SC113 — 16-Core CFD HPC Node with 12-Channel Memory Bandwidth for ANSYS Fluent

The SC113 node is built around the one thing CFD solvers are most often starved of: memory bandwidth. It runs a single AMD EPYC 9135 — 16 cores and 32 threads with an all-core boost of 4.25 GHz — on AMD's full server platform, with twelve DDR5 memory channels feeding it. That's six times the memory path of the dual-channel desktop-class nodes in our lineup. Paired with 128 GB of RAM and 2 × 1.88 TB NVMe SSD, it's the tier to choose when your solve spends more time waiting on data than computing it.

Clock speed determines how fast a core executes. Core count determines how many ways a mesh can be split. Bandwidth determines whether those cores actually have data to work on — and on a real CFD case, that's often the binding constraint.

Specifications

Two details matter beyond the core count. First, the 4.25 GHz boost is an all-core figure — the clock the node actually holds with every rank loaded, not a single-core peak that collapses under a full parallel job. Second, twelve memory channels give roughly 600 GB/s of per-socket bandwidth, an order of magnitude beyond what a dual-channel desktop platform delivers. For a memory-bound solver, that's the spec that decides wall-clock time.

What this node is built for

SC113 is the bandwidth tier: server-class memory throughput behind a solid 16-core configuration. It suits:

• Bandwidth-bound solves — coupled solvers, large steady cases, and anything where cores spend more time fetching data than computing on it. This is where SC113 pulls clearly ahead of desktop-platform nodes with similar core counts.

• Cases that scale across 16 cores without starving — on a dual-channel node, 16 cores compete for a narrow path to RAM; here each core has real bandwidth behind it.

• Large meshes with heavy per-cell physics — multiphase, combustion, many transported scalars: cases that move a lot of data per iteration.

• Sustained, long-running jobs — an all-core boost that holds under full load matters more over a multi-day transient run than a peak clock that only appears on one core.

• Transient, LES, and DES runs writing heavy data — NVMe storage keeps large data files and frequent autosaves from slowing down the solve.

Bandwidth vs. clock speed vs. core count: choosing SC113

SC113 has the same core count as our other 16-core tiers but a lower peak clock — 4.25 GHz all-core rather than a 5.7 GHz single-core boost — and vastly more memory bandwidth. Choose SC113 when:

• Your case is memory-bound rather than compute-bound: heavy physics, large per-cell data, or a solver that moves a lot between cores and RAM each iteration, and

• You want a clock that holds steady across all 16 cores through a long run, rather than a high peak on one.

If your case is compute-bound and gated by single-thread speed, a high-clock desktop-platform tier will finish it sooner. If it scales past 16 cores and needs more parallel breadth, our 24-core node offers that with strong bandwidth too. And if capacity is the constraint, our high-memory nodes are the right tool. As a rough guide, Fluent needs about 1 GB of RAM per million cells for a typical steady, single-phase case, so with 128 GB, SC113 handles meshes well beyond what a 32 GB node can hold.

Tell us your cell count, physics, and where your case actually bottlenecks, and we'll point you to the right tier.

Fast NVMe storage

The two 1.88 TB NVMe SSDs give substantially higher read/write throughput than SATA SSDs, plus nearly 3.76 TB of combined capacity. That means faster loading of large case and data files, quicker autosaves, and room to keep a full DOE sweep's output on disk without constantly clearing space.

How parallel solving works on this node

When you run Fluent in parallel, the solver partitions your mesh across all 16 cores (using algorithms such as METIS), each core solving its share of cells and exchanging boundary data each iteration via MPI. On a single node, that communication runs through fast on-board channels rather than a network, keeping parallel overhead low. What sets SC113 apart is what happens between those exchanges: with twelve memory channels, cores aren't queuing for a narrow path to RAM as the partition count rises, so the 16 cores keep working rather than stalling.

Why rent instead of buy

For engineers whose cases are limited by memory throughput rather than raw clock speed, SC113 delivers the bandwidth that desktop-platform nodes can't — and you can still move to a high-clock or many-core tier when a job calls for it.

How it works — you're in full control

SC113 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 SC113 node.

2. Set up your simulation — load your geometry and case, configure your solver settings, and run ANSYS Fluent with all 16 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 SC113 different from our other 16-core nodes? Same core count, but a different platform. SC113 runs on AMD's full server socket with twelve memory channels and a steady all-core boost; our desktop-platform 16-core tiers have two memory channels and a higher single-core peak. SC113 wins on bandwidth-bound cases and long sustained runs; the others win where single-thread speed is the limit.

Why do memory channels matter so much for CFD? Fluent moves a large amount of data between cores and RAM every iteration. When cores outnumber the available memory bandwidth, they sit idle waiting on data no matter how fast their clock is. Twelve channels give each of SC113's 16 cores a genuinely wide path to memory.

Is 4.25 GHz slower than a 5.7 GHz node? On a single thread, yes. But 5.7 GHz figures are single-core peaks — under a full 16-rank parallel load, actual sustained clocks drop. SC113's 4.25 GHz is the all-core figure, held with every core working.

How large a mesh can it handle? With 128 GB of RAM, plan for meshes well into the tens to low hundreds of millions of cells for simpler steady single-phase cases, fewer for transient or multiphase physics (about 1 GB of RAM per million cells is a good starting estimate). Send us your case for a precise answer.

What if I need more cores or memory than this? We offer many-core and high-memory nodes for cases that scale further or need more capacity — we'll recommend the right tier, or a multi-node setup, for your specific case.

Is my data kept confidential? Yes.

Get started with SC113

If your solves are held back by memory throughput rather than clock speed, SC113's twelve memory channels are built for exactly that.

Contact MR CFD to run your ANSYS Fluent case on SC113 — and we'll help you choose between bandwidth, clock speed, and core count for your simulation.