From ca1fec6774f2df695def1b05ce67e6b9c810589e Mon Sep 17 00:00:00 2001
From: Matthew Kotila <matthew.r.kotila@gmail.com>
Date: Wed, 31 Jan 2024 17:50:47 -0800
Subject: [PATCH] Remove periodic concurrency feature

---
 src/c++/perf_analyzer/CMakeLists.txt          |   6 +-
 src/c++/perf_analyzer/command_line_parser.cc  | 121 +------------
 src/c++/perf_analyzer/command_line_parser.h   |   9 +-
 src/c++/perf_analyzer/docs/cli.md             |  26 +--
 .../docs/inference_load_modes.md              |  36 +---
 src/c++/perf_analyzer/docs/llm.md             |  79 +--------
 src/c++/perf_analyzer/infer_context.cc        |   9 +-
 src/c++/perf_analyzer/infer_context.h         |  15 +-
 src/c++/perf_analyzer/inference_profiler.h    |  15 +-
 src/c++/perf_analyzer/perf_analyzer.cc        |  14 +-
 .../periodic_concurrency_manager.cc           | 121 -------------
 .../periodic_concurrency_manager.h            |  93 ----------
 .../periodic_concurrency_worker.cc            |  71 --------
 .../periodic_concurrency_worker.h             |  80 ---------
 .../perf_analyzer/test_command_line_parser.cc | 162 +-----------------
 15 files changed, 15 insertions(+), 842 deletions(-)
 delete mode 100644 src/c++/perf_analyzer/periodic_concurrency_manager.cc
 delete mode 100644 src/c++/perf_analyzer/periodic_concurrency_manager.h
 delete mode 100644 src/c++/perf_analyzer/periodic_concurrency_worker.cc
 delete mode 100644 src/c++/perf_analyzer/periodic_concurrency_worker.h

diff --git a/src/c++/perf_analyzer/CMakeLists.txt b/src/c++/perf_analyzer/CMakeLists.txt
index bebdba4d5..f6cd1cb0d 100644
--- a/src/c++/perf_analyzer/CMakeLists.txt
+++ b/src/c++/perf_analyzer/CMakeLists.txt
@@ -1,4 +1,4 @@
-# Copyright 2020-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# Copyright 2020-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
 # modification, are permitted provided that the following conditions
@@ -68,8 +68,6 @@ set(
   sequence_manager.cc
   profile_data_collector.cc
   profile_data_exporter.cc
-  periodic_concurrency_manager.cc
-  periodic_concurrency_worker.cc
 )
 
 set(
@@ -110,8 +108,6 @@ set(
   request_record.h
   profile_data_collector.h
   profile_data_exporter.h
-  periodic_concurrency_manager.h
-  periodic_concurrency_worker.h
 )
 
 add_executable(
diff --git a/src/c++/perf_analyzer/command_line_parser.cc b/src/c++/perf_analyzer/command_line_parser.cc
index 711f1714e..b1fe9a015 100644
--- a/src/c++/perf_analyzer/command_line_parser.cc
+++ b/src/c++/perf_analyzer/command_line_parser.cc
@@ -1,4 +1,4 @@
-// Copyright 2022-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2022-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -113,8 +113,6 @@ CLParser::Usage(const std::string& msg)
   std::cerr << "\t--measurement-interval (-p) <measurement window (in msec)>"
             << std::endl;
   std::cerr << "\t--concurrency-range <start:end:step>" << std::endl;
-  std::cerr << "\t--periodic-concurrency-range <start:end:step>" << std::endl;
-  std::cerr << "\t--request-period <number of responses>" << std::endl;
   std::cerr << "\t--request-rate-range <start:end:step>" << std::endl;
   std::cerr << "\t--request-distribution <\"poisson\"|\"constant\">"
             << std::endl;
@@ -301,34 +299,6 @@ CLParser::Usage(const std::string& msg)
              "not be 0 for sequence models while using asynchronous mode.",
              18)
       << std::endl;
-  std::cerr
-      << FormatMessage(
-             "--periodic-concurrency-range <start:end:step>: Determines the "
-             "range of concurrency levels in the similar but slightly "
-             "different manner as the --concurrency-range. Perf Analyzer will "
-             "start from the concurrency level of 'start' and increase by "
-             "'step' each time. Unlike --concurrency-range, the 'end' "
-             "indicates the *total* number of concurrency since the 'start' "
-             "(including) and will stop increasing once the cumulative number "
-             "of concurrent requests has reached the 'end'. The user can "
-             "specify *when* to periodically increase the concurrency level "
-             "using the --request-period option. The concurrency level will "
-             "periodically increase for every n-th response specified by "
-             "--request-period. Since this disables stability check in Perf "
-             "Analyzer and reports response timestamps only, the user must "
-             "provide --profile-export-file to specify where to dump all the "
-             "measured timestamps. The default values of 'start', 'end', and "
-             "'step' are 1.",
-             18)
-      << std::endl;
-  std::cerr
-      << FormatMessage(
-             "--request-period <n>: Indicates the number of responses that "
-             "each request must receive before new, concurrent requests are "
-             "sent when --periodic-concurrency-range is specified. Default "
-             "value is 10.",
-             18)
-      << std::endl;
   std::cerr
       << FormatMessage(
              "--request-parameter <name:value:type>: Specifies a custom "
@@ -872,9 +842,7 @@ CLParser::ParseCommandLine(int argc, char** argv)
       {"output-tensor-format", required_argument, 0, 56},
       {"version", no_argument, 0, 57},
       {"profile-export-file", required_argument, 0, 58},
-      {"periodic-concurrency-range", required_argument, 0, 59},
-      {"request-period", required_argument, 0, 60},
-      {"request-parameter", required_argument, 0, 61},
+      {"request-parameter", required_argument, 0, 59},
       {0, 0, 0, 0}};
 
   // Parse commandline...
