The industry’s first deep learning server to be built by scientists, for scientists



 The Mark 1 is the first deep learning server designed from the ground up by Valkyrie to solve some of the most challenging problems in applied science. It is designed to train, test and adapt machine learning solutions in the blink of an eye.  85X faster than current pc leaders in neural network training and testing,  4.5x faster than common cloud-based Machine learning platforms

- The Mark1 brings multi-GPU performance to your data scientists

- High Performance, Multi-User environment

- Optimized Processors, GPUs, RAM and Storage to meet customer needs

- Convenient. Upon installation, the system is fully capable of running popular Python, R, TensorFlow and other data science training and inference.


Our worry-free installation process adds the capability of the Mark 1 without the headache of traditional hardware procurement. Our team has honed a seamless, white-glove process to ensure we're up and running without issue.

- Solution optimization across HW and SW technologies

     -GPU: architectures, capabilities, virtualization                  middleware


     -Processor: Multi-threading, PCIe lanes, DDR                    channels, etc


     -OS tuning and security hardening (as needed)


- Customer Site Optimization (provide support prior to and during installation)


      -Network: assess networking options for                             deployment

      -SW/HW: assess data sets, performance                             adjustments, HA storage needs, security