This repository contains a structured, automated testing and development environment optimized explicitly for running TensorFlow pipelines on an NVIDIA GeForce RTX 5080 (Blackwell Architecture) under Windows Subsystem for Linux (WSL 2) and Docker Desktop.
| File Name | Execution Type | Primary Role / Description |
|---|---|---|
validate.bash |
Core Script | Basic infrastructure check (Validates Host Driver, WSL, and Docker Engine). |
hardware_unit_test.py |
Python Logic | Silent, binary unit test. Run by health.bash to execute a fast tensor calculation. |
health.bash |
Wrapper Script | Triggers hardware_unit_test.py and returns a clean, warning-free [PASS] or [FAIL] plaque. |
heavy_model_benchmark.py |
Python Logic | Heavy ResNet workflow script utilizing data streaming generators to fully stress the GPU. |
heavy.bash |
Wrapper Script | Triggers heavy_model_benchmark.py to push Alienware Command Center gauges high without memory crashes. |
run.bash |
Interactive Tool | Launches a persistent, Blackwell-optimized Jupyter Notebook development workspace. |
Run this chained combination right after a major Windows driver update or system reboot to verify your virtualization paths are alive and healthy. The && gate ensures that the deep tensor calculations will only execute if your core components pass inspection.
./validate.bash && ./health.bash- Execution Time: ~5 seconds.
- Visual Output: A clean, warning-free configuration plaque displaying a clear
[PASS]status indicator. - Ollama Coexistence: Safe. This operation executes in a temporary container sandbox and releases hardware lines instantly.
Run this pipeline whenever you want to see your RTX 5080 work under a heavy computational load, push fan curves, or capture high-utilization screenshots inside your Alienware Command Center dashboard.
./heavy.bash- Execution Time: ~45 seconds (10 epochs, 100 calculation steps each).
- VRAM Protection: Enforces memory-streaming arrays to protect your hardware pipeline against fragmentation and memory-exhaustion crashes.
Run this script only when you are sitting down to actively open up a web browser tab, write custom machine learning algorithms, or work visually.
./run.bash- Action Required: Copy the resulting
http://127.0.0...link from your command console and paste it into your web browser. - Safe Shutdown: When finished coding, select
File -> Close and Haltinside your browser tab, then pressCtrl + Ctwice inside your terminal prompt to safely release your VRAM resources back to the host system.
To prevent the terminal windows from freezing up or printing walls of low-level driver timing chatter (gpu_timer.cc) or compiler optimization adjustments (+ptx85), all scripts inside this suite automatically deploy these optimized background flags:
TF_FORCE_GPU_ALLOW_GROWTH=true: Restricts TensorFlow from immediately hoarding 100% of your card's memory cache, allowing local LLMs like Ollama to comfortably run alongside your code.TF_CPP_MIN_LOG_LEVEL=3: Completely blocks standard informational, warning, and basic error logs, leaving your screenshots pristine.NVIDIA_DISABLE_REQUIRE=true: Prevents legacy container runtime check blocks from stalling initialization threads when encountering newer Blackwell core setups.--ipc=host&--ulimit memlock=-1: Expands container memory allocation thresholds globally to eliminate multi-threaded data loading bottlenecks.