@@ -1538,56 +1506,6 @@ CLParser::ParseCommandLine(int argc, char** argv)
           break;
         }
         case 59: {
-          params_->is_using_periodic_concurrency_mode = true;
-          std::string arg = optarg;
-          std::vector<std::string> values{SplitString(arg)};
-          if (values.size() < 2) {
-            Usage(
-                "Failed to parse --periodic-concurrency-range. Both <start> "
-                "and <end> values must be provided.");
-          } else if (values.size() > 3) {
-            Usage(
-                "Failed to parse --periodic-concurrency-range. The value does "
-                "not match <start:end:step>.");
-          }
-
-          for (size_t i = 0; i < values.size(); ++i) {
-            uint64_t val = std::stoull(values[i]);
-            if (i == 0) {
-              params_->periodic_concurrency_range.start = val;
-            } else if (i == 1) {
-              params_->periodic_concurrency_range.end = val;
-            } else if (i == 2) {
-              params_->periodic_concurrency_range.step = val;
-            }
-          }
-
-          Range<uint64_t> range{params_->periodic_concurrency_range};
-          if (range.step == 0) {
-            Usage(
-                "Failed to parse --periodic-concurrency-range. The <step> "
-                "value must be > 0.");
-          } else if (range.start > range.end) {
-            Usage(
-                "Failed to parse --periodic-concurrency-range. The <start> "
-                "must be <= <end>.");
-          } else if ((range.end - range.start) % range.step != 0) {
-            Usage(
-                "Failed to parse --periodic-concurrency-range. The <step> "
-                "value must be a factor of the range size (<end> - <start>).");
-          }
-          break;
-        }
-        case 60: {
-          std::string request_period{optarg};
-          if (std::stoi(request_period) > 0) {
-            params_->request_period = std::stoull(request_period);
-          } else {
-            Usage("Failed to parse --request-period. The value must be > 0");
-          }
-          break;
-        }
-        case 61: {
           std::string arg = optarg;
           std::vector<std::string> values{SplitString(arg)};
           if (values.size() != 3) {
@@ -1766,46 +1684,13 @@ CLParser::VerifyOptions()
   }
 
   std::vector<bool> load_modes{
-      params_->is_using_periodic_concurrency_mode,
       params_->using_concurrency_range, params_->using_request_rate_range,
       params_->using_custom_intervals};
   if (std::count(load_modes.begin(), load_modes.end(), true) > 1) {
     Usage(
         "Cannot specify more then one inference load mode. Please choose only "
         "one of the following modes: --concurrency-range, "
-        "--periodic-concurrency-range, --request-rate-range, or "
-        "--request-intervals.");
-  }
-
-  if (params_->is_using_periodic_concurrency_mode && !params_->streaming) {
-    Usage(
-        "The --periodic-concurrency-range option requires bi-directional gRPC "
-        "streaming.");
-  }
-
-  if (params_->is_using_periodic_concurrency_mode &&
-      (params_->profile_export_file == "")) {
-    Usage(
-        "Must provide --profile-export-file when using the "
-        "--periodic-concurrency-range option.");
-  }
-
-  if (params_->is_using_periodic_concurrency_mode) {
-    if (params_->periodic_concurrency_range.end == pa::NO_LIMIT) {
-      std::cerr
-          << "WARNING: The maximum attainable concurrency will be limited by "
-             "max_threads specification."
-          << std::endl;
-      params_->periodic_concurrency_range.end = params_->max_threads;
-    } else {
-      if (params_->max_threads_specified) {
-        std::cerr << "WARNING: Overriding max_threads specification to ensure "
-                     "requested concurrency range."
-                  << std::endl;
-      }
-      params_->max_threads = std::max(
-          params_->max_threads, params_->periodic_concurrency_range.end);
-    }
+        "--request-rate-range, or --request-intervals.");
   }
 
   if (params_->request_parameters.size() > 0 &&
diff --git a/src/c++/perf_analyzer/command_line_parser.h b/src/c++/perf_analyzer/command_line_parser.h
index 9ff4869ff..89c309a13 100644
--- a/src/c++/perf_analyzer/command_line_parser.h
+++ b/src/c++/perf_analyzer/command_line_parser.h
@@ -1,4 +1,4 @@
-// Copyright 2022-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2022-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -131,8 +131,7 @@ struct PerfAnalyzerParameters {
   {
     return (
         using_concurrency_range || using_old_options ||
-        !(using_request_rate_range || using_custom_intervals ||
-          is_using_periodic_concurrency_mode));
+        !(using_request_rate_range || using_custom_intervals));
   }
 
   // Sets the threshold for PA client overhead.
@@ -150,10 +149,6 @@ struct PerfAnalyzerParameters {
 
   // The profile export file path.
   std::string profile_export_file{""};
-
-  bool is_using_periodic_concurrency_mode{false};
-  Range<uint64_t> periodic_concurrency_range{1, 1, 1};
-  uint64_t request_period{10};
 };
 
 using PAParamsPtr = std::shared_ptr<PerfAnalyzerParameters>;
diff --git a/src/c++/perf_analyzer/docs/cli.md b/src/c++/perf_analyzer/docs/cli.md
index 5961224c8..b60c02f84 100644
--- a/src/c++/perf_analyzer/docs/cli.md
+++ b/src/c++/perf_analyzer/docs/cli.md
@@ -1,5 +1,5 @@
 <!--
-Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+Copyright 2023-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 
 Redistribution and use in source and binary forms, with or without
 modification, are permitted provided that the following conditions
@@ -180,30 +180,6 @@ until the latency threshold is met. 'end' and `--latency-threshold` cannot
 both be `0`. 'end' cannot be `0` for sequence models while using asynchronous
 mode.
