Continuously monitor performance improvements and degradation
At Stream HPC we build high performance software and libraries for 15 years. We therefore know the problem of triaging the cause of an unexpected performance change. It could be a minor code-change or a driver-update, but it’ll always take time to get back in control. For years we had our in-house tools, and found our customers wanted these tools too.
We have worked very hard to make these tools usable for for software makers and AI-developers. Currently we’re testing it on our own projects and with select alpha customers only. Watch this space or follow us on LinkedIn.
Some highlights
High-level
Full software benchmarking shows your QoS promises are still true for all supported GPUs.
Historic values
Graphs of historic runs and easily comparing what changed.
Device support promises
Test and Benchmark more devices than you have, on our private cloud and public cloud partners.
Functional level benchmarking
Go as deep as required to find what cause performance fluctuations.
Finding the variables that matter
Let the system find out if the drivers, OS or code caused a speed-up or speed-down.
Embedded and phones
Select embedded devices and phones are available for fully automated benchmarking and testing.
Stay updated what will be next
Go to https://newsletter.streamhpc.net/forms/nfrm_BzLMaLXv to subscribe to our newsletter and be the first to know. If you want to be informed with latency, follow us on LinkedIn.