SNUAILAB gains GPU capacity while controlling costs with OCI
The global vision AI company runs GPU-intensive workloads on OCI after a proof of concept estimates savings of up to 50% on compute vs. other clouds.
South Korea | High Technology
“As demand for 3D inspection and AI inference grows, Oracle Cloud Infrastructure helps us add GPU capacity while keeping compute and data transfer costs more predictable.”
About SNUAILAB
SNUAILAB is a machine vision software company that develops SMVT, a node-based platform for building 2D, 3D, and AI inspection pipelines.
Business problem: Fixed GPU capacity constrained peak workloads
SMVT was running in an on-premises environment with fixed GPU capacity. As demand for 3D inspection and AI inference grew, SNUAILAB needed weeks to procure and install additional hardware to support peak workloads, creating bottlenecks.
The company required a cloud option that could expand GPU capacity for peak workloads while making storage and data transfer costs more predictable.
Solutions and results: SNUAILAB increases flexibility and cost efficiency with OCI
After a proof of concept comparing the compute, GPU, MySQL, and networking capabilities and cost efficiency of two major cloud providers, SNUAILAB selected Oracle Cloud Infrastructure (OCI) for its price-performance, GPU virtual machine options, managed Oracle MySQL HeatWave, and networking capabilities.
SNUAILAB now uses OCI GPU virtual machines and Oracle Cloud Infrastructure Kubernetes Engine to support SMVT workloads. The proof of concept validated the business case for running SMVT on OCI, estimating that compute costs would be up to 50% lower with OCI compared with the other cloud provider, and outbound data transfer costs would be approximately one-quarter of the cost of the competing cloud.
About the customer
SNUAILAB develops SMVT, a node-based 2D, 3D, and AI vision inspection platform for image processing, measurement, 3D analysis, AI inference, and Python scripting.