 
-#### `--periodic-concurrency-range=<start:end:step>`
-
-Specifies the range of concurrency levels in the similar but slightly different
-manner as the `--concurrency-range`. Perf Analyzer will start from the
-concurrency level of 'start' and increase by 'step' each time. Unlike
-`--concurrency-range`, the 'end' indicates the *total* number of concurrency
-since the 'start' (including) and will stop increasing once the cumulative
-number of concurrent requests has reached the 'end'. The user can specify
-*when* to periodically increase the concurrency level using the
-`--request-period` option. The concurrency level will periodically increase for
-every `n`-th response specified by `--request-period`. Since this disables
-stability check in Perf Analyzer and reports response timestamps only, the user
-must provide `--profile-export-file` to specify where to dump all the measured
-timestamps.
-
-The default values of 'start', 'end', and 'step' are `1`.
-
-#### `--request-period=<n>`
-
-Specifies the number of responses that each request must receive before new,
-concurrent requests are sent when `--periodic-concurrency-range` is specified.
-
-Default value is `10`.
-
 #### `--request-parameter=<name:value:type>`
 
 Specifies a custom parameter that can be sent to a Triton backend as part of
diff --git a/src/c++/perf_analyzer/docs/inference_load_modes.md b/src/c++/perf_analyzer/docs/inference_load_modes.md
index 83fa83eb0..ecabf9c83 100644
--- a/src/c++/perf_analyzer/docs/inference_load_modes.md
+++ b/src/c++/perf_analyzer/docs/inference_load_modes.md
@@ -1,5 +1,5 @@
 <!--
-Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+Copyright 2023-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 
 Redistribution and use in source and binary forms, with or without
 modification, are permitted provided that the following conditions
@@ -40,40 +40,6 @@ example, when using
 will to attempt to have 4 outgoing inference requests at all times during
 profiling.
 
-## Periodic Concurrency Mode
-
-In periodic concurrency mode, Perf Analyzer will periodically launch a new set
-of inference requests until the total number of inference requests that has been
-launched since the beginning reaches N requests.
-
-For example, when using `--periodic-concurrency-range 10:100:30`, Perf Analyzer
-will start with 10 concurrent requests and for every step, it will launch 30 new
-inference requests until the total number of requests launched since the
-beginning reaches 100. Additionally, the user can also specify *when* to launch
-the new requests by specifying `--request-period M`. This will set Perf Analyzer
-to launch a new set of requests whenever *all* of the latest set of launched
-concurrent requests received M number of responses back from the server.
-
-The user can also specify custom parameters to the model using
-`--request-parameter <name:value:type>` option.
-For instance, passing `--request-parameter max_tokens:256:uint` will set an
-additional parameter `max_tokens` of type `int` to 256 as part of the request.
-
-```bash
-perf_analyzer -m <model_name> -i grpc --async --streaming \
-    --profile-export-file profile.json \
-    --periodic-concurrency-range 10:100:30 \
-    --request-period 10 \
-    --request-parameter max_tokens:256:int
-```
-
-> **Note**
->
-> The periodic concurrency mode is currently supported only by gRPC protocol and
-> with [decoupled models](https://github.com/triton-inference-server/server/blob/main/docs/user_guide/decoupled_models.md).
-> Additionally, the user must also specify a file where Perf Analyzer could dump all the
-> profiled data using `--profile-export-file`.
-
 ## Request Rate Mode
 
 In request rate mode, Perf Analyzer attempts to send N inference requests per
diff --git a/src/c++/perf_analyzer/docs/llm.md b/src/c++/perf_analyzer/docs/llm.md
index da916d78c..a5f5b6233 100644
--- a/src/c++/perf_analyzer/docs/llm.md
+++ b/src/c++/perf_analyzer/docs/llm.md
@@ -1,5 +1,5 @@
 <!--
-Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+Copyright 2023-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 
 Redistribution and use in source and binary forms, with or without
 modification, are permitted provided that the following conditions
@@ -224,80 +224,3 @@ python profile.py -m ensemble -b trtllm --prompt-size-range 100 500 200 --max-to
 #   * Avg first token latency: 16.0468 ms
 #  ...
 ```
-
-## Benchmark 3: Profiling In-Flight Batching
-
-In this benchmarking scenario, we want to measure the effect of in-flight
-batch size on token-to-token (T2T) latency. We systematically issue requests to
-the server of fixed input sizes and request the model to compute a fixed amount
-of tokens in order to increase the in-flight batch size over time.
-
-#### Example
-
-In this benchmark, we will run Perf Analyzer in
-[periodic concurrency mode](inference_load_modes.md#periodic-concurrency-mode)
-that periodically launches a new concurrent request to the model using
-`--periodic-concurrency-range START END STEP` option.
-In this example, Perf Analyzer starts with a single request and launches the new
-ones until the total number reaches 100.
-You can also specify the timing of the new requests:
-Setting `--request-period` to 32 (as shown below) will make Perf Analyzer to
-wait for all the requests to receive 32 responses before launching new requests.
-Run the following command inside the client container.
-
-```bash
-# Install matplotlib to generate the benchmark plot
-pip install matplotlib
-
-# Run Perf Analyzer
-# trtllm: -m ensemble -b trtllm
-# vllm: -m vllm_model -b vllm
-python profile.py -m ensemble -b trtllm --prompt-size-range 10 10 1 --periodic-concurrency-range 1 100 1 --request-period 32 --max-tokens 1024 --ignore-eos
-
-# [ BENCHMARK SUMMARY ]
-# Prompt size: 10
-#   * Max first token latency: 125.7212 ms
-#   * Min first token latency: 18.4281 ms
-#   * Avg first token latency: 61.8372 ms
-#   ...
-# Saved in-flight batching benchmark plots @ 'inflight_batching_benchmark-*.png'.
