Uploaded February 2022 | Updated November 2026, 5 hours ago
I like rapid HPC prototyping with interactive graphics. So much in fact that I even use them on the Leibnitz Rechenzentrum (LRZ) Compute Cloud - part of Germany's 2nd fastest supercomputer - over SSH in WSL, where FluidX3D runs on a Nvidia Tesla V100 GPU that does not even have a video output.
That's right, real time stereoscopic 3D raytracing in a remote Linux console.
This is a real time demo of a hydraulic jump simulation with interactive raytraced graphics: youtu.be/NQPgumd3Ei8
The console graphics are demonstrated here: youtu.be/tehGAbRB-sw
The simulation software used is FluidX3D, an OpenCL implementation of the lattice Boltzmann method. FluidX3D runs at peak hardware efficiency on the worlds fastest data-center GPUs (8799/5232 MLUPs/s on the Nvidia A100/V100 @ D3Q19 SRT FP32), on gaming GPUs and CPUs and it even runs my smartphone.
The fluid simulation is done with the lattice Boltzmann method (LBM) with the Volume-of-Fluid (VoF) extension and piecewise linear interface construction (PLIC) for surface tension.
Raindrop simulations: doi.org/10.1186/s43591-021-00018-8
Volume-of-Fluid: doi.org/10.3390/computation10020021
Esoteric-Pull: doi.org/10.3390/computation10060092
FP32/FP16 mixed precision: researchgate.net/publication/362275548_Accuracy_and_performance_of_the_lattice_Boltzmann_method_with_64-bit_32-bit_and_customized_16-bit_number_formats
FluidX3D: doi.org/10.15495/EPub_UBT_00005400
VoF-LBM provides a 3D grid with fluid fill levels, used as isovalues for marching-cubes, which generates a contiguous triangle mesh representing the water surface.
Marching-cubes: paulbourke.net/geometry/polygonise
With raytracing I shoot light rays from the camera through the 3D isogrid. For each grid cell that a ray traverses, I run marching-cubes to generate the triangles on-the-fly, then check for ray-triangle intersections with any of these triangles. If the ray intersects, I don't use the flat-face-surface nromal of the triangle, as this would lead to visual artifacts on the reflected/refracted rays. Instead, I interpolate the surface normal at the intersection point in order to have a smoothly appearing surface. How does surface normal interpolation work with marching-cubes? I get the normals directly at the lattice points with gradient (central derivative) on the isogrid, then do trilinear interpolation to the intersection point. This is only computed once, after the intersection point has been found.
The 3D lattice that the LBM provides already is one of the ideal acceleration structures for raytracing. Any ray only has to check for possible intersections with isosurface triangles contained within the traversed grid cells. This makes it - when the lattice is not too large - even run in real time on any OpenCL 1.2 capable GPU with enough memory. No RTX or "professional" GPU required.
Also see the most realistic raindrop simulation ever done: youtu.be/9Rfu9ZqW894
Timestamps:
0:00 intro
0:02 hardware showcase
0:17 starting FluidX3D
0:25 real time 3D graphics in the remote console
1:02 raytracing in console
2:09 outro
#FluidX3D #OpenCL #Raytracing #GPU #LBM
I like rapid HPC prototyping with interactive graphics. So much in fact that I even use them on the Leibnitz Rechenzentrum (LRZ) Compute Cloud - part of Germany's 2nd fastest supercomputer - over SSH in WSL, where FluidX3D runs on a Nvidia Tesla V100 GPU that does not even have a video output.
That's right, real time stereoscopic 3D raytracing in a remote Linux console.
This is a real time demo of a hydraulic jump simulation with interactive raytraced graphics: youtu.be/NQPgumd3Ei8
The console graphics are demonstrated here: youtu.be/tehGAbRB-sw
The simulation software used is FluidX3D, an OpenCL implementation of the lattice Boltzmann method. FluidX3D runs at peak hardware efficiency on the worlds fastest data-center GPUs (8799/5232 MLUPs/s on the Nvidia A100/V100 @ D3Q19 SRT FP32), on gaming GPUs and CPUs and it even runs my smartphone.
The fluid simulation is done with the lattice Boltzmann method (LBM) with the Volume-of-Fluid (VoF) extension and piecewise linear interface construction (PLIC) for surface tension.
Raindrop simulations: doi.org/10.1186/s43591-021-00018-8
Volume-of-Fluid: doi.org/10.3390/computation10020021
Esoteric-Pull: doi.org/10.3390/computation10060092
FP32/FP16 mixed precision: researchgate.net/publication/362275548_Accuracy_and_performance_of_the_lattice_Boltzmann_method_with_64-bit_32-bit_and_customized_16-bit_number_formats
FluidX3D: doi.org/10.15495/EPub_UBT_00005400
VoF-LBM provides a 3D grid with fluid fill levels, used as isovalues for marching-cubes, which generates a contiguous triangle mesh representing the water surface.
