Update .gitignore and benchmark scripts for GPU support; enhance TTS service handling and session management

2025-04-13 09:39:17 +00:00 · 2025-01-30 05:47:28 -07:00 · 2025-01-30 05:47:28 -07:00 · 2dfad62b5b
commit 2dfad62b5b
parent f61f79981d
18 changed files with 522 additions and 2877 deletions
--- a/.gitignore
+++ b/.gitignore
@ -67,4 +67,5 @@ examples/*.flac
 examples/*.acc
 examples/*.ogg
 examples/speech.mp3
-examples/phoneme_examples/output/*.wav
+examples/phoneme_examples/output/*.wav
+examples/assorted_checks/benchmarks/output_audio/*
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@ -2,6 +2,34 @@

 Notable changes to this project will be documented in this file.

+## [v0.1.4] - 2025-01-30
+### Added
+- Smart Chunking System:
+  - New text_processor with smart_split for improved sentence boundary detection
+  - Dynamically adjusts chunk sizes based on sentence structure, using phoneme/token information in an intial pass
+  - Should avoid ever going over the 510 limit per chunk, while preserving natural cadence
+- Web UI Added (To Be Replacing Gradio):
+  - Integrated streaming with tempfile generation
+  - Download links available in X-Download-Path header
+  - Configurable cleanup triggers for temp files
+- Debug Endpoints:
+  - /debug/threads for thread information and stack traces
+  - /debug/storage for temp file and output directory monitoring
+  - /debug/system for system resource information
+  - /debug/session_pools for ONNX/CUDA session status
+- Automated Model Management:
+  - Auto-download from releases page
+  - Included download scripts for manual installation
+  - Pre-packaged voice models in repository
+
+### Changed
+- Significant architectural improvements:
+  - Multi-model architecture support
+  - Enhanced concurrency handling
+  - Improved streaming header management
+  - Better resource/session pool management
+
+
 ## [v0.1.2] - 2025-01-23
 ### Structural Improvements
 - Models can be manually download and placed in api/src/models, or use included script
--- a/README.md
+++ b/README.md
@ -3,51 +3,58 @@
 </p>

 # <sub><sub>_`FastKoko`_ </sub></sub>
-[![Tests](https://img.shields.io/badge/tests-104%20passed-darkgreen)]()
+[![Tests](https://img.shields.io/badge/tests-100%20passed-darkgreen)]()
 [![Coverage](https://img.shields.io/badge/coverage-49%25-grey)]()
 [![Tested at Model Commit](https://img.shields.io/badge/last--tested--model--commit-a67f113-blue)](https://huggingface.co/hexgrad/Kokoro-82M/tree/c3b0d86e2a980e027ef71c28819ea02e351c2667) [![Try on Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Try%20on-Spaces-blue)](https://huggingface.co/spaces/Remsky/Kokoro-TTS-Zero)

-> Pre-release. Not fully tested
+> Support for Kokoro-82M v1.0 coming very soon!

 Dockerized FastAPI wrapper for [Kokoro-82M](https://huggingface.co/hexgrad/Kokoro-82M) text-to-speech model
 - OpenAI-compatible Speech endpoint, with inline voice combination, and mapped naming/models for strict systems
 - NVIDIA GPU accelerated or CPU inference (ONNX, Pytorch) 
 - very fast generation time
-  - 35x-100x+ real time speed via 4060Ti+
-  - 5x+ real time speed via M3 Pro CPU
- streaming support w/ variable chunking to control latency, (new) improved concurrency
+  - ~35x-100x+ real time speed via 4060Ti+
+  - ~5x+ real time speed via M3 Pro CPU
+- streaming support & tempfile generation
 - phoneme based dev endpoints
 - (new) Integrated web UI on localhost:8880/web
+- (new) Debug endpoints for monitoring threads, storage, and session pools

-> [!Tip]
-> You can try the new beta version from the `v0.1.2-pre` branch now:
-<table>
-  <tr>
-    <td>
-      <img src="https://github.com/user-attachments/assets/440162eb-1918-4999-ab2b-e2730990efd0" width="100%" alt="Voice Analysis Comparison" style="border: 2px solid #333; padding: 5px;">
-    </td>
-    <td>
-      <ul>
-        <li>Integrated web UI (on localhost:8880/web)</li>
-        <li>Better concurrency handling, baked in models and voices</li>
-        <li>Voice name/model mappings to OAI standard</li>
-        <pre> # with:
-docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:latest # CPU
-docker run --gpus all -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-gpu:latest # Nvidia GPU
-        </pre>
-      </ul>
-    </td>
-  </tr>
-</table>

-<details open>
-<summary>Quick Start</summary>
+## Get Started

-The service can be accessed through either the API endpoints or the Gradio web interface.
+<details >
+<summary>Quickest Start (docker run)</summary>
+
+
+Pre built images are available to run, with arm/multi-arch support, and baked in models
+Refer to the core/config.py file for a full list of variables which can be managed via the environment
+
+```bash
+
+docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:v0.1.4 # CPU, or:
+docker run --gpus all -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-gpu:v0.1.4 #NVIDIA GPU
+```
+
+Once running, access:
+- API Documentation: http://localhost:8880/docs
+- Web Interface: http://localhost:8880/web
+
+<div align="center" style="display: flex; justify-content: center; gap: 20px;">
+  <img src="assets/docs-screenshot.png" width="48%" alt="API Documentation" style="border: 2px solid #333; padding: 10px;">
+  <img src="assets/webui-screenshot.png" width="48%" alt="Web UI Screenshot" style="border: 2px solid #333; padding: 10px;">
+</div>
+
+</details>
+
+
+<details>
+
+<summary>Quick Start (docker compose) </summary>

 1. Install prerequisites, and start the service using Docker Compose (Full setup including UI):
   - Install [Docker](https://www.docker.com/products/docker-desktop/)
-
+   - 
   - Clone the repository:
        ```bash
        git clone https://github.com/remsky/Kokoro-FastAPI.git
@ -60,20 +67,18 @@ The service can be accessed through either the API endpoints or the Gradio web i
        # python ../scripts/download_model.py --type pth  # for GPU
        # python ../scripts/download_model.py --type onnx # for CPU
        ```
-        
+
+        ```bash
+        Or directly via UV
+        ./start-cpu.sh
+        ./start-gpu.sh 
+        ```
+
      Once started:
     - The API will be available at http://localhost:8880
     - The *Web UI* can be tested at http://localhost:8880/web
     - The Gradio UI (deprecating) can be accessed at http://localhost:7860
-        
-    __Or__ running the API alone using Docker (model + voice packs baked in) (Most Recent):
-          
-    ```bash
-    docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:v0.1.2 # CPU 
-    docker run --gpus all -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-gpu:v0.1.2 # Nvidia GPU
-    ```
-        
-        
+
 2. Run locally as an OpenAI-Compatible Speech Endpoint
    ```python
    from openai import OpenAI
@ -91,13 +96,50 @@ The service can be accessed through either the API endpoints or the Gradio web i
        response.stream_to_file("output.mp3")
    
