DX 3.0 Advanced Software System
DX 3.0 is an advanced server software stack that provides a virtual disaggregation capability for GPU servers. The base layer includes a server software operating system and an isolated resource partitioning environment, wrapped into a machine deployment mechanism driven by the Drut Software Platform (DSP).
Virtual Disaggregation
DX 3.0
The base layer includes a server software operating system and an isolated resource partitioning environment, wrapped into a machine deployment mechanism driven by the Drut Software Platform (DSP). GPU servers are created by the Drut Fabric Manager (DFM), building vPODs out of the available machine partitions.
vPOD
A vPOD is constructed out of a number of CPU slices, memory slices, operating systems, GPUs and NICs deployed with a user’s AI/ML model of choice. vPODs can exist within the server and across servers over the network fabric, whether Ethernet, InfiniBand, or using Drut’s photonic fabric. For the user, we glue all this together with our AI/ML workbench software and automate the configuration process. It is the easy button for AI/ML.
Features and Benefits of Virtual Disaggregation
Virtual Disaggregation
Key Benefits
- GPU Isolation: Allow users to run isolated GPU workloads in multi-GPU servers
- Resource Configurability: Efficiently allocate GPUs in multi-GPU servers
- Flexible Resource Groups: Divide multi-GPU server and combine GPUs from different machines
- Existing Server Augmentation: Combines old and new GPUs into virtual pods, optimizes GPU usage
- Diurnal Resource Grouping: Lets you adjust GPU configurations for time of day workloads
- Cost efficiency: ML/AI utilization at a significantly lower cost
Features
- Hardware and Software: Complete system solution
- GPU Server Isolation: Dynamically isolate GPUs in multi-GPU server to create physically isolated resource groupings
- User Isolation: Partitions GPU machines into specific user/workload instances
- GPU Allocations: Add GPUs to existing vPODs
- Off the shelf servers : Compatibility with industry standards server, GPUs and NICs
Why Choose DX 3.0
- Affordability: Cut down on AI deployment costs significantly, making advanced AI accessible to your business without breaking the bank
- Scalability: Easily scale your AI operations as your needs grow, without the hassle of major hardware investments
- Efficiency: Optimize your GPU usage, ensuring every dollar spent translates to actual performance gains
- Versatility: Whether you’re deploying real-time analytics, machine learning models, or conducting AI research, DX 3.0 is equipped to handle a wide range of applications
Contact Drut For More Information
Coming Soon
Look out for lots more coming from Drut soon!