ScaleWorX provides integrated, data-centric infrastructure solutions for AI and data-intensive computing workloads. Their platforms, like ScalePOD® AI, accelerate AI model training and large-scale data analysis by optimizing data access and simplifying system deployment.
Funding
Funding not disclosed
Founders
Product
Problem
The increasing scale of data in fields like AI, big data analytics, and scientific research necessitates IT infrastructure that is designed around data rather than solely around compute resources. Traditional architectures often create data silos and require complex data movement, hindering efficient access and processing for data-intensive workloads.
Solution
ScaleWorX provides integrated infrastructure solutions for data-intensive computing and AI workloads, focusing on a data-centric design. Their offerings include specialized hardware and software platforms, such as ScalePOD® AI and ScalePOD® Barreleye™, engineered to accelerate AI model training, inference, and large-scale data analysis. These solutions aim to simplify system deployment, enhance data accessibility, and optimize the entire data lifecycle from acquisition to utilization across on-premises, cloud, and hybrid environments. By leveraging advanced system integration capabilities and expertise in data-centric computing, ScaleWorX enables organizations to achieve higher investment efficiency and better performance for their demanding computational tasks.
Target Audience
The primary customers are organizations and research institutions dealing with large-scale data processing and AI development, including those in HPC, genomics, autonomous driving, and manufacturing sectors.
Features
- **ScalePOD® AI**: A building-block infrastructure solution scalable up to 1024 nodes, optimized for AI workloads.
- **Data-Centric Architecture**: Designed around data and data lifecycle management, moving away from compute-centric HPC designs.
- **Turnkey AI Solutions**: Pre-packaged systems with pre-loaded OS (Ubuntu, RHEL), NVIDIA CUDA, TensorFlow, PyTorch, Llama 2, and Open OnDemand for immediate use.
- **High-Performance Compute Options**: Supports NVIDIA H100 Tensor Core GPUs and NVIDIA Grace Hopper Superchips, integrated with AMD EPYC 9004 or 4th Gen Intel Xeon processors.
- **NVIDIA AI Enterprise Compatibility**: Includes over 100 frameworks, pre-trained models, and development tools optimized for NVIDIA GPUs.
- **ScalePOD® Barreleye™**: An integrated monitoring appliance for DDN EXAScaler® file systems and storage, providing detailed statistics on user, group, application, and job levels.
- **Troubleshooting and Sizing Capabilities**: Barreleye™ facilitates performance monitoring, root cause analysis for storage latency, and capacity planning through historical data analysis.
- **High-Speed Networking**: Supports NVIDIA Quantum 400Gb/s NDR and 200Gb/s HDR InfiniBand switches.
- **High-Performance Storage**: Integrates DDN® A³I™ solutions like the AI400X2, offering scalable capacity and IOPS with NVIDIA GPUDirect® Storage support.
- **System Validation**: Offers rack-level system validation through partnerships with hardware vendors to ensure correct investment decisions.