    ```
+</details>
+<summary>Direct Run (via uv) </summary>

-  <div align="center">
-    <div style="display: flex; justify-content: center; gap: 20px;">
-      <img src="assets/beta_web_ui.png" width="45%" alt="Beta Web UI" style="border: 2px solid #333; padding: 10px;">
-      <img src="ui/GradioScreenShot.png" width="45%" alt="Voice Analysis Comparison" style="border: 2px solid #333; padding: 10px;">
-    </div>
-  </div>
+1. Install prerequisites ():
+   - Install [astral-uv](https://docs.astral.sh/uv/)
+   - Clone the repository:
+        ```bash
+        git clone https://github.com/remsky/Kokoro-FastAPI.git
+        cd Kokoro-FastAPI
+
+        # if you are missing any models, run:
+        # python ../scripts/download_model.py --type pth  # for GPU
+        # python ../scripts/download_model.py --type onnx # for CPU
+        ```
+
+        Start directly via UV (with hot-reload)
+        ```bash
+        ./start-cpu.sh OR
+        ./start-gpu.sh 
+        ```
+
+      Once started:
+     - The API will be available at http://localhost:8880
+     - The *Web UI* can be tested at http://localhost:8880/web
+     - The Gradio UI (deprecating) can be accessed at http://localhost:7860
+
+2. Run locally as an OpenAI-Compatible Speech Endpoint
+    ```python
+    from openai import OpenAI
+    client = OpenAI(
+        base_url="http://localhost:8880/v1",
+        api_key="not-needed"
+        )
+
+    with client.audio.speech.with_streaming_response.create(
+        model="kokoro", 
+        voice="af_sky+af_bella", #single or multiple voicepack combo
+        input="Hello world!",
+        response_format="mp3"
+    ) as response:
+        response.stream_to_file("output.mp3")
+    
+    ```
+</details>

 ## Features 
 <details>
@ -211,13 +253,12 @@ If you only want the API, just comment out everything in the docker-compose.yml

 Currently, voices created via the API are accessible here, but voice combination/creation has not yet been added

-Running the UI Docker Service
+Running the UI Docker Service [deprecating]
   - If you only want to run the Gradio web interface separately and connect it to an existing API service:
      ```bash
      docker run -p 7860:7860 \
        -e API_HOST=<api-hostname-or-ip> \
        -e API_PORT=8880 \
-        ghcr.io/remsky/kokoro-fastapi-ui:v0.1.0
      ```

     - Replace `<api-hostname-or-ip>` with:
@ -236,7 +277,7 @@ environment:

 When running the Docker image directly:
 ```bash
-docker run -p 7860:7860 -e DISABLE_LOCAL_SAVING=true ghcr.io/remsky/kokoro-fastapi-ui:latest
+docker run -p 7860:7860 -e DISABLE_LOCAL_SAVING=true ghcr.io/remsky/kokoro-fastapi-ui:v0.1.4
 ```
 </details>

@ -247,7 +288,7 @@ docker run -p 7860:7860 -e DISABLE_LOCAL_SAVING=true ghcr.io/remsky/kokoro-fasta
 # OpenAI-compatible streaming
 from openai import OpenAI
 client = OpenAI(
-    base_url="http://localhost:8880", api_key="not-needed")
+    base_url="http://localhost:8880/v1", api_key="not-needed")

 # Stream to file
 with client.audio.speech.with_streaming_response.create(
@ -329,17 +370,17 @@ Benchmarking was performed on generation via the local API using text lengths up
 </p>

 Key Performance Metrics:
- Realtime Speed: Ranges between 25-50x (generation time to output audio length)
+- Realtime Speed: Ranges between 35x-100x (generation time to output audio length)
 - Average Processing Rate: 137.67 tokens/second (cl100k_base)
 </details>
 <details>
 <summary>GPU Vs. CPU</summary>

 ```bash
-# GPU: Requires NVIDIA GPU with CUDA 12.1 support (~35x realtime speed)
+# GPU: Requires NVIDIA GPU with CUDA 12.1 support (~35x-100x realtime speed)
 docker compose up --build

-# CPU: ONNX optimized inference (~2.4x realtime speed)
+# CPU: ONNX optimized inference (~5x+ realtime speed on M3 Pro)
 docker compose -f docker-compose.cpu.yml up --build
 ```
 *Note: Overall speed may have reduced somewhat with the structural changes to accomodate streaming. Looking into it* 
@ -359,36 +400,61 @@ Convert text to phonemes and/or generate audio directly from phonemes:
 ```python
 import requests

-# Convert text to phonemes
-response = requests.post(
-    "http://localhost:8880/dev/phonemize",
-    json={
-        "text": "Hello world!",
-        "language": "a"  # "a" for American English
-    }
-)
-result = response.json()
-phonemes = result["phonemes"]  # Phoneme string e.g  ðɪs ɪz ˈoʊnli ɐ tˈɛst
-tokens = result["tokens"]      # Token IDs including start/end tokens 
+def get_phonemes(text: str, language: str = "a"):
+    """Get phonemes and tokens for input text"""
+    response = requests.post(
+        "http://localhost:8880/dev/phonemize",
+        json={"text": text, "language": language}  # "a" for American English
+    )
+    response.raise_for_status()
+    result = response.json()
+    return result["phonemes"], result["tokens"]

-# Generate audio from phonemes
-response = requests.post(
-    "http://localhost:8880/dev/generate_from_phonemes",
-    json={
-        "phonemes": phonemes,
-        "voice": "af_bella",
-        "speed": 1.0
-    }
-)
+def generate_audio_from_phonemes(phonemes: str, voice: str = "af_bella"):
+    """Generate audio from phonemes"""
+    response = requests.post(
+        "http://localhost:8880/dev/generate_from_phonemes",
+        json={"phonemes": phonemes, "voice": voice},
+        headers={"Accept": "audio/wav"}
+    )
+    if response.status_code != 200:
+        print(f"Error: {response.text}")
+        return None
+    return response.content

-# Save WAV audio
-with open("speech.wav", "wb") as f:
-    f.write(response.content)
+# Example usage
+text = "Hello world!"
+try:
+    # Convert text to phonemes
+    phonemes, tokens = get_phonemes(text)
+    print(f"Phonemes: {phonemes}")  # e.g. ðɪs ɪz ˈoʊnli ɐ tˈɛst
+    print(f"Tokens: {tokens}")      # Token IDs including start/end tokens
+
+    # Generate and save audio
+    if audio_bytes := generate_audio_from_phonemes(phonemes):
+        with open("speech.wav", "wb") as f:
+            f.write(audio_bytes)
+        print(f"Generated {len(audio_bytes)} bytes of audio")
+except Exception as e:
+    print(f"Error: {e}")
 ```