-```
-
-The resulting plot will look like
-
-<img src="examples/inflight_batching_benchmark.png" width="600">
-
-The plot demonstrates how the average T2T latency changes across the entire
-benchmark process as we increase the number of requests.
-To observe the change, we first align the responses of every requests and then
-split them into multiple segments of responses.
-For instance, assume we ran the following benchmark command:
-
-```bash
-# trtllm: -m ensemble -b trtllm
-# vllm: -m vllm_model -b vllm
-python profile.py -m ensemble -b trtllm --periodic-concurrency-range 1 4 1 --request-period 32 --max-tokens 1024 --ignore-eos
-```
-
-We start from a single request and increment up to 4 requests one by one for
-every 32 responses (defined by `--request-period`).
-For each request, there are total 1024 generated responses (defined by `--max-tokens`).
-We align these total 1024 generated responses and split them by request period,
-giving us 1024/32 = 32 total segments per request as shown below:
-
-```
-          32 responses (=request period)
-            ┌────┐
-request 1   ──────┊──────┊──────┊──────┊─ ··· ─┊──────┊
-request 2         ┊──────┊──────┊──────┊─ ··· ─┊──────┊──────┊
-request 3         ┊      ┊──────┊──────┊─ ··· ─┊──────┊──────┊──────┊
-request 4         ┊      ┊      ┊──────┊─ ··· ─┊──────┊──────┊──────┊──────
-
-segment #     1      2      3       4     ···     32     33     34     35
-```
-
-Then for each segment, we compute the mean of T2T latencies of the responses.
-This will allow us to visualize the change in T2T latency as the number of
-requests increase, filling up the inflight batch slots, and as they terminate.
-See [profile.py](examples/profile.py) for more details.
-
diff --git a/src/c++/perf_analyzer/infer_context.cc b/src/c++/perf_analyzer/infer_context.cc
index 588d4057a..65e943df6 100644
--- a/src/c++/perf_analyzer/infer_context.cc
+++ b/src/c++/perf_analyzer/infer_context.cc
@@ -1,4 +1,4 @@
-// Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2023-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -259,7 +259,6 @@ InferContext::AsyncCallbackFuncImpl(cb::InferResult* result)
           return;
         }
         it->second.response_times_.push_back(std::chrono::system_clock::now());
-        num_responses_++;
         if (is_null_response == true) {
           it->second.has_null_last_response_ = true;
         }
@@ -269,7 +268,6 @@ InferContext::AsyncCallbackFuncImpl(cb::InferResult* result)
           return;
         }
         if (is_final_response) {
-          has_received_final_response_ = is_final_response;
           thread_stat_->request_records_.emplace_back(
               it->second.start_time_, it->second.response_times_,
               it->second.sequence_end_, it->second.delayed_,
@@ -282,13 +280,8 @@ InferContext::AsyncCallbackFuncImpl(cb::InferResult* result)
     }
   }
 
-  if (worker_callback_) {
-    worker_callback_(id_);
-  }
-
   if (is_final_response) {
     total_ongoing_requests_--;
-    num_responses_ = 0;
 
     if (async_callback_finalize_func_ != nullptr) {
       async_callback_finalize_func_(id_);
diff --git a/src/c++/perf_analyzer/infer_context.h b/src/c++/perf_analyzer/infer_context.h
index 912cd2ca0..56ed9fae8 100644
--- a/src/c++/perf_analyzer/infer_context.h
+++ b/src/c++/perf_analyzer/infer_context.h
@@ -1,4 +1,4 @@
-// Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2023-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -116,28 +116,18 @@ class InferContext {
   // object and have not returned
   uint GetNumOngoingRequests() { return total_ongoing_requests_; }
 
-  // Returns the number of responses for the current request
-  uint64_t GetNumResponsesForCurrentRequest() { return num_responses_; }
-
   // Register a function that will get called after every async request returns
   void RegisterAsyncCallbackFinalize(std::function<void(uint32_t)> callback)
   {
     async_callback_finalize_func_ = callback;
   }
 
-  void RegisterWorkerCallback(std::function<void(uint32_t)> worker_callback)
-  {
-    worker_callback_ = worker_callback;
-  }
-
   // TODO REFACTOR TMA-1043 this should be in memory class
   void SetNumActiveThreads(size_t num_threads)
   {
     num_active_threads_ = num_threads;
   }
 
-  bool HasReceivedFinalResponse() { return has_received_final_response_; }
-
  protected:
   /// A helper function to issue inference request to the server.
   /// \param request_id The unique id to be associated with the request.
