Rack validation and optimization software that generates tokens per second output under AI workloads — closing the gap between L11/L12 testing and the revenue performance operators depend on.
Deployed & validated with industry leaders
Key Features
XPerf’s proprietary algorithms and automated workflow compress full rack validation from weeks to hours or days.
Combines inference and training AI workloads with XPerf’s proprietary test suite to stress GPUs, memory, interconnects, network fabric, and thermal systems, exposing hardware and firmware issues that conventional methods miss.
The only validation tool that identifies the root cause of detected issues and prescribes specific remediation and optimization steps, cutting engineering triage time and getting racks ready faster.
Every rack ships with a ClusterReady validation report documenting tokens per second output across inference and training workloads — providing a performance baseline for revenue generation.
Workflow
ClusterReady is designed for rapid on-premise deployment in minutes in supported environments.
Stress-test racks using a combination of inference and training AI workloads alongside XPerf’s proprietary test suite — exercising GPUs, memory, interconnects, network fabric, and thermal systems under conditions that mirror live production.
ClusterReady measures tokens per second output, giving a concrete, revenue-linked performance baseline for every rack. When issues are detected, XPerf’s proprietary algorithms identify the root cause and prescribe specific remediation or optimization steps.
Proven Results
Deployed with MiTAC Computing on AMD Instinct™ MI325X racks
Get Started
Commercially available now. Designed for rapid on-premise deployment in minutes in supported environments.
macOS: Right-click the app and select Open to bypass Gatekeeper on first launch.
About
Founded by ex-Intel engineers with extensive experience deploying clusters with thousands of accelerators, XPerf Inc. is an AI infrastructure software company specializing in GPU cluster performance validation and optimization, taking an AI-native approach to solving cluster operation problems. Its flagship product, XPerf ClusterReady, runs 10× faster than typical L11/L12 validation workflows using AI workloads and XPerf’s proprietary algorithms.