 See `examples/phoneme_examples/generate_phonemes.py` for a sample script.
 </details>

+<details>
+<summary>Debug Endpoints</summary>
+
+Monitor system state and resource usage with these endpoints:
+
+- `/debug/threads` - Get thread information and stack traces
+- `/debug/storage` - Monitor temp file and output directory usage
+- `/debug/system` - Get system information (CPU, memory, GPU)
+- `/debug/session_pools` - View ONNX session and CUDA stream status
+
+Useful for debugging resource exhaustion or performance issues.
+</details>
+
 ## Known Issues

 <details>
--- a/api/src/inference/session_pool.py
+++ b/api/src/inference/session_pool.py
@ -119,13 +119,17 @@ class BaseSessionPool:
            # Clean expired sessions
            self._cleanup_expired()
            
+            # TODO: Change session tracking to use unique IDs instead of model paths
+            # This would allow multiple instances of the same model
+            
            # Check if session exists and is valid
            if model_path in self._sessions:
                session_info = self._sessions[model_path]
                session_info.last_used = time.time()
                return session_info.session
                
-            # Check if we can create new session
+            # TODO: Modify session limit check to count instances per model path
+            # Rather than total sessions across all models
            if len(self._sessions) >= self._max_size:
                raise RuntimeError(
                    f"Maximum number of sessions reached ({self._max_size}). "
--- a/api/src/services/tts_service.py
+++ b/api/src/services/tts_service.py
@ -50,6 +50,10 @@ class TTSService:
            try:
                # Handle stream finalization
                if is_last:
+                    # Skip format conversion for raw audio mode
+                    if not output_format:
+                        return np.array([], dtype=np.float32)
+                    
                    return await AudioService.convert_audio(
                        np.array([0], dtype=np.float32),  # Dummy data for type checking
                        24000,
@ -229,7 +233,7 @@ class TTSService:
        try:
            # Use streaming generator but collect all chunks
            async for chunk in self.generate_audio_stream(
-                text, voice, speed, output_format=None
+                text, voice, speed,  # Default to WAV for raw audio
            ):
                if chunk is not None:
                    chunks.append(chunk)
--- a/assets/docs-screenshot.png
+++ b/assets/docs-screenshot.png
--- a/assets/webui-screenshot.png
+++ b/assets/webui-screenshot.png
--- a/examples/assorted_checks/benchmarks/benchmark_tts_rtf.py
+++ b/examples/assorted_checks/benchmarks/benchmark_tts_rtf.py
@ -68,7 +68,7 @@ def main():
    # Initialize system monitor
    monitor = SystemMonitor(interval=1.0)  # 1 second interval
    # Set prefix for output files (e.g. "gpu", "cpu", "onnx", etc.)
-    prefix = "cpu"
+    prefix = "gpu"
    # Generate token sizes
    if "gpu" in prefix:
        token_sizes = generate_token_sizes(
--- a/examples/assorted_checks/benchmarks/output_data/cpu_benchmark_results_rtf.json
+++ b/examples/assorted_checks/benchmarks/output_data/cpu_benchmark_results_rtf.json
--- a/examples/assorted_checks/benchmarks/output_data/cpu_benchmark_stats_rtf.txt
+++ b/examples/assorted_checks/benchmarks/output_data/cpu_benchmark_stats_rtf.txt
@ -1,23 +1,23 @@
 === Benchmark Statistics (with correct RTF) ===

-Total tokens processed: 1800
-Total audio generated (s): 568.53
-Total test duration (s): 306.02
-Average processing rate (tokens/s): 5.75
-Average RTF: 0.55
-Average Real Time Speed: 1.81
+Total tokens processed: 1500
+Total audio generated (s): 427.90
+Total test duration (s): 10.84
+Average processing rate (tokens/s): 133.35
+Average RTF: 0.02
+Average Real Time Speed: 41.67

 === Per-chunk Stats ===

-Average chunk size (tokens): 600.00
-Min chunk size (tokens): 300
-Max chunk size (tokens): 900
-Average processing time (s): 101.89
-Average output length (s): 189.51
+Average chunk size (tokens): 300.00
+Min chunk size (tokens): 100
+Max chunk size (tokens): 500
+Average processing time (s): 2.13
+Average output length (s): 85.58

 === Performance Ranges ===

-Processing rate range (tokens/s): 5.30 - 6.26
-RTF range: 0.51x - 0.59x
-Real Time Speed range: 1.69x - 1.96x
+Processing rate range (tokens/s): 102.04 - 159.74
+RTF range: 0.02x - 0.03x
+Real Time Speed range: 33.33x - 50.00x