@@ -201,9 +191,6 @@ class InferContext {
   std::reference_wrapper<const bool> execute_{execute_placeholder_};
 
   std::shared_ptr<SequenceManager> sequence_manager_{nullptr};
-  uint64_t num_responses_{0};
-  std::function<void(uint32_t)> worker_callback_{nullptr};
-  bool has_received_final_response_{false};
 
 #ifndef DOCTEST_CONFIG_DISABLE
   friend NaggyMockInferContext;
diff --git a/src/c++/perf_analyzer/inference_profiler.h b/src/c++/perf_analyzer/inference_profiler.h
index 913b23ded..ca76e74e0 100644
--- a/src/c++/perf_analyzer/inference_profiler.h
+++ b/src/c++/perf_analyzer/inference_profiler.h
@@ -1,4 +1,4 @@
-// Copyright 2020-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2020-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -43,7 +43,6 @@
 #include "metrics_manager.h"
 #include "model_parser.h"
 #include "mpi_utils.h"
-#include "periodic_concurrency_manager.h"
 #include "profile_data_collector.h"
 #include "request_rate_manager.h"
 
@@ -307,18 +306,6 @@ class InferenceProfiler {
     return cb::Error::Success;
   }
 
-  cb::Error ProfilePeriodicConcurrencyMode()
-  {
-    auto& manager{dynamic_cast<PeriodicConcurrencyManager&>(*manager_)};
-    std::vector<RequestRecord> request_records{manager.RunExperiment()};
-    // FIXME - Refactor collector class to not need ID or window in the case of
-    // periodic concurrency mode
-    InferenceLoadMode id{1, 0.0};
-    collector_->AddWindow(id, 0, UINT64_MAX);
-    collector_->AddData(id, std::move(request_records));
-    return cb::Error::Success;
-  }
-
   bool IncludeServerStats() { return include_server_stats_; }
 
  private:
diff --git a/src/c++/perf_analyzer/perf_analyzer.cc b/src/c++/perf_analyzer/perf_analyzer.cc
index 46b665757..29ec3a4cf 100644
--- a/src/c++/perf_analyzer/perf_analyzer.cc
+++ b/src/c++/perf_analyzer/perf_analyzer.cc
@@ -1,4 +1,4 @@
-// Copyright 2022-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2022-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -27,7 +27,6 @@
 #include "perf_analyzer.h"
 
 #include "perf_analyzer_exception.h"
-#include "periodic_concurrency_manager.h"
 #include "report_writer.h"
 #include "request_rate_manager.h"
 
@@ -209,13 +208,6 @@ PerfAnalyzer::CreateAnalyzerObjects()
             factory, &manager, params_->request_parameters),
         "failed to create concurrency manager");
 
-  } else if (params_->is_using_periodic_concurrency_mode) {
-    manager = std::make_unique<pa::PeriodicConcurrencyManager>(
-        params_->async, params_->streaming, params_->batch_size,
-        params_->max_threads, params_->max_concurrency,
-        params_->shared_memory_type, params_->output_shm_size, parser_, factory,
-        params_->periodic_concurrency_range, params_->request_period,
-        params_->request_parameters);
   } else if (params_->using_request_rate_range) {
     if ((params_->sequence_id_range != 0) &&
         (params_->sequence_id_range < params_->num_of_sequences)) {
@@ -377,8 +369,6 @@ PerfAnalyzer::Profile()
     err = profiler_->Profile<size_t>(
         params_->concurrency_range.start, params_->concurrency_range.end,
         params_->concurrency_range.step, params_->search_mode, perf_statuses_);
-  } else if (params_->is_using_periodic_concurrency_mode) {
-    err = profiler_->ProfilePeriodicConcurrencyMode();
   } else {
     err = profiler_->Profile<double>(
         params_->request_rate_range[pa::SEARCH_RANGE::kSTART],
@@ -402,7 +392,7 @@ PerfAnalyzer::Profile()
 void
 PerfAnalyzer::WriteReport()
 {
-  if (!perf_statuses_.size() || params_->is_using_periodic_concurrency_mode) {
+  if (!perf_statuses_.size()) {
     return;
   }
 
diff --git a/src/c++/perf_analyzer/periodic_concurrency_manager.cc b/src/c++/perf_analyzer/periodic_concurrency_manager.cc
deleted file mode 100644
index 1a5527b7b..000000000
--- a/src/c++/perf_analyzer/periodic_concurrency_manager.cc
+++ /dev/null
@@ -1,121 +0,0 @@
-// Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-//
-// Redistribution and use in source and binary forms, with or without
-// modification, are permitted provided that the following conditions
-// are met:
-//  * Redistributions of source code must retain the above copyright
-//    notice, this list of conditions and the following disclaimer.
-//  * Redistributions in binary form must reproduce the above copyright
-//    notice, this list of conditions and the following disclaimer in the
-//    documentation and/or other materials provided with the distribution.
-//  * Neither the name of NVIDIA CORPORATION nor the names of its
-//    contributors may be used to endorse or promote products derived
-//    from this software without specific prior written permission.
-//
-// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
-// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
-// PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
-// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
-// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
-// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
-// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
-// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
-#include "periodic_concurrency_manager.h"
-
-namespace triton { namespace perfanalyzer {
-
-std::vector<RequestRecord>
-PeriodicConcurrencyManager::RunExperiment()
-{
-  AddConcurrentRequests(concurrency_range_.start);
-  WaitForRequestsToFinish();
-  return GetRequestRecords();
-}
-
-std::shared_ptr<IWorker>
-PeriodicConcurrencyManager::MakeWorker(
-    std::shared_ptr<ThreadStat> thread_stat,
-    std::shared_ptr<PeriodicConcurrencyWorker::ThreadConfig> thread_config)
-{
-  uint32_t id = workers_.size();
-  auto worker = std::make_shared<PeriodicConcurrencyWorker>(
-      id, thread_stat, thread_config, parser_, data_loader_, factory_,
-      on_sequence_model_, async_, max_concurrency_, using_json_data_,
-      streaming_, batch_size_, wake_signal_, wake_mutex_, active_threads_,
-      execute_, infer_data_manager_, sequence_manager_, request_period_,
-      period_completed_callback_, request_completed_callback_);
-  return worker;
-};
-
-void
-PeriodicConcurrencyManager::AddConcurrentRequests(
-    uint64_t num_concurrent_requests)
-{
-  for (size_t i = 0; i < num_concurrent_requests; i++) {
-    AddConcurrentRequest(i);
-  }
-  num_incomplete_periods_ = num_concurrent_requests;
-}
-
-void
-PeriodicConcurrencyManager::AddConcurrentRequest(size_t seq_stat_index_offset)
