OnStak offers a GPU Analytics Application that harnesses the computational power of NVIDIA’s Graphics Processing Units (GPUs) for high-performance computing tasks. I recently had the opportunity to chat with James Schneider about OnStak’s recently released Splunk app that gives you insights into GPU utilization. James gives us more information about the data insights surfaced, correlation capabilities, alerting, and future capabilities such as network fabric.

Transcript

Auto-generated captions, lightly cleaned. Speakers are not separately labeled.

hey everybody we’re out here at Splunk Tech Summit I’ve been talking to James here about on stack and monitoring GPU can you give me some more information about what we’re doing there yeah absolutely Jason so on Stack’s a great Cisco partner for a long time and they recently published uh their first uh application on Splunk base uh that where you can have a centralized view of all the resources associated with your GPU workloads and the host they’re running on as well as uh you know the different jobs that you’re running through those uh the host and the gpus so essentially we’re uh you know it’s the very first app it’s out there on Splunk base um this you can correlate this data to be able to understand you know if the GPU is running hot set alerts so that uh you know if um memory usage gets saturated you can proactively be alerted before it gets to a critical State and uh this is just the first version that’s come out one of the interesting things they’re already in the works with for the next one is correlating the underlying Network fabric uh up to uh you know associated with the gpus and the host that they’re running on so if let’s say uh the network is an issue you can pinpoint that quickly yeah that sounds like a really Incredible use case especially in the age of AI that we’re in where can I go get some more information about this integration see you just uh hop up on Splunk base and and you can either search for on stack or for GPU it’ll pop right up it’s the on stack uh GPU performance analytics application incredible thank you [Music]