--- a/examples/assorted_checks/benchmarks/output_data/gpu_benchmark_results_rtf.json
+++ b/examples/assorted_checks/benchmarks/output_data/gpu_benchmark_results_rtf.json
@ -2,616 +2,240 @@
  "results": [
    {
      "tokens": 150,
-      "processing_time": 2.36,
-      "output_length": 45.9,
-      "rtf": 0.05,
-      "elapsed_time": 2.44626
+      "processing_time": 1.18,
+      "output_length": 43.7,
+      "rtf": 0.03,
+      "elapsed_time": 1.20302
    },
    {
      "tokens": 300,
-      "processing_time": 4.94,
-      "output_length": 96.425,
-      "rtf": 0.05,
-      "elapsed_time": 7.46073
+      "processing_time": 2.27,
+      "output_length": 86.75,
+      "rtf": 0.03,
+      "elapsed_time": 3.49958
    },
    {
      "tokens": 450,
-      "processing_time": 8.94,
-      "output_length": 143.1,
-      "rtf": 0.06,
-      "elapsed_time": 16.55036
+      "processing_time": 3.49,
+      "output_length": 125.9,
+      "rtf": 0.03,
+      "elapsed_time": 7.03862
    },
    {
      "tokens": 600,
-      "processing_time": 19.78,
-      "output_length": 188.675,
-      "rtf": 0.1,
-      "elapsed_time": 36.69352
+      "processing_time": 4.64,
+      "output_length": 169.325,
+      "rtf": 0.03,
+      "elapsed_time": 11.71062
    },
    {
      "tokens": 750,
-      "processing_time": 19.89,
-      "output_length": 236.7,
-      "rtf": 0.08,
-      "elapsed_time": 56.77695
+      "processing_time": 5.07,
+      "output_length": 212.3,
+      "rtf": 0.02,
+      "elapsed_time": 16.83186
    },
    {
      "tokens": 900,
-      "processing_time": 16.83,
-      "output_length": 283.425,
-      "rtf": 0.06,
-      "elapsed_time": 73.8079
+      "processing_time": 6.66,
+      "output_length": 258.0,
+      "rtf": 0.03,
+      "elapsed_time": 23.54135
    }
  ],
  "system_metrics": [
    {
-      "timestamp": "2025-01-06T00:43:20.888295",
-      "cpu_percent": 36.92,
-      "ram_percent": 68.6,
-      "ram_used_gb": 43.6395263671875,
-      "gpu_memory_used": 7022.0,
-      "relative_time": 0.09646010398864746
+      "timestamp": "2025-01-30T05:06:38.733338",
+      "cpu_percent": 0.0,
+      "ram_percent": 18.6,
+      "ram_used_gb": 5.284908294677734,
+      "gpu_memory_used": 1925.0,
+      "relative_time": 0.039948463439941406
    },
-    {
-      "timestamp": "2025-01-06T00:43:21.983741",
-      "cpu_percent": 22.29,
-      "ram_percent": 68.6,
-      "ram_used_gb": 43.642677307128906,
-      "gpu_memory_used": 7021.0,
-      "relative_time": 1.1906661987304688
-    },
-    {
-      "timestamp": "2025-01-06T00:43:23.078293",
-      "cpu_percent": 27.39,
-      "ram_percent": 68.6,
-      "ram_used_gb": 43.61421203613281,
-      "gpu_memory_used": 7190.0,
-      "relative_time": 2.264479160308838
-    },
-    {
-      "timestamp": "2025-01-06T00:43:24.151445",
-      "cpu_percent": 20.28,
-      "ram_percent": 68.6,
-      "ram_used_gb": 43.65406036376953,
-      "gpu_memory_used": 7193.0,
-      "relative_time": 3.349093198776245
-    },
-    {
-      "timestamp": "2025-01-06T00:43:25.237021",
-      "cpu_percent": 23.03,
-      "ram_percent": 68.6,
-      "ram_used_gb": 43.647274017333984,
-      "gpu_memory_used": 7191.0,
-      "relative_time": 4.413560628890991
-    },
-    {
-      "timestamp": "2025-01-06T00:43:26.300255",
-      "cpu_percent": 23.62,
-      "ram_percent": 68.6,
-      "ram_used_gb": 43.642295837402344,
-      "gpu_memory_used": 7185.0,
-      "relative_time": 5.484973430633545
-    },
-    {
-      "timestamp": "2025-01-06T00:43:27.377319",
-      "cpu_percent": 46.04,
-      "ram_percent": 68.7,
-      "ram_used_gb": 43.7291374206543,
-      "gpu_memory_used": 7178.0,
-      "relative_time": 6.658120632171631
-    },
-    {
-      "timestamp": "2025-01-06T00:43:28.546053",
-      "cpu_percent": 29.79,
-      "ram_percent": 68.7,
-      "ram_used_gb": 43.73202133178711,
-      "gpu_memory_used": 7177.0,
-      "relative_time": 7.725939035415649
-    },
-    {
-      "timestamp": "2025-01-06T00:43:29.613327",
-      "cpu_percent": 18.19,
-      "ram_percent": 68.8,
-      "ram_used_gb": 43.791343688964844,
-      "gpu_memory_used": 7177.0,
-      "relative_time": 8.800285577774048
-    },
-    {
-      "timestamp": "2025-01-06T00:43:30.689097",
-      "cpu_percent": 22.29,
-      "ram_percent": 68.9,
-      "ram_used_gb": 43.81514358520508,
-      "gpu_memory_used": 7176.0,
-      "relative_time": 9.899119853973389
-    },
-    {
-      "timestamp": "2025-01-06T00:43:31.786443",
-      "cpu_percent": 32.59,
-      "ram_percent": 68.9,
-      "ram_used_gb": 43.834510803222656,
-      "gpu_memory_used": 7189.0,
-      "relative_time": 11.042734384536743
-    },
-    {
-      "timestamp": "2025-01-06T00:43:32.929720",
-      "cpu_percent": 42.48,
-      "ram_percent": 68.8,
-      "ram_used_gb": 43.77507019042969,
-      "gpu_memory_used": 7192.0,
-      "relative_time": 12.117269277572632
-    },
-    {
-      "timestamp": "2025-01-06T00:43:34.004481",
-      "cpu_percent": 26.33,
-      "ram_percent": 68.8,
-      "ram_used_gb": 43.77891159057617,
-      "gpu_memory_used": 7192.0,
-      "relative_time": 13.19830870628357
-    },
-    {
-      "timestamp": "2025-01-06T00:43:35.086024",
-      "cpu_percent": 26.53,
-      "ram_percent": 68.8,
-      "ram_used_gb": 43.77515411376953,
-      "gpu_memory_used": 7192.0,
-      "relative_time": 14.29457426071167
-    },
-    {
-      "timestamp": "2025-01-06T00:43:36.183496",
-      "cpu_percent": 40.33,
-      "ram_percent": 68.9,
-      "ram_used_gb": 43.81095886230469,
-      "gpu_memory_used": 7192.0,
-      "relative_time": 15.402768850326538
-    },
-    {
-      "timestamp": "2025-01-06T00:43:37.290635",
-      "cpu_percent": 43.6,
-      "ram_percent": 69.0,
-      "ram_used_gb": 43.87236022949219,
-      "gpu_memory_used": 7190.0,
-      "relative_time": 16.574281930923462
-    },
-    {
-      "timestamp": "2025-01-06T00:43:38.462164",
-      "cpu_percent": 85.74,
-      "ram_percent": 69.0,
-      "ram_used_gb": 43.864280700683594,
-      "gpu_memory_used": 6953.0,
-      "relative_time": 17.66074824333191
-    },
-    {
-      "timestamp": "2025-01-06T00:43:39.548295",