-{
-  threads_stat_.emplace_back(std::make_shared<ThreadStat>());
-  threads_config_.emplace_back(
-      std::make_shared<ConcurrencyWorker::ThreadConfig>(
-          threads_config_.size(), 1, seq_stat_index_offset));
-  workers_.emplace_back(
-      MakeWorker(threads_stat_.back(), threads_config_.back()));
-  threads_.emplace_back(&IWorker::Infer, workers_.back());
-  active_threads_++;
-}
-
-void
-PeriodicConcurrencyManager::PeriodCompletedCallback()
-{
-  std::lock_guard<std::mutex> lock(period_completed_callback_mutex_);
-  num_incomplete_periods_--;
-  if (num_incomplete_periods_ == 0) {
-    steps_completed_++;
-    uint64_t num_requests_sent{steps_completed_ * concurrency_range_.step};
-    if (num_requests_sent < concurrency_range_.end) {
-      AddConcurrentRequests(concurrency_range_.step);
-    }
-  }
-}
-
-void
-PeriodicConcurrencyManager::RequestCompletedCallback()
-{
-  std::lock_guard<std::mutex> lock(request_completed_callback_mutex_);
-  num_completed_requests_++;
-  if (num_completed_requests_ == concurrency_range_.end) {
-    all_requests_completed_promise_.set_value(true);
-  }
-}
-
-void
-PeriodicConcurrencyManager::WaitForRequestsToFinish()
-{
-  std::future<bool> all_requests_completed_future{
-      all_requests_completed_promise_.get_future()};
-  all_requests_completed_future.get();
-}
-
-std::vector<RequestRecord>
-PeriodicConcurrencyManager::GetRequestRecords()
-{
-  std::vector<RequestRecord> request_records{};
-  for (const auto& thread_stat : threads_stat_) {
-    request_records.insert(
-        request_records.end(), thread_stat->request_records_.cbegin(),
-        thread_stat->request_records_.cend());
-  }
-  return request_records;
-}
-
-}}  // namespace triton::perfanalyzer
diff --git a/src/c++/perf_analyzer/periodic_concurrency_manager.h b/src/c++/perf_analyzer/periodic_concurrency_manager.h
deleted file mode 100644
index db612fd96..000000000
--- a/src/c++/perf_analyzer/periodic_concurrency_manager.h
+++ /dev/null
@@ -1,93 +0,0 @@
-// Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-//
-// Redistribution and use in source and binary forms, with or without
-// modification, are permitted provided that the following conditions
-// are met:
-//  * Redistributions of source code must retain the above copyright
-//    notice, this list of conditions and the following disclaimer.
-//  * Redistributions in binary form must reproduce the above copyright
-//    notice, this list of conditions and the following disclaimer in the
-//    documentation and/or other materials provided with the distribution.
-//  * Neither the name of NVIDIA CORPORATION nor the names of its
-//    contributors may be used to endorse or promote products derived
-//    from this software without specific prior written permission.
-//
-// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
-// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
-// PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
-// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
-// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
-// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
-// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
-// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-#pragma once
-
-#include <future>
-
-#include "concurrency_manager.h"
-#include "periodic_concurrency_worker.h"
-
-namespace triton { namespace perfanalyzer {
-
-/// @brief Concurrency manager for periodically increasing concurrency by a step
-/// amount based on the number of responses received (request period) by the
-/// latest N (step or start concurrency for first-issued concurrent requests)
-/// concurrent requests/workers.
-class PeriodicConcurrencyManager : public ConcurrencyManager {
- public:
-  PeriodicConcurrencyManager(
-      const bool async, const bool streaming, const int32_t batch_size,
-      const size_t max_threads, const size_t max_concurrency,
-      const SharedMemoryType shared_memory_type, const size_t output_shm_size,
-
-      const std::shared_ptr<ModelParser>& parser,
-      const std::shared_ptr<cb::ClientBackendFactory>& factory,
-      const Range<uint64_t> concurrency_range, const uint64_t request_period,
-      const std::unordered_map<std::string, cb::RequestParameter>&
-          request_parameters)
-      : ConcurrencyManager(
-            async, streaming, batch_size, max_threads, max_concurrency,
-            shared_memory_type, output_shm_size, parser, factory,
-            request_parameters),
-        concurrency_range_(concurrency_range), request_period_(request_period)
-  {
-  }
-
-  std::vector<RequestRecord> RunExperiment();
-
- private:
-  std::shared_ptr<IWorker> MakeWorker(
-      std::shared_ptr<ThreadStat> thread_stat,
-      std::shared_ptr<PeriodicConcurrencyWorker::ThreadConfig> thread_config)
-      override;
-
-  void AddConcurrentRequests(uint64_t num_concurrent_requests);
-
-  void AddConcurrentRequest(size_t seq_stat_index_offset);
-
-  void PeriodCompletedCallback();
-
-  void RequestCompletedCallback();
-
-  void WaitForRequestsToFinish();
-
-  std::vector<RequestRecord> GetRequestRecords();
-
-  Range<uint64_t> concurrency_range_{1, 1, 1};
-  uint64_t request_period_{0};
-  uint64_t steps_completed_{0};
-  uint64_t num_incomplete_periods_{0};
-  uint64_t num_completed_requests_{0};
-  std::mutex period_completed_callback_mutex_{};
-  std::mutex request_completed_callback_mutex_{};
-  std::promise<bool> all_requests_completed_promise_{};
-  std::function<void()> period_completed_callback_{
-      std::bind(&PeriodicConcurrencyManager::PeriodCompletedCallback, this)};
-  std::function<void()> request_completed_callback_{
-      std::bind(&PeriodicConcurrencyManager::RequestCompletedCallback, this)};
-};
-
-}}  // namespace triton::perfanalyzer
diff --git a/src/c++/perf_analyzer/periodic_concurrency_worker.cc b/src/c++/perf_analyzer/periodic_concurrency_worker.cc
deleted file mode 100644
index 9af3a9d87..000000000
--- a/src/c++/perf_analyzer/periodic_concurrency_worker.cc
+++ /dev/null
@@ -1,71 +0,0 @@
-// Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-//
-// Redistribution and use in source and binary forms, with or without
-// modification, are permitted provided that the following conditions
-// are met:
-//  * Redistributions of source code must retain the above copyright
-//    notice, this list of conditions and the following disclaimer.
