The NameWhat VANA Means
VANA stands for "Voltage Activated Neural Architecture." It describes how the engine thinks about computation: voltage pathways activating across a structure that resembles a neural system, not a sequential pipeline.
A traditional CPU processes instructions one at a time along a single line. It does this very fast, but the model is the same regardless of how many cores you stack. VANA changes the model. Data activates across many paths at once. The hardware does the work simultaneously, not sequentially.
How It WorksThe Runtime Engine
When you write Sanguis code, it doesn't get interpreted line by line. It gets transposed into a "connectome", a graph of neurons and the connections between them. That connectome is a .vana file. VANA owns execution from there. The graph "settles" on the GPU in parallel every tick. No instruction pointer. No sequential step-through. The computation is the settling.
Today, that surface is the GPU. VANA does not emulate a CPU on the GPU. It does not translate instructions. It treats the GPU as the processor and maps the computation to what the silicon can actually do. The CPU does not sync every tick. It observes.
The PipelineFrom Source to Silicon
Every Sanguis program follows the same path from your keyboard to the GPU. Here's what happens at each stage:
You Write .sang
You write Sanguis code using the 9 keywords: circuit, hold, memory, when, but, emit, append, through, wire. This is source code, same as writing Python or C. The difference is what happens next.
Transposer Builds the Connectome
The transposer reads your .sang file and builds a .vana connectome: neurons, connections between them, gates, and emit mappings. The connectome is a text file. VANA loads it and takes over from there.
Blood Bridge Loads It Onto the GPU
The "blood bridge" is an async CUDA kernel that uploads the connectome to your GPU and starts the organism running. It doesn't sync with the CPU every tick; it streams. The CPU does some very light work to feed data in, and then gets out of the way. The GPU is where the organism lives.
The Graph Settles
Every tick, the GPU settles the entire connectome in parallel. Neurons hold values. Connections carry signals. Gates fire when conditions are met. The graph relaxes toward a stable state, and that stable state IS your answer.
Output Comes Back
When the organism settles, motor neurons emit their values back through the blood bridge. Those values are your program's output: numbers, text, whatever your code produces. The CPU reads them and hands them to you. The round trip is done.
Beyond the GPUHardware Independence
A new CPU core asks how to make the same thing faster. VANA asks, what if the hardware did something different. Today the engine runs on CUDA cores. Tomorrow it can run on custom firmware on a dedicated board. Eventually it can run on a SoC where the language and the chip share the same structure.
The hardware may change underneath, but the language, and support for older hardware, does not. VANA is not tied to any preset configuration in silicon, firmware, or software. It is meant to serve as the bridge between Sanguis coding language and native hardware support for GPU-based compiled programs, until custom SoC allows full native integration.
Where It GoesConnectome Architectures
VANA isn't just a faster runtime. It's a different way for hardware to compute. The connectome maps to the same kind of parallel structure your brain uses, not a single line of instructions.
The connectome isn't a prototype. It's running today, settling real programs on real silicon that wasn't built for it.
The VANA connectome runtime executes Sanguis programs on consumer GPUs today. The transposer builds neural graphs from .sang source. The blood bridge runs the full organism on the GPU. Over 260 tests pass. Benchmark results on the Benchmarks page show real performance data from this hardware.