-      "cpu_percent": 23.88,
-      "ram_percent": 68.8,
-      "ram_used_gb": 43.75236129760742,
-      "gpu_memory_used": 4722.0,
-      "relative_time": 18.739423036575317
-    },
-    {
-      "timestamp": "2025-01-06T00:43:40.626692",
-      "cpu_percent": 59.24,
-      "ram_percent": 68.7,
-      "ram_used_gb": 43.720741271972656,
-      "gpu_memory_used": 4723.0,
-      "relative_time": 19.846031665802002
-    },
-    {
-      "timestamp": "2025-01-06T00:43:41.733597",
-      "cpu_percent": 41.74,
-      "ram_percent": 68.4,
-      "ram_used_gb": 43.53546142578125,
-      "gpu_memory_used": 4722.0,
-      "relative_time": 20.920310020446777
-    },
-    {
-      "timestamp": "2025-01-06T00:43:42.808191",
-      "cpu_percent": 35.43,
-      "ram_percent": 68.3,
-      "ram_used_gb": 43.424468994140625,
-      "gpu_memory_used": 4726.0,
-      "relative_time": 22.00457763671875
-    },
-    {
-      "timestamp": "2025-01-06T00:43:43.891669",
-      "cpu_percent": 43.81,
-      "ram_percent": 68.2,
-      "ram_used_gb": 43.38311004638672,
-      "gpu_memory_used": 4727.0,
-      "relative_time": 23.08402943611145
-    },
-    {
-      "timestamp": "2025-01-06T00:43:44.971246",
-      "cpu_percent": 58.13,
-      "ram_percent": 68.0,
-      "ram_used_gb": 43.27970886230469,
-      "gpu_memory_used": 4731.0,
-      "relative_time": 24.249765396118164
-    },
-    {
-      "timestamp": "2025-01-06T00:43:46.137626",
-      "cpu_percent": 66.76,
-      "ram_percent": 68.0,
-      "ram_used_gb": 43.23844528198242,
-      "gpu_memory_used": 4731.0,
-      "relative_time": 25.32853865623474
-    },
-    {
-      "timestamp": "2025-01-06T00:43:47.219723",
-      "cpu_percent": 27.95,
-      "ram_percent": 67.8,
-      "ram_used_gb": 43.106136322021484,
-      "gpu_memory_used": 4734.0,
-      "relative_time": 26.499221563339233
-    },
-    {
-      "timestamp": "2025-01-06T00:43:48.386913",
-      "cpu_percent": 73.13,
-      "ram_percent": 67.7,
-      "ram_used_gb": 43.049781799316406,
-      "gpu_memory_used": 4736.0,
-      "relative_time": 27.592528104782104
-    },
-    {
-      "timestamp": "2025-01-06T00:43:49.480407",
-      "cpu_percent": 50.63,
-      "ram_percent": 67.6,
-      "ram_used_gb": 43.007415771484375,
-      "gpu_memory_used": 4736.0,
-      "relative_time": 28.711266040802002
-    },
-    {
-      "timestamp": "2025-01-06T00:43:50.599220",
-      "cpu_percent": 92.36,
-      "ram_percent": 67.5,
-      "ram_used_gb": 42.9685173034668,
-      "gpu_memory_used": 4728.0,
-      "relative_time": 29.916289567947388
-    },
-    {
-      "timestamp": "2025-01-06T00:43:51.803667",
-      "cpu_percent": 83.07,
-      "ram_percent": 67.5,
-      "ram_used_gb": 42.96232986450195,
-      "gpu_memory_used": 4724.0,
-      "relative_time": 31.039498805999756
-    },
-    {
-      "timestamp": "2025-01-06T00:43:52.927208",
-      "cpu_percent": 90.61,
-      "ram_percent": 67.5,
-      "ram_used_gb": 42.96202850341797,
-      "gpu_memory_used": 5037.0,
-      "relative_time": 32.2381911277771
-    },
-    {
-      "timestamp": "2025-01-06T00:43:54.128135",
-      "cpu_percent": 89.47,
-      "ram_percent": 67.5,
-      "ram_used_gb": 42.94692611694336,
-      "gpu_memory_used": 5085.0,
-      "relative_time": 33.35147500038147
+    {
+      "timestamp": "2025-01-30T05:06:39.774003",
+      "cpu_percent": 13.37,
+      "ram_percent": 18.6,
+      "ram_used_gb": 5.2852630615234375,
+      "gpu_memory_used": 3047.0,
+      "relative_time": 1.0883615016937256
    },
-    {
-      "timestamp": "2025-01-06T00:43:55.238967",
-      "cpu_percent": 60.01,
-      "ram_percent": 67.4,
-      "ram_used_gb": 42.88222122192383,
-      "gpu_memory_used": 5085.0,
-      "relative_time": 34.455963373184204
-    },
-    {
-      "timestamp": "2025-01-06T00:43:56.344164",
-      "cpu_percent": 62.12,
-      "ram_percent": 67.3,
-      "ram_used_gb": 42.81411361694336,
-      "gpu_memory_used": 5083.0,
-      "relative_time": 35.549962282180786
-    },
-    {
-      "timestamp": "2025-01-06T00:43:57.437566",
-      "cpu_percent": 53.56,
-      "ram_percent": 67.3,
-      "ram_used_gb": 42.83011245727539,
-      "gpu_memory_used": 5078.0,
-      "relative_time": 36.66783380508423
-    },
-    {
-      "timestamp": "2025-01-06T00:43:58.554923",
-      "cpu_percent": 80.27,
-      "ram_percent": 67.3,
-      "ram_used_gb": 42.79304504394531,
-      "gpu_memory_used": 5069.0,
-      "relative_time": 37.77330660820007
+    {
+      "timestamp": "2025-01-30T05:06:40.822449",
+      "cpu_percent": 13.68,
+      "ram_percent": 18.7,
+      "ram_used_gb": 5.303462982177734,
+      "gpu_memory_used": 3040.0,
+      "relative_time": 2.12058687210083
    },
-    {
-      "timestamp": "2025-01-06T00:43:59.660456",
-      "cpu_percent": 72.33,
-      "ram_percent": 67.2,
-      "ram_used_gb": 42.727474212646484,
-      "gpu_memory_used": 5079.0,
-      "relative_time": 38.885955810546875
+    {
+      "timestamp": "2025-01-30T05:06:41.854375",
+      "cpu_percent": 15.39,
+      "ram_percent": 18.7,
+      "ram_used_gb": 5.306262969970703,
+      "gpu_memory_used": 3326.0,
+      "relative_time": 3.166278600692749
    },
-    {
-      "timestamp": "2025-01-06T00:44:00.773867",
-      "cpu_percent": 59.29,
-      "ram_percent": 66.9,
-      "ram_used_gb": 42.566131591796875,
-      "gpu_memory_used": 5079.0,
-      "relative_time": 39.99704432487488
-    },
-    {
-      "timestamp": "2025-01-06T00:44:01.884399",
-      "cpu_percent": 43.52,
-      "ram_percent": 66.5,
-      "ram_used_gb": 42.32980728149414,
-      "gpu_memory_used": 5079.0,
-      "relative_time": 41.13008522987366
-    },
-    {
-      "timestamp": "2025-01-06T00:44:03.018905",
-      "cpu_percent": 84.46,
-      "ram_percent": 66.5,
-      "ram_used_gb": 42.28911590576172,
-      "gpu_memory_used": 5087.0,
-      "relative_time": 42.296770095825195
-    },
-    {
-      "timestamp": "2025-01-06T00:44:04.184606",
-      "cpu_percent": 88.27,
-      "ram_percent": 66.3,
-      "ram_used_gb": 42.16263961791992,
-      "gpu_memory_used": 5091.0,
-      "relative_time": 43.42832589149475
+    {
