Focus Deep Learning Suite
This tool can get pipelines running in minutes. It guides you to install Anaconda and then add and test TensorFlow, PyTorch, and Jupyter Notebook toolkits as needed. Develop with industry-standard PyCharm installed from the JetBrains Toolbox.
We continuously validate, curate, and pin the Deep Learning Suite, graphics drivers, CUDA, cuDNN, TensorRT, and OptiX to always provide excellent performance, reliability, and compatibility across ML and GPU workloads. Improvements and maintenance are provided seamlessly through normal software updates.
Benchmarks
We tested using Stable Diffusion 1.5 and Qwen3-0-6b tests. The TSMC “4N” lithography and fifth-generation Tensor cores make the 50-series GPUs faster, cooler, and more efficient.
| Model | Zr GEN 1 (18.0") | Zr GEN 1 (18.0") | M2 GEN 6 (16.0") | M2 GEN 5 (17.3") | M2 GEN 5 (17.3") | M2 GEN 5 (15.6") |
|---|---|---|---|---|---|---|
| GPU | RTX 5090 M | RTX 5080 M | RTX 5070 Ti M | RTX 4090 M | RTX 4080 M | RTX 4070 M |
| Stable Diffusion 1.5 | 143 Images | 116 Images | 92 Images | 127 Images | 118 Images | 80 Images |
| Qwen03-0-6b | 544 Gen | 534 Gen | 482 Gen | 579 Gen | 528 Gen | 459 Gen |
| Video RAM | 24 GB | 16 GB | 12 GB | 16 GB | 12 GB | 8 GB |
GPU accelerated performance is generally around 30x greater than CPU-only ML pipelines on the same device. Stable Diffusion and Qwen3-0-6b are found on the United Compute score board in their respective tabs.