Merge pull request #393 from NVIDIA/am/tdef-in-scenario

Control workload settings from scenario

Author: amaslenn

conf/common/test_scenario/nccl_tests_inline.toml

@@ -0,0 +1,51 @@
+# SPDX-FileCopyrightText: NVIDIA CORPORATION & AFFILIATES
+# Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name = "nccl-tests"
+
+[[Tests]]
+id = "nccl.all_reduce"
+num_nodes = 2
+time_limit = "00:20:00"
+
+name = "nccl-all_reduce"
+description = "all_reduce"
+test_template_name = "NcclTest"
+
+  [Tests.cmd_args]
+  subtest_name = "all_reduce_perf_mpi"
+  ngpus = 1
+  minbytes = "128"
+  maxbytes = "512"
+  iters = 50
+  warmup_iters = 10
+
+[[Tests]]
+id = "nccl.all_gather"
+num_nodes = 2
+time_limit = "00:20:00"
+
+name = "nccl-all_gather"
+description = "all_gather"
+test_template_name = "NcclTest"
+
+  [Tests.cmd_args]
+  subtest_name = "all_gather_perf_mpi"
+  ngpus = 1
+  minbytes = "128"
+  maxbytes = "512"
+  iters = 50
+  warmup_iters = 10

conf/experimental/test/nemo_launcher_nemotron_15b_bf16_16_node.toml

@@ -23,25 +23,40 @@ num_nodes = "2"
 
 [[Tests]]
 id = "nemo_launcher_nemotron_15b_bf16_4_node"
-test_name = "nemo_launcher_nemotron_15b_bf16_4_node"
+test_name = "nemo_launcher_nemotron_15b_bf16_2_node"
 num_nodes = "4"
 
+  [Tests.cmd_args.training.model]
+  global_batch_size = 128
+
 [[Tests]]
 id = "nemo_launcher_nemotron_15b_bf16_8_node"
-test_name = "nemo_launcher_nemotron_15b_bf16_8_node"
+test_name = "nemo_launcher_nemotron_15b_bf16_2_node"
 num_nodes = "8"
 
+  [Tests.cmd_args.training.model]
+  global_batch_size = 256
+
 [[Tests]]
 id = "nemo_launcher_nemotron_15b_bf16_16_node"
-test_name = "nemo_launcher_nemotron_15b_bf16_16_node"
+test_name = "nemo_launcher_nemotron_15b_bf16_2_node"
 num_nodes = "16"
 
+  [Tests.cmd_args.training.model]
+  global_batch_size = 512
+
 [[Tests]]
 id = "nemo_launcher_nemotron_15b_bf16_32_node"
-test_name = "nemo_launcher_nemotron_15b_bf16_32_node"
+test_name = "nemo_launcher_nemotron_15b_bf16_2_node"
 num_nodes = "32"
 
+  [Tests.cmd_args.training.model]
+  global_batch_size = 1024
+
 [[Tests]]
 id = "nemo_launcher_nemotron_15b_bf16_64_node"
-test_name = "nemo_launcher_nemotron_15b_bf16_64_node"
+test_name = "nemo_launcher_nemotron_15b_bf16_2_node"
 num_nodes = "64"
+
+  [Tests.cmd_args.training.model]
+  global_batch_size = 2048

conf/experimental/test/nemo_launcher_nemotron_15b_bf16_32_node.toml

@@ -18,5 +18,8 @@ name = "nemo_launcher_nemotron_15b_bf16_16_node"
 
 [[Tests]]
 id = "nemo_launcher_nemotron_15b_bf16_16_node"
-test_name = "nemo_launcher_nemotron_15b_bf16_16_node"
+test_name = "nemo_launcher_nemotron_15b_bf16_2_node"
 num_nodes = "16"
+
+  [Tests.cmd_args.training.model]
+  global_batch_size = 512

conf/experimental/test/nemo_launcher_nemotron_15b_bf16_4_node.toml

@@ -18,5 +18,8 @@ name = "nemo_launcher_nemotron_15b_bf16_32_node"
 
 [[Tests]]
 id = "nemo_launcher_nemotron_15b_bf16_32_node"
-test_name = "nemo_launcher_nemotron_15b_bf16_32_node"
+test_name = "nemo_launcher_nemotron_15b_bf16_2_node"
 num_nodes = "32"
+
+  [Tests.cmd_args.training.model]
+  global_batch_size = 1024