+      "timestamp": "2025-01-30T05:06:42.900882",
+      "cpu_percent": 14.19,
+      "ram_percent": 18.8,
+      "ram_used_gb": 5.337162017822266,
+      "gpu_memory_used": 2530.0,
+      "relative_time": 4.256956577301025
    },
-    {
-      "timestamp": "2025-01-06T00:44:05.315967",
-      "cpu_percent": 80.91,
-      "ram_percent": 65.9,
-      "ram_used_gb": 41.9491081237793,
-      "gpu_memory_used": 5089.0,
-      "relative_time": 44.52496290206909
+    {
+      "timestamp": "2025-01-30T05:06:43.990792",
+      "cpu_percent": 12.63,
+      "ram_percent": 18.8,
+      "ram_used_gb": 5.333805084228516,
+      "gpu_memory_used": 3331.0,
+      "relative_time": 5.2854602336883545
    },
-    {
-      "timestamp": "2025-01-06T00:44:06.412298",
-      "cpu_percent": 41.68,
-      "ram_percent": 65.6,
-      "ram_used_gb": 41.72716522216797,
-      "gpu_memory_used": 5090.0,
-      "relative_time": 45.679444313049316
+    {
+      "timestamp": "2025-01-30T05:06:45.019134",
+      "cpu_percent": 14.14,
+      "ram_percent": 18.8,
+      "ram_used_gb": 5.334297180175781,
+      "gpu_memory_used": 3332.0,
+      "relative_time": 6.351738929748535
    },
-    {
-      "timestamp": "2025-01-06T00:44:07.566964",
-      "cpu_percent": 73.02,
-      "ram_percent": 65.5,
-      "ram_used_gb": 41.64710998535156,
-      "gpu_memory_used": 5091.0,
-      "relative_time": 46.81710481643677
+    {
+      "timestamp": "2025-01-30T05:06:46.085997",
+      "cpu_percent": 12.78,
+      "ram_percent": 18.8,
+      "ram_used_gb": 5.351467132568359,
+      "gpu_memory_used": 2596.0,
+      "relative_time": 7.392607688903809
    },
-    {
-      "timestamp": "2025-01-06T00:44:08.704786",
-      "cpu_percent": 75.38,
-      "ram_percent": 65.4,
-      "ram_used_gb": 41.59475326538086,
-      "gpu_memory_used": 5097.0,
-      "relative_time": 47.91444158554077
+    {
+      "timestamp": "2025-01-30T05:06:47.127113",
+      "cpu_percent": 14.7,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.367542266845703,
+      "gpu_memory_used": 3341.0,
+      "relative_time": 8.441826343536377
    },
-    {
-      "timestamp": "2025-01-06T00:44:09.802745",
-      "cpu_percent": 42.21,
-      "ram_percent": 65.2,
-      "ram_used_gb": 41.45526885986328,
-      "gpu_memory_used": 5111.0,
-      "relative_time": 49.04095649719238
+    {
+      "timestamp": "2025-01-30T05:06:48.176033",
+      "cpu_percent": 13.47,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.361263275146484,
+      "gpu_memory_used": 3339.0,
+      "relative_time": 9.500520706176758
    },
-    {
-      "timestamp": "2025-01-06T00:44:10.928231",
-      "cpu_percent": 65.65,
-      "ram_percent": 64.4,
-      "ram_used_gb": 40.93437957763672,
-      "gpu_memory_used": 5111.0,
-      "relative_time": 50.14311861991882
+    {
+      "timestamp": "2025-01-30T05:06:49.234332",
+      "cpu_percent": 15.84,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.3612213134765625,
+      "gpu_memory_used": 3339.0,
+      "relative_time": 10.53744649887085
+    },
+    {
+      "timestamp": "2025-01-30T05:06:50.271159",
+      "cpu_percent": 14.89,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.379688262939453,
+      "gpu_memory_used": 3646.0,
+      "relative_time": 11.570110321044922
+    },
+    {
+      "timestamp": "2025-01-30T05:06:51.303841",
+      "cpu_percent": 15.71,
+      "ram_percent": 19.0,
+      "ram_used_gb": 5.390773773193359,
+      "gpu_memory_used": 3037.0,
+      "relative_time": 12.60651707649231
    },
    {
-      "timestamp": "2025-01-06T00:44:12.036249",
-      "cpu_percent": 28.51,
-      "ram_percent": 64.1,
-      "ram_used_gb": 40.749881744384766,
-      "gpu_memory_used": 5107.0,
-      "relative_time": 51.250269651412964
-    },
+      "timestamp": "2025-01-30T05:06:52.340383",
+      "cpu_percent": 15.46,
+      "ram_percent": 19.0,
+      "ram_used_gb": 5.389518737792969,
+      "gpu_memory_used": 3319.0,
+      "relative_time": 13.636165380477905
+    },
    {
-      "timestamp": "2025-01-06T00:44:13.137586",
-      "cpu_percent": 52.99,
-      "ram_percent": 64.2,
-      "ram_used_gb": 40.84278869628906,
-      "gpu_memory_used": 5104.0,
-      "relative_time": 52.34805965423584
-    },
+      "timestamp": "2025-01-30T05:06:53.370342",
+      "cpu_percent": 13.12,
+      "ram_percent": 19.0,
+      "ram_used_gb": 5.391136169433594,
+      "gpu_memory_used": 3320.0,
+      "relative_time": 14.67578935623169
+    },
    {
-      "timestamp": "2025-01-06T00:44:14.235248",
-      "cpu_percent": 34.55,
-      "ram_percent": 64.1,
-      "ram_used_gb": 40.7873420715332,
-      "gpu_memory_used": 5097.0,
-      "relative_time": 53.424301862716675
-    },
+      "timestamp": "2025-01-30T05:06:54.376175",
+      "cpu_percent": 14.98,
+      "ram_percent": 19.0,
+      "ram_used_gb": 5.390045166015625,
+      "gpu_memory_used": 3627.0,
+      "relative_time": 15.70747685432434
+    },
    {
-      "timestamp": "2025-01-06T00:44:15.311386",
-      "cpu_percent": 39.07,
-      "ram_percent": 64.2,
-      "ram_used_gb": 40.860008239746094,
-      "gpu_memory_used": 5091.0,
-      "relative_time": 54.50679922103882
-    },
+      "timestamp": "2025-01-30T05:06:55.441172",
+      "cpu_percent": 13.45,
+      "ram_percent": 19.0,
+      "ram_used_gb": 5.394947052001953,
+      "gpu_memory_used": 1937.0,
+      "relative_time": 16.758784770965576
+    },
    {
-      "timestamp": "2025-01-06T00:44:16.393626",
-      "cpu_percent": 31.02,
-      "ram_percent": 64.3,
-      "ram_used_gb": 40.884307861328125,
-      "gpu_memory_used": 5093.0,
-      "relative_time": 55.57431173324585
-    },
+      "timestamp": "2025-01-30T05:06:56.492442",
+      "cpu_percent": 17.03,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.361682891845703,
+      "gpu_memory_used": 3041.0,
+      "relative_time": 17.789713144302368
+    },
    {
-      "timestamp": "2025-01-06T00:44:17.461449",
-      "cpu_percent": 24.53,
-      "ram_percent": 64.3,
-      "ram_used_gb": 40.89955520629883,
-      "gpu_memory_used": 5070.0,
-      "relative_time": 56.660638093948364
-    },
+      "timestamp": "2025-01-30T05:06:57.523536",