-//  * Redistributions in binary form must reproduce the above copyright
-//    notice, this list of conditions and the following disclaimer in the
-//    documentation and/or other materials provided with the distribution.
-//  * Neither the name of NVIDIA CORPORATION nor the names of its
-//    contributors may be used to endorse or promote products derived
-//    from this software without specific prior written permission.
-//
-// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
-// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
-// PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
-// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
-// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
-// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
-// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
-// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
-#include "periodic_concurrency_worker.h"
-
-namespace triton { namespace perfanalyzer {
-
-void
-PeriodicConcurrencyWorker::Infer()
-{
-  CreateCtxIdTracker();
-  ReserveContexts();
-  RunInference();
-}
-
-std::shared_ptr<InferContext>
-PeriodicConcurrencyWorker::CreateInferContext()
-{
-  std::shared_ptr infer_context{std::make_shared<InferContext>(
-      id_, ctxs_.size(), async_, streaming_, on_sequence_model_,
-      using_json_data_, batch_size_, thread_stat_, data_loader_, parser_,
-      factory_, execute_, infer_data_manager_, sequence_manager_)};
-  infer_context->RegisterWorkerCallback(worker_callback_);
-  return infer_context;
-}
-
-void
-PeriodicConcurrencyWorker::WorkerCallback(uint32_t infer_context_id)
-{
-  if (ctxs_.at(infer_context_id)->GetNumResponsesForCurrentRequest() ==
-      request_period_) {
-    period_completed_callback_();
-  }
-  if (ctxs_.at(infer_context_id)->HasReceivedFinalResponse()) {
-    bool has_not_completed_period{
-        ctxs_.at(infer_context_id)->GetNumResponsesForCurrentRequest() <
-        request_period_};
-    if (has_not_completed_period) {
-      throw std::runtime_error(
-          "Request received final response before request period was reached. "
-          "Request period must be at most the total number of responses "
-          "received by any request.");
-    }
-    request_completed_callback_();
-  }
-}
-
-}}  // namespace triton::perfanalyzer
diff --git a/src/c++/perf_analyzer/periodic_concurrency_worker.h b/src/c++/perf_analyzer/periodic_concurrency_worker.h
deleted file mode 100644
index 7242219b9..000000000
--- a/src/c++/perf_analyzer/periodic_concurrency_worker.h
+++ /dev/null
@@ -1,80 +0,0 @@
-// Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-//
-// Redistribution and use in source and binary forms, with or without
-// modification, are permitted provided that the following conditions
-// are met:
-//  * Redistributions of source code must retain the above copyright
-//    notice, this list of conditions and the following disclaimer.
-//  * Redistributions in binary form must reproduce the above copyright
-//    notice, this list of conditions and the following disclaimer in the
-//    documentation and/or other materials provided with the distribution.
-//  * Neither the name of NVIDIA CORPORATION nor the names of its
-//    contributors may be used to endorse or promote products derived
-//    from this software without specific prior written permission.
-//
-// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
-// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
-// PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
-// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
-// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
-// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
-// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
-// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-#pragma once
-
-#include <thread>
-
-#include "concurrency_worker.h"
-
-namespace triton { namespace perfanalyzer {
-
-/// @brief Worker class for periodic concurrency mode. Issues one request only
-/// and waits for all responses to come in. Notifies manager when N responses
-/// (request period) have been received. Notifies manager when final response
-/// has been received.
-class PeriodicConcurrencyWorker : public ConcurrencyWorker {
- public:
-  PeriodicConcurrencyWorker(
-      uint32_t id, std::shared_ptr<ThreadStat> thread_stat,
-      std::shared_ptr<ThreadConfig> thread_config,
-      const std::shared_ptr<ModelParser> parser,
-      std::shared_ptr<DataLoader> data_loader,
-      const std::shared_ptr<cb::ClientBackendFactory> factory,
-      const bool on_sequence_model, const bool async,
-      const size_t max_concurrency, const bool using_json_data,
-      const bool streaming, const int32_t batch_size,
-      std::condition_variable& wake_signal, std::mutex& wake_mutex,
-      size_t& active_threads, bool& execute,
-      const std::shared_ptr<IInferDataManager>& infer_data_manager,
-      std::shared_ptr<SequenceManager> sequence_manager,
-      uint64_t request_period, std::function<void()> period_completed_callback,
-      std::function<void()> request_completed_callback)
-      : ConcurrencyWorker(
-            id, thread_stat, thread_config, parser, data_loader, factory,
-            on_sequence_model, async, max_concurrency, using_json_data,
-            streaming, batch_size, wake_signal, wake_mutex, active_threads,
-            execute, infer_data_manager, sequence_manager),
-        request_period_(request_period),
-        period_completed_callback_(period_completed_callback),
-        request_completed_callback_(request_completed_callback)
-  {
-  }
-
-  void Infer() override;
-
-  std::shared_ptr<InferContext> CreateInferContext() override;
-
-  void WorkerCallback(uint32_t infer_context_id);
-
- private:
-  uint64_t request_period_{0};
-  std::function<void()> period_completed_callback_{nullptr};
-  std::function<void()> request_completed_callback_{nullptr};
