SC96 — Single-Socket 16-Core High-Frequency CFD Node with 128 GB Memory and ~938 TB Storage for ANSYS Fluent

The SC96 node completely redefines what a CFD storage and utility node looks like. While our other tiers focus on massive parallel scaling, SC96 solves the other half of the CFD equation: extreme data retention and high-speed post-processing. A single AMD EPYC 9135 processor provides 16 cores and 32 threads with a rapid 4.3 GHz boost clock. Memory sits at 128 GB, but the centerpiece of this machine is its storage array: forty 23.44 TB SATA SSDs, combining for roughly 938 TB — nearly a full Petabyte — of usable capacity.

This is a highly specialized configuration. With an incredible 8 GB of RAM per core and blistering single-thread speed, it is purpose-built as an active data warehouse. It allows you to run ultra-long transient workflows, house multi-year DOE simulation databases, or perform intense post-processing and animation rendering on colossal datasets without ever having to move files across a network.

Specifications

Compared to the rest of the fleet, SC96 operates in a class of its own regarding storage. While nodes like the SC99 or SC92 offer an already massive 22 TB, SC96 delivers more than forty times that capacity. It retains a highly capable 16-core compute foundation, providing 8 GB of RAM per core — perfect for opening massive result files, rendering heavy visuals, and running complex single-thread tasks.

What this node is built for

SC96 is the extreme-capacity data lake and post-processing tier. It suits:

• Massive simulation data warehouses — Archiving thousands of full-scale parametric sweeps (DOEs) or years of project data natively on a fast, live machine rather than cold tape storage.

• Ultra-long transient campaigns — Running continuous, months-long time-step or acoustic (CAA) simulations where the solver writes terabytes of data.

• Heavy post-processing and visualization — The 4.3 GHz boost clock and 128 GB of RAM make opening, manipulating, and rendering massive 3D results in CFD-Post or EnSight incredibly fast and fluid.

• Compute-heavy physics on smaller meshes — Running heavy combustion, DPM, or multi-phase physics on moderate meshes (under ~30 million cells) where 8 GB per core ensures the solver never starves for memory.

The single-socket and massive storage advantage

Typically, handling nearly a Petabyte of storage requires a dedicated network-attached storage (NAS) appliance, which throttles your workflow with network latency every time you load a result file. SC96 brings the storage directly into the compute node.

Because it relies on a single processor, there is no cross-socket latency (NUMA overhead). The single AMD EPYC 9135 handles the immense I/O traffic from the 40-drive array natively, and its 12 memory channels feed the 16 cores flawlessly. You get the simplicity of a single-socket workstation wrapped around the storage volume of a small data center.

Choosing SC96

Choose SC96 when:

• You are generating or storing hundreds of terabytes of result data that must remain live and instantly accessible, or

• Your primary bottleneck is the time it takes to read, load, and post-process massive simulation files, or

• You are running long-duration transient workflows on moderately sized meshes that simply cannot run out of disk space, or

• You need a dedicated, highly responsive node to render complex animations from massive datasets.

If your primary goal is slashing solve times on a 150-million cell mesh, you should choose a high-core-count tier like the SC92 or SC91. SC96 is not designed for massive MPI parallelism; it is designed to hold, manage, and process the immense amount of data that high-core-count simulations produce.

Storage: petabyte-scale capacity

Forty 23.44 TB SATA SSDs working in parallel completely eliminate storage constraints. While a single SATA drive does not match NVMe for peak sequential speed, striping 40 of them together provides massive aggregate throughput. For CFD, this means you can dump massive transient autosaves, load colossal case files in seconds, and execute complex read/write operations without a hitch. Nearly 938 TB of solid-state storage gives you the freedom to run continuous, heavy-write simulations indefinitely.

How parallel solving and processing works on this node

When running Fluent on SC96, the solver partitions your mesh across 16 extremely fast cores. While 16 cores represents our lowest parallel width, the 4.3 GHz boost clock and 8 GB of RAM per core ensure that smaller, heavily complex meshes run brilliantly. More importantly, when it comes time to analyze the data, those same high-frequency cores dramatically accelerate single-threaded post-processing tasks, such as calculating complex derived variables, tracing particle paths, and rendering high-resolution contour animations.

Why rent instead of buy

How it works — you're in full control

SC96 is yours to drive directly. You connect to the node over a secure remote desktop connection and run everything yourself, exactly as you would on a local workstation:

1. Connect via remote desktop — you get full remote access to the SC96 node.

2. Set up your simulation or data workflow — load your geometry, run your solvers, or aggregate terabytes of results from other runs into the massive SSD array.

3. Run and monitor in real time — start, pause, adjust, and post-process your solutions with complete control throughout.

4. Keep your results live — utilize the ~938 TB storage to hold your data indefinitely during your rental, making it the ultimate hub for large-scale analysis.

You have full authority over the node for the duration of your rental: we provide the hardware, the petabyte-scale storage, and the support, and you run the simulations your way.

Frequently asked questions

Why only 16 cores on a node with nearly 1,000 TB of storage? SC96 is an I/O and post-processing specialist, not a massive parallel compute engine. Tasks like loading massive datasets, rendering visuals, and writing continuous transient files depend on massive disk capacity, memory size (128 GB), and high single-core clock speeds (4.3 GHz) rather than a high core count.

Is SATA storage too slow for this much data? While a single SATA drive is slower than NVMe, an array of 40 SSDs accessed in parallel provides incredible aggregate bandwidth. This easily handles the periodic autosaves and heavy read/write loads of Fluent without the extreme cost footprint of a petabyte of NVMe storage.

Can I use this node purely for post-processing data generated on other nodes? Yes. SC96 is exceptionally well-suited as a central data warehouse and visualization hub for results generated by our higher-core-count compute tiers (like SC91 or SC93).

How large a mesh can it handle? With 128 GB of RAM, it can physically hold meshes up to roughly 100 million cells. However, with 16 cores, it is computationally best suited for meshes under 30 million cells if you are actively solving, or for post-processing massive meshes that have already been solved on wider nodes.

Is my data kept confidential? Yes.

Get started with SC96

If your simulations are generating more data than standard workstations can possibly hold, or you are tired of network latency choking your post-processing workflows, SC96 provides the petabyte-scale capacity and single-thread speed to handle your largest datasets effortlessly.

Contact MR CFD to run your ANSYS Fluent case on SC96 — and we'll help you leverage ultimate storage capacity for your simulations.