+      "cpu_percent": 13.76,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.360996246337891,
+      "gpu_memory_used": 3321.0,
+      "relative_time": 18.838542222976685
+    },
    {
-      "timestamp": "2025-01-06T00:44:18.547558",
-      "cpu_percent": 19.93,
-      "ram_percent": 64.3,
-      "ram_used_gb": 40.92641830444336,
-      "gpu_memory_used": 5074.0,
-      "relative_time": 57.736456871032715
-    },
+      "timestamp": "2025-01-30T05:06:58.572158",
+      "cpu_percent": 15.94,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.3652801513671875,
+      "gpu_memory_used": 3323.0,
+      "relative_time": 19.86689043045044
+    },
    {
-      "timestamp": "2025-01-06T00:44:19.624478",
-      "cpu_percent": 15.63,
-      "ram_percent": 64.3,
-      "ram_used_gb": 40.92564392089844,
-      "gpu_memory_used": 5082.0,
-      "relative_time": 58.81701683998108
-    },
+      "timestamp": "2025-01-30T05:06:59.600551",
+      "cpu_percent": 15.67,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.363399505615234,
+      "gpu_memory_used": 3630.0,
+      "relative_time": 20.89712619781494
+    },
    {
-      "timestamp": "2025-01-06T00:44:20.705184",
-      "cpu_percent": 29.86,
-      "ram_percent": 64.4,
-      "ram_used_gb": 40.935394287109375,
-      "gpu_memory_used": 5082.0,
-      "relative_time": 59.88701677322388
-    },
+      "timestamp": "2025-01-30T05:07:00.631315",
+      "cpu_percent": 15.37,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.3663482666015625,
+      "gpu_memory_used": 3629.0,
+      "relative_time": 22.01374316215515
+    },
    {
-      "timestamp": "2025-01-06T00:44:21.775463",
-      "cpu_percent": 43.55,
-      "ram_percent": 64.4,
-      "ram_used_gb": 40.9350471496582,
-      "gpu_memory_used": 5080.0,
-      "relative_time": 60.96005439758301
-    },
-    {
-      "timestamp": "2025-01-06T00:44:22.847939",
-      "cpu_percent": 26.66,
-      "ram_percent": 64.4,
-      "ram_used_gb": 40.94179916381836,
-      "gpu_memory_used": 5076.0,
-      "relative_time": 62.02673673629761
-    },
-    {
-      "timestamp": "2025-01-06T00:44:23.914337",
-      "cpu_percent": 22.46,
-      "ram_percent": 64.4,
-      "ram_used_gb": 40.9537467956543,
-      "gpu_memory_used": 5076.0,
-      "relative_time": 63.10581707954407
-    },
-    {
-      "timestamp": "2025-01-06T00:44:24.993313",
-      "cpu_percent": 28.07,
-      "ram_percent": 64.4,
-      "ram_used_gb": 40.94577407836914,
-      "gpu_memory_used": 5076.0,
-      "relative_time": 64.18998432159424
-    },
-    {
-      "timestamp": "2025-01-06T00:44:26.077028",
-      "cpu_percent": 26.1,
-      "ram_percent": 64.4,
-      "ram_used_gb": 40.98012161254883,
-      "gpu_memory_used": 5197.0,
-      "relative_time": 65.28782486915588
-    },
-    {
-      "timestamp": "2025-01-06T00:44:27.175228",
-      "cpu_percent": 35.17,
-      "ram_percent": 64.6,
-      "ram_used_gb": 41.0831184387207,
-      "gpu_memory_used": 5422.0,
-      "relative_time": 66.37566781044006
-    },
-    {
-      "timestamp": "2025-01-06T00:44:28.265025",
-      "cpu_percent": 55.14,
-      "ram_percent": 64.9,
-      "ram_used_gb": 41.25740432739258,
-      "gpu_memory_used": 5512.0,
-      "relative_time": 67.48023676872253
-    },
-    {
-      "timestamp": "2025-01-06T00:44:29.367776",
-      "cpu_percent": 53.84,
-      "ram_percent": 65.0,
-      "ram_used_gb": 41.36682891845703,
-      "gpu_memory_used": 5616.0,
-      "relative_time": 68.57096815109253
-    },
-    {
-      "timestamp": "2025-01-06T00:44:30.458301",
-      "cpu_percent": 33.42,
-      "ram_percent": 65.3,
-      "ram_used_gb": 41.5602912902832,
-      "gpu_memory_used": 5724.0,
-      "relative_time": 69.66709041595459
-    },
-    {
-      "timestamp": "2025-01-06T00:44:31.554329",
-      "cpu_percent": 50.81,
-      "ram_percent": 65.5,
-      "ram_used_gb": 41.66044616699219,
-      "gpu_memory_used": 5827.0,
-      "relative_time": 70.75874853134155
-    },
-    {
-      "timestamp": "2025-01-06T00:44:32.646414",
-      "cpu_percent": 34.34,
-      "ram_percent": 65.6,
-      "ram_used_gb": 41.739715576171875,
-      "gpu_memory_used": 5843.0,
-      "relative_time": 71.86718988418579
-    },
-    {
-      "timestamp": "2025-01-06T00:44:33.754223",
-      "cpu_percent": 44.32,
-      "ram_percent": 66.0,
-      "ram_used_gb": 42.005794525146484,
-      "gpu_memory_used": 5901.0,
-      "relative_time": 72.95793795585632
-    },
-    {
-      "timestamp": "2025-01-06T00:44:34.848852",
-      "cpu_percent": 48.36,
-      "ram_percent": 66.5,
-      "ram_used_gb": 42.3160514831543,
-      "gpu_memory_used": 5924.0,
-      "relative_time": 74.35109186172485
-    },
-    {
-      "timestamp": "2025-01-06T00:44:36.240235",
-      "cpu_percent": 58.06,
-      "ram_percent": 67.5,
-      "ram_used_gb": 42.95722198486328,
-      "gpu_memory_used": 5930.0,
-      "relative_time": 75.47581958770752
-    },
-    {
-      "timestamp": "2025-01-06T00:44:37.363208",
-      "cpu_percent": 46.82,
-      "ram_percent": 67.6,
-      "ram_used_gb": 42.97764587402344,
-      "gpu_memory_used": 6364.0,
-      "relative_time": 76.58708119392395
-    },
+      "timestamp": "2025-01-30T05:07:01.747500",
+      "cpu_percent": 13.79,
+      "ram_percent": 18.9,
+      "ram_used_gb": 5.367362976074219,
+      "gpu_memory_used": 3620.0,
+      "relative_time": 23.05113124847412
+    },
    {
-      "timestamp": "2025-01-06T00:44:38.474408",
-      "cpu_percent": 50.93,
-      "ram_percent": 67.9,
-      "ram_used_gb": 43.1597900390625,
-      "gpu_memory_used": 6426.0,
-      "relative_time": 77.6842532157898
+      "timestamp": "2025-01-30T05:07:02.784828",
+      "cpu_percent": 10.16,
+      "ram_percent": 19.1,
+      "ram_used_gb": 5.443946838378906,
+      "gpu_memory_used": 1916.0,
+      "relative_time": 24.08937978744507
    }
  ],
-  "test_duration": 82.49591493606567
+  "test_duration": 26.596059799194336
 }
--- a/examples/assorted_checks/benchmarks/output_data/gpu_benchmark_stats_rtf.txt
+++ b/examples/assorted_checks/benchmarks/output_data/gpu_benchmark_stats_rtf.txt
@ -1,23 +1,23 @@
 === Benchmark Statistics (with correct RTF) ===

