SC94 — Dual-Socket 384-Core CFD HPC Node with 128 GB Memory and 3.76 TB NVMe Storage for ANSYS Fluent

The SC94 node introduces an unprecedented level of extreme compute density to our lineup. Powered by two AMD EPYC 9965 processors, this dual-socket configuration deploys an astonishing 384 physical cores and 768 threads with a 3.7 GHz boost clock. Paired with a focused 128 GB of memory and two 1.88 TB NVMe SSDs (roughly 3.76 TB of total usable capacity), this node is a highly specialized powerhouse built for one specific purpose: hyper-scaling.

By packing 384 cores alongside 128 GB of RAM and 24 total memory channels, SC94 is engineered for simulations that fit within standard memory footprints but require the absolute maximum number of processors to slash time-to-solution. It is the ultimate environment for slicing moderately sized meshes into incredibly small, ultra-fast partitions.

Specifications

Compared to our high-memory nodes, SC94 flips the equation entirely. Instead of offering huge memory capacities to hold colossal cases, it offers extreme core volume for medium-sized cases. With roughly 0.33 GB of RAM per core, this node pushes the limits of MPI scaling, aggressively driving down the number of cells each core must calculate to achieve lightning-fast iterations.

What this node is built for

SC94 is the extreme compute-density tier. It suits:

• Hyper-scaled steady-state runs — Cases that fit within 128 GB but need to be solved in minutes rather than hours by distributing the workload across nearly 400 cores.

• Compute-heavy physics on moderate meshes — Highly complex reactions, combustion, or multi-species flows on 50 to 100 million cell meshes that demand immense mathematical throughput.

• Rapid design iteration — When your mesh isn't overwhelmingly large, but your deadline is impossibly tight, SC94 provides the brute-force parallel width to power through.

• Bandwidth-bound solves — 24 memory channels across two sockets feed the 384 cores rapidly, ensuring that despite the massive core count, memory throughput remains exceptionally high.

Understanding the dual-socket layout

A two-socket machine is not simply a bigger one-socket machine. Each processor has its own memory controllers and its own local RAM, and a core reaching data attached to the other socket pays a latency penalty crossing the interconnect between them.

For Fluent, this is highly manageable and worth knowing about. Partitioning the mesh so each socket works mostly on locally-held data — and pinning MPI ranks accordingly — keeps that cross-socket traffic low. Run without any NUMA awareness and you can lose some of the bandwidth advantage the second socket provides. We configure this for you as part of the setup, so in practice it's a detail we handle rather than one you have to manage.

Choosing SC94

Choose SC94 when:

• Your case fits comfortably inside 128 GB of RAM (typically under ~100-120 million cells), and

• You want the absolute maximum parallel speed available to minimize turnaround time, or

• You are running compute-heavy physics that require massive processing power per cell.

If your case needs more than 128 GB of RAM, you must step up to our 256 GB or 384 GB nodes (like SC92 or SC91). If you are generating tens of terabytes of transient data, you need a high-storage node (like SC93). But if your bottleneck is pure compute time on a moderately sized mesh, SC94 throws 384 cores at the problem to crush solve times.

Storage: high-speed NVMe

Two 1.88 TB NVMe SSDs provide roughly 3.76 TB of incredibly fast storage. While it lacks the massive 22+ TB bulk capacity of our SATA or 6-drive NVMe arrays, this configuration is perfectly sized for steady-state workloads, design optimization loops, and short transients. The NVMe drives ensure that when Fluent writes autosaves or final case data, it happens in seconds, keeping your 384 cores focused on the math rather than waiting on disk operations.

How parallel solving works on this node

When you run Fluent in parallel on SC94, the solver partitions your mesh across 384 cores. This is where this node shines: if you run a 100-million cell mesh, each core is only responsible for roughly 260,000 cells. This extremely low cell-per-core ratio means each rank finishes its calculations near-instantaneously before exchanging boundary data via MPI. Backed by the massive 24-channel memory bandwidth of a dual-socket layout, this hyper-partitioning strategy yields dramatic speedups for cases with strong parallel scaling characteristics.

Why rent instead of buy

How it works — you're in full control

SC94 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 SC94 node.

2. Set up your simulation — load your geometry and case, configure your solver settings, and run ANSYS Fluent with all 384 cores at your command.

3. Run and monitor in real time — start, pause, adjust, and watch the solution as it progresses, with complete control throughout.

4. Keep your results — post-process on the node or download your solution and data files whenever you're ready.

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

Frequently asked questions

How is SC94 different from the SC91 node? While SC91 is a flagship compute node with 192 cores, SC94 doubles that to an incredible 384 cores. However, SC94 trades away memory and storage to achieve this (128 GB RAM and 3.76 TB storage vs SC91’s 384 GB RAM and 22.5 TB storage). SC94 is a specialist tool for smaller meshes that need extreme computational speed.

Is 128 GB of RAM a bottleneck for 384 cores? It acts as a filter for the type of case you can run. 128 GB limits you to meshes of roughly 100-120 million cells. Because you have 384 cores, you are forced into a very aggressive partitioning ratio (under 300k cells per core). As long as your case scales well at this ratio, it is incredibly fast. If your physics require more memory per core, our 256 GB or 384 GB nodes are required.

Will my case actually scale to 384 cores? Fluent scales exceptionally well, but scaling efficiency depends on your physics and mesh size. If your mesh is too small (e.g., 5 million cells), 384 cores will create too much MPI communication overhead. SC94 is best utilized for dense, compute-heavy models in the 50 to 100+ million cell range where the sheer math outweighs the communication overhead.

Is my data kept confidential? Yes.

Get started with SC94

If your models fit within 128 GB of memory but you want to slash your solve times by leveraging the absolute highest core count available on a single node, SC94 is designed for exactly that.

Contact MR CFD to run your ANSYS Fluent case on SC94 — and we'll help you unleash 384 physical cores on your simulations.