Marching-cubes: paulbourke.net/geometry/polygonise
With raytracing I shoot light rays from the camera through the 3D isogrid. For each grid cell that a ray traverses, I run marching-cubes to generate the triangles on-the-fly, then check for ray-triangle intersections with any of these triangles. If the ray intersects, I don't use the flat-face-surface nromal of the triangle, as this would lead to visual artifacts on the reflected/refracted rays. Instead, I interpolate the surface normal at the intersection point in order to have a smoothly appearing surface. How does surface normal interpolation work with marching-cubes? I get the normals directly at the lattice points with gradient (central derivative) on the isogrid, then do trilinear interpolation to the intersection point. This is only computed once, after the intersection point has been found.
The 3D lattice that the LBM provides already is one of the ideal acceleration structures for raytracing. Any ray only has to check for possible intersections with isosurface triangles contained within the traversed grid cells. This makes it - when the lattice is not too large - even run in real time on any OpenCL 1.2 capable GPU with enough memory. No RTX or "professional" GPU required.
Also see the most realistic raindrop simulation ever done: youtu.be/9Rfu9ZqW894
Timestamps:
0:00 intro
0:02 hardware showcase
0:17 starting FluidX3D
0:25 real time 3D graphics in the remote console
1:02 raytracing in console
2:09 outro
#FluidX3D #OpenCL #Raytracing #GPU #LBM



![The History of PhysX3D [GRAVITY SIMULATION]
Starting completely from scrap, I decided to write an experimental program in 2014, which has since then turned into the most efficient gravity simulation ever written in Java.
Im 19 years old now, being considered one of the best young programmers in Bavaria, Germany.
More information at: http://www.projectphysx.de
Music: Sex Whales & Phantom Sage - One Day [NCS Release] The History of PhysX3D [GRAVITY SIMULATION]](https://i.ytimg.com/vi/8_2Rbzszn4c/mqdefault.jpg)
![FluidX3D [FLUID SIMULATION] 3.5 Million Particles through porous Media in Real Time
FluidX3D is a GPU-accelerated real time 3D fluid simulation based on the lattice Boltzmann method.
Line3D, my own graphics engine, got some upgrades. I transferred my rendering kernel to OpenCL. The particles now are not anymore rendered by the CPU one by one, but by the GPU all at the same time, so all the particle positions stay in video memory. Only the completely rendered frame is sent over the PCI-e bus. This dramatically reduces memory transfer time.
Even though rendering slows down the LBM simulation because it occupies part of the GPU, 3.5 Million LBM points and particles still work in real time at around 32fps.
For more information about FluidX3D and my other projects, visit
http://www.projectphysx.de
Music: Anzo - Raindance [Tasty Release] FluidX3D [FLUID SIMULATION] 3.5 Million Particles through porous Media in Real Time](https://i.ytimg.com/vi/9HZkO0caFBk/mqdefault.jpg)




![DroneEngine [QUADCOPTER SIMULATION] August Update: sensor fusion demo
Ich kann in meinem Programm jetzt verschiedene Sensoren simulieren. Die Messungen der Sensoren sind fehlerbehaftet und richten sich nach Sensor-spezifischen Operationsfrequenzen. Die Messergebisse werden mittels selbst entwickelten sensor-fusion-Algorithmen kombiniert, um zusammen ein möglichst genaues Ergebnis, in dem Fall die drei Winkel und die Positionsdaten der Drohne, zu liefern, welches in die Steueralgorithmen eingespeist wird.
Song: Sebastian Forslund - Geronimo DroneEngine [QUADCOPTER SIMULATION] August Update: sensor fusion demo](https://i.ytimg.com/vi/BN34lgJXmcA/mqdefault.jpg)
![PhysX3D [GRAVITY SIMULATION] Early Universe Simulation
I have implemented GPU computation via aparapi in my PhysX3D project. At high body numbers the Nvidia GeForce GTX 960M (1.40Tflops, 640 shader units @ 1.097GHz) is 7x faster compared to the Intel Core i7-4720HQ (0.44Tflops, 8 threads @ 3.56GHz).
The simulation is being calculated in real time. At 10k bodies the GPU accomplishes 5 steps per second with Runge-Kutta 4th order.
The program currently is 3953 lines of code.
Special thanks to ThePhysics1234 for helping me out with aparapi.
https://www.youtube.com/user/ThePhysics1234
More information at: http://www.projectphysx.de
Music: Chris Zabriskie - Cylinder Three PhysX3D [GRAVITY SIMULATION] Early Universe Simulation](https://i.ytimg.com/vi/BOiFRbcMDNk/mqdefault.jpg)