Comino has launched the Grando H100 Server, specifically designed for AI and high-performance computing (HPC) workloads.

At the heart of the Grando H100 is the powerful AMD Threadripper PRO 7995WX processor, which boasts 96 cores and 192 threads utilizing Zen 4 architecture.

The server is further enhanced with dual Nvidia H100 NVL GPUs, providing a total of 188GB of GPU memory, which is crucial for advanced AI computations.

Supporting up to 2TB of DDR5 memory, the Grando H100 is ideal for extensive data processing tasks, making it suitable for large-scale computational needs.

To maintain stability during heavy workloads, the server features an advanced liquid cooling system, which operates quietly and includes four 1600W redundant power supplies.

This cooling design incorporates a centralized water block and quick-disconnect fittings to ensure optimal thermal performance.

The server is particularly well-suited for AI model training and inference, as well as data-intensive simulations in sectors like finance and healthcare.

Testing by StorageReview highlighted the Grando H100's strengths in AI-driven tasks and CPU-intensive workloads, particularly in multi-core rendering and compression.

Pros of the Grando H100 include exceptional computing power, advanced cooling capabilities, and expansive memory and GPU options.

This potential underperformance in specific traditional rendering tasks is a noted con, particularly against specialized systems.

As enterprises increasingly seek powerful solutions for managing massive datasets, the demand for servers like the Grando H100 is on the rise, aligning with trends towards specialized hardware for AI and HPC.

However, recent benchmarks indicate that while the Grando H100 excels in AI-specific computations, it may underperform in traditional rendering tasks compared to systems optimized for those applications.