 Total tokens processed: 3150
-Total audio generated (s): 994.22
-Total test duration (s): 73.81
-Average processing rate (tokens/s): 49.36
-Average RTF: 0.07
-Average Real Time Speed: 15.00
+Total audio generated (s): 895.98
+Total test duration (s): 23.54
+Average processing rate (tokens/s): 133.43
+Average RTF: 0.03
+Average Real Time Speed: 35.29

 === Per-chunk Stats ===

 Average chunk size (tokens): 525.00
 Min chunk size (tokens): 150
 Max chunk size (tokens): 900
-Average processing time (s): 12.12
-Average output length (s): 165.70
+Average processing time (s): 3.88
+Average output length (s): 149.33

 === Performance Ranges ===

-Processing rate range (tokens/s): 30.33 - 63.56
-RTF range: 0.05x - 0.10x
-Real Time Speed range: 10.00x - 20.00x
+Processing rate range (tokens/s): 127.12 - 147.93
+RTF range: 0.02x - 0.03x
+Real Time Speed range: 33.33x - 50.00x

--- a/examples/assorted_checks/benchmarks/output_plots/cpu_processing_time_rtf.png
+++ b/examples/assorted_checks/benchmarks/output_plots/cpu_processing_time_rtf.png
--- a/examples/assorted_checks/benchmarks/output_plots/cpu_realtime_factor_rtf.png
+++ b/examples/assorted_checks/benchmarks/output_plots/cpu_realtime_factor_rtf.png
--- a/examples/assorted_checks/benchmarks/output_plots/cpu_system_usage_rtf.png
+++ b/examples/assorted_checks/benchmarks/output_plots/cpu_system_usage_rtf.png
--- a/examples/assorted_checks/benchmarks/output_plots/gpu_processing_time_rtf.png
+++ b/examples/assorted_checks/benchmarks/output_plots/gpu_processing_time_rtf.png
--- a/examples/assorted_checks/benchmarks/output_plots/gpu_realtime_factor_rtf.png
+++ b/examples/assorted_checks/benchmarks/output_plots/gpu_realtime_factor_rtf.png
--- a/examples/assorted_checks/benchmarks/output_plots/gpu_system_usage_rtf.png
+++ b/examples/assorted_checks/benchmarks/output_plots/gpu_system_usage_rtf.png