-  std::function<void(uint32_t)> worker_callback_{std::bind(
-      &PeriodicConcurrencyWorker::WorkerCallback, this, std::placeholders::_1)};
-};
-
-}}  // namespace triton::perfanalyzer
diff --git a/src/c++/perf_analyzer/test_command_line_parser.cc b/src/c++/perf_analyzer/test_command_line_parser.cc
index fd0d8af16..ff0da78a4 100644
--- a/src/c++/perf_analyzer/test_command_line_parser.cc
+++ b/src/c++/perf_analyzer/test_command_line_parser.cc
@@ -1,4 +1,4 @@
-// Copyright 2022-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2022-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -176,19 +176,6 @@ CHECK_PARAMS(PAParamsPtr act, PAParamsPtr exp)
   CHECK_STRING(act->filename, act->filename);
   CHECK(act->mpi_driver != nullptr);
   CHECK_STRING(act->memory_type, exp->memory_type);
-  CHECK(
-      act->is_using_periodic_concurrency_mode ==
-      exp->is_using_periodic_concurrency_mode);
-  CHECK(
-      act->periodic_concurrency_range.start ==
-      exp->periodic_concurrency_range.start);
-  CHECK(
-      act->periodic_concurrency_range.end ==
-      exp->periodic_concurrency_range.end);
-  CHECK(
-      act->periodic_concurrency_range.step ==
-      exp->periodic_concurrency_range.step);
-  CHECK(act->request_period == exp->request_period);
   CHECK(act->request_parameters.size() == exp->request_parameters.size());
   for (auto act_param : act->request_parameters) {
     auto exp_param = exp->request_parameters.find(act_param.first);
@@ -1173,153 +1160,6 @@ TEST_CASE("Testing Command Line Parser")
     }
   }
 
-  SUBCASE("Option : --periodic-concurrency-range")
-  {
-    char* option_name = "--periodic-concurrency-range";
-
-    // Add required args that specifies where to dump profiled data
-    args.insert(
-        args.end(), {"-i", "grpc", "--async", "--streaming",
-                     "--profile-export-file", "profile.json"});
-    exp->protocol = cb::ProtocolType::GRPC;
-    exp->async = true;
-    exp->streaming = true;
-    exp->url = "localhost:8001";  // gRPC url
-
-    SUBCASE("start provided")
-    {
-      args.push_back(option_name);
-      args.push_back("100");  // start
-
-      int argc = args.size();
-      char* argv[argc];
-      std::copy(args.begin(), args.end(), argv);
-
-      expected_msg = CreateUsageMessage(
-          option_name, "Both <start> and <end> values must be provided.");
-      CHECK_THROWS_WITH_AS(
-          act = parser.Parse(argc, argv), expected_msg.c_str(),
-          PerfAnalyzerException);
-
-      check_params = false;
-    }
-
-    exp->max_threads = 400;
-
-    CheckValidRange(
-        args, option_name, parser, act, exp->is_using_periodic_concurrency_mode,
-        exp->periodic_concurrency_range);
-
-    CheckInvalidRange(args, option_name, parser, act, check_params);
-
-    SUBCASE("more than one load mode")
-    {
-      args.push_back(option_name);
-      args.push_back("100:400");
-      args.push_back("--concurrency-range");
-      args.push_back("10:40");
-
-      int argc = args.size();
-      char* argv[argc];
-      std::copy(args.begin(), args.end(), argv);
-
-      expected_msg =
-          "Cannot specify more then one inference load mode. Please choose "
-          "only one of the following modes: --concurrency-range, "
-          "--periodic-concurrency-range, --request-rate-range, or "
-          "--request-intervals.";
-      CHECK_THROWS_WITH_AS(
-          act = parser.Parse(argc, argv), expected_msg.c_str(),
-          PerfAnalyzerException);
-
-      check_params = false;
-    }
-
-    SUBCASE("no export file specified")
-    {
-      // Remove the export file args
-      args.pop_back();
-      args.pop_back();
-
-      args.push_back(option_name);
-      args.push_back("100:400");
-
-      int argc = args.size();
-      char* argv[argc];
-      std::copy(args.begin(), args.end(), argv);
-
-      expected_msg =
-          "Must provide --profile-export-file when using the "
-          "--periodic-concurrency-range option.";
-      CHECK_THROWS_WITH_AS(
-          act = parser.Parse(argc, argv), expected_msg.c_str(),
-          PerfAnalyzerException);
-
-      check_params = false;
-    }
-
-    SUBCASE("step is not factor of range size")
-    {
-      args.push_back(option_name);
-      args.push_back("100:400:7");
-
-      int argc = args.size();
-      char* argv[argc];
-      std::copy(args.begin(), args.end(), argv);
-
-      expected_msg = CreateUsageMessage(
-          option_name,
-          "The <step> value must be a factor of the range size (<end> - "
-          "<start>).");
-      CHECK_THROWS_WITH_AS(
-          act = parser.Parse(argc, argv), expected_msg.c_str(),
-          PerfAnalyzerException);
-
-      check_params = false;
-    }
-
-    SUBCASE("step is zero")
-    {
-      args.push_back(option_name);
-      args.push_back("10:400:0");
-
-      int argc = args.size();
-      char* argv[argc];
-      std::copy(args.begin(), args.end(), argv);
-
-      expected_msg =
-          CreateUsageMessage(option_name, "The <step> value must be > 0.");
-      CHECK_THROWS_WITH_AS(
-          act = parser.Parse(argc, argv), expected_msg.c_str(),
-          PerfAnalyzerException);
-
-      check_params = false;
-    }
-  }
-
-  SUBCASE("Option : --request-period")
-  {
-    expected_msg =
-        CreateUsageMessage("--request-period", "The value must be > 0");
-    CHECK_INT_OPTION("--request-period", exp->request_period, expected_msg);
-
-    SUBCASE("set to 0")
-    {
-      args.push_back("--request-period");
-      args.push_back("0");
-
-      int argc = args.size();
-      char* argv[argc];
-      std::copy(args.begin(), args.end(), argv);
-
-      CHECK_THROWS_WITH_AS(
-          act = parser.Parse(argc, argv), expected_msg.c_str(),
-          PerfAnalyzerException);
-
-      check_params = false;
-    }
-  }
-
   SUBCASE("Option : --request-parameter")
   {
     char* option_name = "--request-parameter";