Next-gen CFD hardware accelerator

CFD needs to evolve, that’s why CFD SEMI gives every engineer a tool to replace RANS with LES technology…
in a way that hasn’t been seen before.

Learn more about algorithm


Based on the high-order Discontinuous Galerkin method on unstructured meshes wherein every step is optimized to efficiently perform on ASIC. Explicit approach ensures locality of the method which allows easy and efficient parallelisation of computations. Applied approximation using polynomials with P=2 (27 DOF in each element), corresponds to refinement mesh 3 times in each direction (3D) and provides a third order of accuracy.


Chip designed for one task – quick solving LES equations using Discontinuous Galerkin method, wherein every part is designed to get the maximum performance of the given algorithm. That Solver-On-Chip provides the best performance for one task – quick and efficient solving CFD problems. It will enable creating high parallel systems composed of hundreds or even thousands of circuits/devices with a linear speed-up factor.direction (3D) and provides a third order of accuracy.

Hyper SoC

Hyper Scalability enables to create systems with thousands of chips with a linear speed-up factor. This was achieved thanks to a special optimization of the algorithm itself (explicit), which ensures the possibility of combining any number of computational units without a decrease in performance. Also fast data exchange interfaces have been optimized for our algorithm.

For automotive

  • 30 days LES computations made in just 24 hours
  • 1 BLN DOF mesh with 40 000 000 elements
  • High-order tool to achieve best energy conversion efficiency

How our technology was build

Based on the equations describing the fluid dynamics, we created a numerical solver. Based on the solver, we created an ASIC. At each of these stages, we knew the target was a hardware solver built into the ASIC.
CFD SEMI believes that revolutionary ideas need a revolutionary approach. Therefore, we adapted the numerical algorithm for solving the Navier-Stokes equations based on the discontinuous Galerkin method in such a way, that it was performed as efficiently as possible in the ASIC system. On the other hand, our ASIC is designed to efficiently execute the algorithm and only this algorithm to achieve performance unattainable by the CPU.
In this way, we have created the world’s first Solver-on-Chip solving N-S equations using the discontinuous Galerkin method in which every smallest detail has been optimized to give every engineer the equipment thanks to which he can replace RANS methods with LES.

Our technology vs CPU

CFD SEMI and its story

Our story

CFD SEMI has been founded on more than two decades of professional experience in IT and CFD. These two pillars denote:
– DCD’s SEMI innovative approach in chip design like e.g. World’s Fastest 8051 CPU, fully scalable 32-bit CPU and 100% safe cryptographic system
– CFD scientists well known for customized CFD solutions like Applied Mathematical Modelling,, Chemical Engineering etc.

Our vision

In CFD SEMI we believe that the power of innovation opens a brave new world for everyone. That’s why we decided to design a fully scalable hardware accelerator (ASIC) with an algorithm based on the high-order Discontinuous Galerkin method. It may seem obvious, but it verges on a Copernican Revolution in comparison with what has become dominant for so many years. Like a PC is one intended for interactive individual use, as opposed to a mainframe computer, the same our solution gives a brave new world in your hands.


Our technology is a breakthrough available for every single engineer and company:
– LES available for every CFD engineer;
– Reducing computation time;
– The high order computation method;
– Money savings in the purchase of a new computational equipment;
– Less power consumption.