zystem joins NVIDIA Inception Program to help AI and ML engineers maximize the efficiency of GPU-bound workloads
We’re excited to share that zystem, Inc.—the company behind zymtrace—has officially joined NVIDIA Inception, a program designed to support and accelerate startups at the forefront of AI and high-performance computing.
This milestone underscores a core principle that’s guided us from the beginning: truly uncover the unknown unknowns in GPU-bound workloads.
GPU monitoring ≠ GPU profiling.
GPU utilization, temperature, power—those metrics are easy to collect, but they don’t explain the inner workings of the GPU or why it’s idling. So we started deeper, closer to the hardware. zymtrace profiles down to CUDA kernels, disassembles CUDA SASS mnemonics, exposes GPU stall reasons, and correlates it all back to the CPU traces that launched the kernels. That’s where real insight lives—and that’s where we started, not with NVML metrics.
Joining NVIDIA Inception gives us access to technical resources, go-to-market support, and a global community of AI-focused engineers, researchers, and founders. It’s a meaningful step forward for zystem as we continue building the next generation of performance tooling for GPU and AI-accelerated workloads.
zymtrace is a lightweight, distributed GPU and CPU continuous profiler purpose-built for heterogeneous, high-performance computing workloads. As AI reshapes the world, introspection at this level isn’t optional—it’s foundational to modern infrastructure. Low-level GPU profiling is the most efficient way to truly understand why GPUs are idling.
We go beyond utilization charts, because real observability for AI-native systems means understanding how and why things happen, not just when.
We’re excited to learn, collaborate, and contribute to this thriving ecosystem.
The NVIDIA logo is a trademark and/or registered trademark of NVIDIA Corporation in the U.S. and other countries
