System Requirements¶
Operating Systems Supported¶
Windows 10, 11
Windows Server 2016+
Redhat 7, 8, 9
Ubuntu 18, 20, 22
Debian 10, 11
GPU Driver Support¶
Windows: NVidia driver version 418.96 or newer
Linux: NVidia driver version 418.39 or newer
OpenGL Support¶
OpenGL is a 3D rendering technology required by both the M-Star Pre-Processor and Post.
OpenGL version 3.3+ is required
This OpenGL usually means you will need to have either integrated or discrete graphics hardware and driver installed. Software rendering is not likely to work and not recommended.
If having issues with 3D rendering program startup, eee Troubleshooting M-Star Pre/Post start up for additional information.
NVidia Hardware¶
NVidia NVLink¶
The NVLink is a high-speed GPU interconnect which offers a significantly faster link for multi-GPU systems. When multiple GPUs are in use, NVlink should always be installed between the GPUs to ensure top performance.
NVidia GPUs¶
The M-Star Solver requires GPUs with Compute Capability 3.5 or newer. Most NVidia GPUs produced since 2014 will be compatible. For a complete list of compute capability, refer to the NVidia CUDA GPU Capability List.
Tesla-based Hardware
Whenever possible, we suggest using NVidia Tesla GPUs. This is enterprise-grade equipment intended to be used for heavy computational loads. They feature TCC mode on Windows, ECC memory, and higher memory capacity.
Quadro-based Hardware
These GPUs work fine and can be a cost-effective means to run M-Star.
Geforce-based Hardware
These are consumer-grade GPUs intended primarily for rendering 3D scenes. These are compatible with the solver and can be highly cost-effective with regard to memory and CUDA cores. Multi-GPU use of Geforce hardware is generally limited to Linux.
Note
Multi-GPU jobs with NVidia RTX 3090 are currently only supported on Linux.
Setup¶
Minimum Requirement¶
For smaller simulation domain sizes, a basic setup can be used:
CPU: Dual core or better
Memory: 16GB
Disk Space: 100GB
GPU: NVIDIA Geforce RTX 2060 6GB
Recommended Workstation¶
Tip
For more guidance on selecting GPUs, see Hardware Guide
CPU: Quad core or better
Memory: 128GB+ (recommend 2x the amount of GPU memory)
Disk 1: 1TB SSD operating system drive
Disk 2: 2TB SSD drive for additional simulation scratch space
Display GPU: Any recent/modern Geforce GPU 2GB+ (required on Windows)
Accelerator GPU 1: NVIDIA RTX A6000 or Tesla V100 or A100
Accelerator GPU 2: NVIDIA RTX A6000 or Tesla V100 or A100 (optional)
Bridge Accelerator GPU 1 & 2 with NVLINK
Recommended Servers¶
M-Star is a GPU compute–intensive application that requires NVidia Tesla–based GPUs for optimal performance. Generally speaking, you want the largest capacity GPUs (CUDA cores and GPU memory). On each compute node you should max out the available GPU slots. You must connect the GPU with NVLINK/NVSwitch to provide the maximum amount of inter-GPU bandwidth. Network connections should be selected to provide the maxium available bandwidth, typically using an Infiniband-based network. Multi-node jobs require GPUDirect RDMA support for optimal performance. We suggest speaking with an HPC representative from a reputable vendor to clarify capability and application requirements.
The general guidelines below are a starting point:
OS: Linux (Redhat/Centos and Ubuntu are supported by M-Star).
CPU: Core count must be equal to or greater than the number of GPUs in the node.
Memory: 512GB+. Generally speaking, more is better, although the M-Star solver is typically limited to the memory on the GPU.
Disk: 200GB–1TB per M-Star user attached over network to all compute nodes. This depends on the specific needs of end users. Disk storage requirements vary based on simulation configuration. You may want to look into high performance file systems typically used in HPC setups.
GPU: NVidia V100, A100, or newer GPUs with the maximum amount of per-GPU memory available.
Compute Nodes: Max out the number of GPUs per compute node. New builds should consider 8x or 16x GPUs per compute node.
Bandwidth: Use NVLINK and NVSwitch in order to provide the maximum amount of bandwidth between GPUs on the compute nodes. See more information here.
Compute Node Interconnect: Infiniband with GPUDirect RDMA support (200Gb/s or better). This is required for multi-node compute jobs.
If multi-node jobs need to be supported, the system should be set up to support GPUDirect RDMA, permitting point-to-point GPU-to-GPU communication across the node interconnect without relying on GPU to CPU memory copies. Refer to the vendor documentation concerning this subject. On Mellanox, this requires loading the nv_peer_mem and gdr_copy kernel modules.
Running on CPUs¶
Although CPU execution is still partially supported, it is not recommended. Many features in M-Star, such as UDF technology, are not supported on CPUs. GPU performance provides orders-of-magnitude performance gain.
Tip
Strongly recommend a dedicated GPU.
For additional information related to system requirements, check out these links: