type>& types, int num_inputs, const Attrs& attrs,
     }
     reporter->Assign(types[1], TupleType(Array<Type>(fields)));
   } else {
-    auto indices = Downcast<Array<ObjectRef>>(param->indices_or_sections);
+    Array<IndexExpr> indices;
+    for (auto i : Downcast<Array<Integer>>(param->indices_or_sections)) {
+      indices.push_back(IntImm(DataType::Int(32), i.as<IntImmNode>()->value));
+    }
     auto begin = IndexExpr(tir::make_zero(DataType::Int(32)));
     std::vector<Type> fields;
     for (unsigned int i = 0; i < indices.size(); ++i) {
-      ICHECK(reporter->Assert(Downcast<IndexExpr>(indices[i]) > begin))
+      ICHECK(reporter->Assert(indices[i] > begin))
           << "indices_or_sections need to be a sorted ascending list";
       std::vector<IndexExpr> oshape(data->shape.begin(), data->shape.end());
-      oshape[axis] = Downcast<IndexExpr>(indices[i]) - begin;
-      begin = Downcast<IndexExpr>(indices[i]);
+      oshape[axis] = indices[i] - begin;
+      begin = indices[i];
       auto vec_type = TensorType(oshape, data->dtype);
       fields.push_back(vec_type);
     }

src/runtime/graph_executor/graph_executor_factory.cc

@@ -53,6 +53,10 @@ PackedFunc GraphExecutorFactory::GetFunction(
       }
       *rv = this->ExecutorCreate(devices);
     });
+  } else if (name == "get_graph_json") {
+    return PackedFunc(
+        [sptr_to_self, this](TVMArgs args, TVMRetValue* rv) { *rv = this->graph_json_; });
+
   } else if (name == "debug_create") {
     return PackedFunc([sptr_to_self, this](TVMArgs args, TVMRetValue* rv) {
       ICHECK_GE(args.size(), 2);

src/target/spirv/spirv_support.cc

@@ -72,6 +72,9 @@ SPIRVSupport::SPIRVSupport(tvm::Target target) {
   if (target->GetAttr<Bool>("supports_float16")) {
     supports_float16 = target->GetAttr<Bool>("supports_float16").value();
   }
+  if (target->GetAttr<Bool>("supports_float64")) {
+    supports_float64 = target->GetAttr<Bool>("supports_float64").value();
+  }
   if (target->GetAttr<Bool>("supports_int8")) {
     supports_int8 = target->GetAttr<Bool>("supports_int8").value();
   }