SIML Compute
Fit more state into GPU memory and measure how quickly a specialized kernel advances it.
Cellular simulation and binary lattice sampling →Scientific computing · Deep SIML Labs
We develop GPU methods that fit larger biological simulations into a fixed memory budget. Our research explores how simulation and optimization can guide the next experiment in biology and materials.
Research publications →MCF-7 signaling model · One RTX 3060
Six seconds of simulated time
Demonstrated results
One study measures interacting cellular models. The other measures binary lattice updates. Each uses its own baseline and answers a different question.
A GPU can run out of memory before a model reaches the population a researcher wants to study. On one RTX 3060, SIML CUDA advanced 8,100,736 interacting MCF-7 signaling patches through six simulated seconds. Triton advanced 6,022,368 under the same 6 GB budget and accuracy requirements.
On one RTX 3060, SIML updated 256 small binary lattices 12.08× as fast as a separately tuned UInt8 CUDA control once the data were on the GPU. A separate capacity test kept one connected lattice of 84.26 billion binary variables updating continuously for ten minutes.
Our technology
Fit more state into GPU memory and measure how quickly a specialized kernel advances it.
Cellular simulation and binary lattice sampling →Model interacting systems and compare candidate experiments. We are investigating how to select the next simulation from its expected scientific value.
Computational experiments and predictive models →Study how ionic transport and polymer structure could inform sensing, compliant actuation, and future computing substrates.
Materials questions and planned physical tests →How we work
Define the constraint and the outcome that would matter.
Specify the system, its assumptions, and a reference.
Advance the model under a declared compute budget.
Match the workload and score both implementations.
Refine the representation or the execution method.
Check numerical agreement. Physical validation of proposed materials is planned.
Technical evaluations
These are potential joint evaluations. Bring a workload, its current implementation, and the outcome that would make a difference to your team.
Partnerships
We welcome research groups and scientific computing teams interested in comparing methods, extending a model, or designing a joint validation study.