RTX 3090
Ampere architecture · 24 GB GPU memory
Operational Research Infrastructure
Heterogeneous
Strixation
Heterogeneous
Local Compute
Meaningful computation. Locally available.
Strixation is IIC's operational heterogeneous local compute and research platform. It combines multiple generations of NVIDIA GPU hardware into one locally controlled Linux environment for experimentation, engineering, development, model execution, data processing, and knowledge-oriented workloads.
Compute should amplify the person doing the thinking.
Six GPUs. Three architectures. One local research platform.
Strixation has moved beyond a planned compute concept. Six heterogeneous NVIDIA GPUs now operate simultaneously under Linux, with deterministic CUDA device mapping and independent CUDA execution verified across every installed compute device.
RTX 3080
Ampere architecture · 10 GB GPU memory
2× Tesla P100
Pascal architecture · 16 GB each
2× GTX 1660 Super
Turing architecture · 6 GB each
~78 GB
Aggregate GPU memory across the current six-device system.
CUDA PASS × 6
CUDA execution independently verified on all six GPUs.
Pascal. Turing. Ampere.
Strixation deliberately combines three NVIDIA compute generations in one research environment.
The objective is not to make unlike hardware appear identical. It is to understand what each resource can contribute, where its capabilities differ, and how meaningful work can be directed toward appropriate locally available computation.
Physical Infrastructure
Built for the workbench.
Strixation is physical IIC infrastructure, assembled and operated locally rather than consumed as an abstract cloud-compute service.
The current platform uses a dedicated high-airflow open-frame chassis, controlled GPU power limits, two PSU power domains, Linux/NVIDIA compute drivers, deterministic device mapping, and remote SSH administration.
Local ownership allows IIC to inspect, modify, test, break, rebuild, and understand the complete compute environment.
Local Control
Own the environment.
Own the environment.
Understand the environment.
Local compute does not eliminate the usefulness of external infrastructure. It provides another resource whose hardware, configuration, evidence, and availability remain directly observable.
IA — Intellectual Amplification
Strixation is being developed around a simple question: how can meaningful computational capacity amplify a person's ability to investigate, learn, test, understand, and create?
IIC uses the term Intellectual Amplification to describe this direction.
IA does not begin with the assumption that computation should replace the person performing the work. Human intention remains the driver. Computation becomes a tool for expanding the scale, speed, depth, or range of what that person can investigate.
Augment. Never replace.
What becomes possible when meaningful computation is locally available?
Strixation is not being assembled merely to possess compute power.
It exists so IIC can ask, test, and revisit a broader question:
What can we learn, build, test, and understand when meaningful computational capacity is locally available?
SPHIROSYSTEM
Computation as a continuity resource.
Rūts demonstrates local observation close to real-world activity. Strixation demonstrates substantial locally controlled computation.
Within the developing SPHIROSYSTEM architecture, Strixation represents a physical compute resource that can eventually participate in cross-compute continuity.
The objective is to allow workloads and capabilities to use appropriate available compute without requiring every edge device to independently carry the entire computational burden.
Observe.
Recognize.
Direct.
Compute.
Return.
Continue.
Established compute. Developing continuity.
Strixation's six-GPU heterogeneous compute environment is operational today.
Cross-device workload orchestration, automated resource scheduling, IA software, and broader SPHIROSYSTEM compute coordination remain development directions and are not presented as completed capability.
What is real now: six heterogeneous GPUs, three NVIDIA architectures, approximately 78 GB aggregate GPU memory, six independently CUDA-qualified compute devices, remote administration, and local ownership and control.