Workspace Weekly: SPECviewperf GPU Benchmark Meets Dynamic Platform Metrics
August 19, 2025
This Workspace Weekly, we are introducing a fresh way to bring high-fidelity GPU performance data directly into your Login Enterprise dashboards. Our SPECviewperf GPU Benchmark workload now runs full 3D graphical benchmarks, captures the results, and uploads them to Platform Metrics, enhanced with automatic time drift correction and secure API key injection for a smoother, safer experience.
The result? You get a visual, correlated view of GPU performance right alongside your EUX scores, VSImax results, application timings, and session metrics, so you can see exactly what’s happening under the hood when user experience changes.
Figure 1: Login Enterprise is driving the virtual user to run a Creo viewset in SpecViewPerf.
The Challenge
When teams run high-intensity workloads like SPECviewperf, they often want to see how GPU performance lines up with the rest of the environment. For example, login times, application responsiveness, and even hypervisor activity. But without a shared clock between test results and platform metrics, aligning those datasets can be messy, especially in multi-site or cloud scenarios. Add in the need to protect API keys or configuration tokens, and it’s easy to see why GPU testing at scale can become a manual, error-prone task.
Our Solution: Enhanced SPECviewperf Workload
Our updated SPECviewperf workload now:
- Runs the full benchmark, parses results, and streams FPS data into Platform Metrics with the correct metric IDs, grouping, and display names for immediate visibility.
- Auto-corrects for clock drift by contacting your Login Enterprise appliance before each run, measuring round-trip time, and applying that drift factor to every data point. Timestamps are automatically aligned to UTC, so every metric matches perfectly with other measurements, and there will be consistency with measurements timestamps in Login Enterprise graphing!
- Uses secure Application credentials to inject variables, like your API access token (sensitive variable!) and viewset name, at runtime, without ever hardcoding them into the script. This keeps secrets safe, reduces setup steps, and works for any scenario where you need to pull in third-party API keys or credentials from a vault.
Don’t be fooled by the Application Credentials being named Username and Password; technically, any value can be stored in here. Think of it as a shortcut to quickly editing a variable in your Login Enterprise workload scripts, especially sensitive ones like API tokens!
Why This Matters
SPECviewperf is already a world-class graphics benchmark, giving you objective measurements across popular GPU-intensive workloads. Now, by streaming those results into Platform Metrics, you can:
- Correlate performance across layers – See GPU framerates alongside VSImax events, EUX dips, or application slowdowns in the same chart.
- Spot cause and effect faster – When user experience degrades, instantly check if it coincided with a GPU load spike, driver update, or hypervisor contention.
- Baseline and compare – Track GPU performance over time, across hardware generations, or before/after configuration changes.
- Extend to anything – This same approach works for other custom metrics, from hypervisor counters to network latency, all plotted in the same interface.
Think of it as giving your performance data a universal translator; everything speaks the same time-synced language, so insights appear faster and with more confidence.
A Quick Example
Imagine a VSImax event triggers during a load test. Instead of switching tools or exporting data, you open Platform Metrics and see:
- EUX score dropped from 9.1 to 7.5
- GPU framerates from SPECviewperf viewset Maya (engine) dipped at the exact same moment
- CPU utilization and hypervisor metrics show a matching spike
With these in one place, you don’t just know something happened, you know why.
Getting Started
- Grab the updated workload from our GitHub repo.
- Check the Knowledge Base for step-by-step setup instructions: SPECviewperf GPU Benchmark Workload.
- Import the workload into Login Enterprise and in the Login Enterprise web UI, set your secure Application credentials for the SpecViewPerf viewset name (set as Username) and your Login Enterprise API Token (set as Password).
Now the API token has end-to-end encryption while still allowing the Platform Metrics to be securely uploaded to your Login Enterprise appliance
- Run a test and watch your GPU data flow into Platform Metrics – perfectly time-aligned and ready to analyze alongside everything else.
Figure 2: SpecViewPerf Platform Metrics displayed over an hour timeframe.
Where to See It
In test results, you’ll find your GPU framerate data grouped with other performance indicators, with granular time breakdowns available for trend analysis. For extra context, pull it up next to EUX or application timings and explore the correlations.
Figure 3: Platform Metrics displayed over an hour timeframe.
Wrapping Up
This isn’t just about running SPECviewperf, it’s about unlocking a deeper view of your environment by pairing GPU benchmarks with the rest of your performance story. With automatic time drift correction, every datapoint stays accurate and trustworthy.
Secure credential injection keeps sensitive values out of scripts and in the right hands. Meanwhile, the Platform Metrics framework gives you the flexibility to capture and visualize virtually any performance measurement you need.
Explore the repo, run it in your environment, and see firsthand how much more insight you gain when your GPU metrics live alongside the rest of your performance data.
If you’re part of the LVSI CUSTOMERS workspace, join the #workspace-weekly Slack channel to share results, ask questions, or suggest new metric ideas.
Workspace Weekly
