Introducing the VyperLab Lead Partner Early Access Program

VyperCore technology, which offers up to 80% reduction in CPU power consumption with its five fold memory-safe acceleration of managed-language applications, is now available in VyperLab, a cloud-based evaluation platform showcasing its innovative compute accelerator technology.

VyperCore created VyperLab and its Lead Partner Early Access Program to foster ongoing collaboration for optimizing and accelerating in-production data center applications.

VyperLab is a next generation compute evaluation platform which allows frictionless evaluation of 3-5x acceleration of existing in-production application code. It offers an in-cloud demonstration of the hardware memory safety capabilities of VyperCore’s architecture and this cloud-hosting also ensures partners’ code and data confidentiality. VyperLab is targeted at early adopters of next generation general-purpose high-performance compute, specifically partners who have significant amounts of in-production Python code which would benefit from significant acceleration, for improved performance or efficiency.

Current state-of-the-art memory allocation management in CPython consumes up to 80% of a processor’s cycles when taking into consideration secondary impacts such as cache pollution. VyperCore technology acceleration is achieved by replacing the software-based memory allocation management, a common feature of all managed-language runtimes, with dedicated hardware inside the CPU. Moving this process to hardware gains both performance and efficiency, and additionally frees up the motherboard CPU for the execution of more suitable tasks.

This program is targeted at lead partners who have significant amounts of in-production Python code which would benefit from significant acceleration, for improved performance or efficiency.

The early access program enables Lead Partners to seamlessly run their existing production Python application code (without modification) on an isolated FPGA instance, deploying VyperCore's prototype technology as a high-performance general-purpose compute accelerator. Metrics built into VyperLab will produce a detailed analysis of the execution of the application both with and without VyperCore's technology, highlighting the acceleration potential. Lead Partners will be able to collaborate directly with the VyperCore architecture team on future development to further optimize the architectural features of the VyperCore Booth processor with the potential of gaining greater performance increases.

The evaluation process is completed over a few steps:

1. Request FPGA capability and service quota increase from cloud vendor
2. Provide cloud account credentials to Vypercore
3. Execute the configuration terraform script, which will:
   1. Provision a new instance of the FPGA
   2. Bootstrap the system installing dependencies such as docker
   3. Download the Docker container and install the launch script
4. Attach to the box and run the launch script
   1. In the process this will load the FPGA with the correct image
5. Use the vypython scripts to run the same code on VyperCore accelerated technology and non-accelerated processor model
   1. Can also launch a Jupyter server and attach to it to run code
6. Send results file to VyperCore
7. Review performance results

Register your interest

Partners who wish to join the lead partner program and have the opportunity to benchmark their current code performance against future acceleration should complete their details in the form below or email lead.partner@vypercore.com for more details.