Optimizing Performance to Minimize Compute Spend
Legacy codebases power the compute-heavy workloads of many large corporations. However, these codebases are often not optimized for today’s computing environments, leading to significant operational expenses. Vorticity’s solutions have demonstrated success in dramatically reducing turnaround times and computing costs, both in the cloud and on-premises.
Here, we present two case studies illustrating how we collaborated with world-leading financial services corporations to accelerate their computing workloads, decrease turnaround times and reduce compute spend.
Quantum Hardware Simulator
A leading global financial institution partnered with Vorticity to optimize their existing code for GPU acceleration. The objective was to boost processing speeds on single GPUs while ensuring seamless scalability across multiple GPUs with minimal performance loss, crucial for high-performance quantum heuristic simulations.
Vorticity focused on kernel optimization and memory allocation, delivering a solution compatible with both older (Nvidia T4) and latest-generation (Nvidia A100) GPUs.
Within a 4-week timeframe, Vorticity achieved a remarkable 2.5x increase in computing speed, empowering the client with faster prototyping and more rigorous evaluation capabilities.
Link to work: https://github.com/jpmorganchase/QOKit
Randomized Numerical Linear Algebra
Randomized Numerical Linear Algebra (RandNLA) is revolutionizing big-data applications with innovative techniques like data sketching and subspace embedding. A leading multinational bank partnered with Vorticity to harness this power by optimizing the PARLA software package for GPU acceleration.
The goal: significantly boost computing speeds on both single and multi-GPU setups. Vorticity delivered, achieving remarkable speedups and ensuring high-quality results across diverse matrix types (dense, s-sparse, low-rank).
Exceeding expectations, Vorticity surpassed the client’s target of 2x-5x speedup, ultimately achieving an astounding 59x improvement within 8 weeks.
Our case studies highlight not only the exceptional speed improvements but also the strategic advantage gained through customized solutions. As the demand for high-performance computing continues to grow, Vorticity stands ready to empower organizations to stay ahead of the curve.
Contact us today to discover how our expertise can elevate your operations and drive substantial cost savings.
See more at Vorticity.xyz
