In et_generator, change 'comm_size' attr from uint64 to int64

[jinsun-yoo]

Summary

This PR makes et_generator use int64, instead of uint64 for 'comm_size' when creating collective nodes.
et_feeder parses comm_size as a int64 datatype(et_feeder.cpp:28). With the current strong datatype conversion, using a uint64 value would make attr.int64_val() parse this value as 0 (appendix 1)

Also added static_cast into uint64_t for comm_size, which #18 seems to have missed.
(@TaekyungHeo , please confirm if the lack of static_cast was intentional)

Test Plan

Created & ran the following test locally, to verify that et_generator using uint64 causes comm_size to be parsed as 0.
I did not add this test into the PR, as this is only used to verify the problem statement, not any of our code.

Generate a test ET using the following code

comm_size = 212480
output_filename = "chakra.1.et"
with open(output_filename, "wb") as et:
    encode_message(et, GlobalMetadata(version="0.0.4"))

    node = get_node("node_with_int64", COMM_COLL_NODE)
    node.attr.append(ChakraAttr(name="comm_size", int64_val=comm_size))
    encode_message(et, node)

    node = get_node("node_with_uint64", COMM_COLL_NODE)
    node.attr.append(ChakraAttr(name="comm_size", uint64_val=comm_size))
    encode_message(et, node)

To generate the following trace

{
  "version": "0.0.4"
}{
  "name": "node_with_int64",
  "type": "COMM_COLL_NODE",
  "attr": [
    {
      "name": "comm_size",
      "int64Val": "212480"
    }
  ]
}{
  "id": "1",
  "name": "node_with_uint64",
  "type": "COMM_COLL_NODE",
  "attr": [
    {
      "name": "comm_size",
      "uint64Val": "212480"
    }
  ]
}

To be tested with the following test

TEST_F(ETFeederTest, ShowUint64DoesNotWorkTest) {
  SetUp("tests/data/chakra.1.et");
  std::shared_ptr<Chakra::ETFeederNode> node;

  node = trace->lookupNode(0);
  uint64_t comm_size = node->comm_size();
  ASSERT_EQ(comm_size, 212480);

  node = trace->lookupNode(1);
  comm_size = node->comm_size();
  ASSERT_EQ(comm_size, 0);
}

Additional Notes

Appendix 1)
From et_def.pb.h:

inline ::google::protobuf::int64 AttributeProto::int64_val() const {
  // @@protoc_insertion_point(field_get:ChakraProtoMsg.AttributeProto.int64_val)
  if (has_int64_val()) {
    return value_.int64_val_;
  }
  return GOOGLE_LONGLONG(0);
}

[github-actions]

MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅

[jinsun-yoo]

Recheck

[jinsun-yoo]

recheck

[srinivas212]

@jinsun-yoo - thanks for the PR. you need to sign the CLA before we can proceed. Can you try the following:

• Join MLC Chakra WG
• If that does not fix it - contact [email protected]

[jinsun-yoo]

Hi @srinivas212 , @mlcommons/wg-chakra , Please have another look at this PR. Thanks!

[tushar-krishna]

The change looks good!