FluidX3D v3.5 (multi-GPU particles)

Thank you for using FluidX3D! Update v3.5 finally brings multi-GPU support for the PARTICLES extension. Now all extensions are supported with multi-GPU. Thanks to @jasonxauat for helping with the implementation approach!

Improvements

• PARTICLES extension now also works with multi-GPU
• faster force spreading if volume force is axis-aligned
• added more documentation for boundary conditions
• updated FAQs
• improved "hydraulic jump" sample setup
• updated GPU driver install instructions

Bug fixes

• disabled zero-copy on ARM iGPUs because CL_MEM_USE_HOST_PTR is broken there

Have fun with the software!
-- Moritz

FluidX3D v3.4 (bug fixes)

Thank you for using FluidX3D! Finally I also have an AMD GPU in my posession, so I can test FluidX3D locally on AMD/Intel/Nvidia GPUs within the same PC, to guarantee full compatibility. This allowed me to identify and fix 2 critical bugs that were coding mistakes on my side yet somehow only exposed with AMD's driver.

Improvements

• updated OpenCL driver install versions
• minor refactoring in stream_collide()

Bug fixes

• fixed bug in insertion-sort in voxelize_mesh() kernel causing crash on AMD GPUs
• fixed bug in voxelize_mesh_on_device() host code causing initialization corruption on AMD GPUs
• fixed dual CU and IPC reporting on AMD RDNA 1-4 GPUs

Have fun with the software!
-- Moritz

PS: Here's a little demo of "SLI"-ing AMD+Intel+Nvidia GPUs with FluidX3D:

FluidX3D v3.3 (faster .vtk export)

Thank you for using FluidX3D! Update v3.3 brings improvements to .vtk export and bug fixes:

Improvements

• .vtk export now converts and writes data in chunks, to reduce memory footprint and time for large memory allocation
• .vtk files now contain original file name as metadata in title
• INTERACTIVE_GRAPHICS_ASCII now renders in 2x vertical resolution but less colors
• updated OpenCL-Wrapper: more robust dp4a detection, fixed core count reporting for RDNA4 GPUs

Bug fixes

• fixed update_moving_boundaries() kernel not being called with flags other than TYPE_S
• fixed corrupted first frame until resizing with INTERACTIVE_GRAPHICS_ASCII
• fixed resolution() function for D2Q9
• fixed missing <chrono> header on some compilers
• fixed bug in split_regex()
• fixed compiler warning with min_int

Have fun with the software!
-- Moritz

FluidX3D v3.2 (fast force/torque summation)

Thank you for using FluidX3D! Update v3.2 brings the much requested GPU-accelerated force/torque summation:

Improvements

• implemented GPU-accelerated force/torque summation (~20x faster than CPU-multithreaded implementation before)
• simplified calculating object force/torque in setups; before:

  lbm.voxelize_mesh_on_device(mesh, TYPE_S|TYPE_X);
  const float3 lbm_com = lbm.calculate_object_center_of_mass(TYPE_S|TYPE_X);
  // ...
  lbm.calculate_force_on_boundaries();
  lbm.F.read_from_device(); // having to copy entire lbm.F from GPU VRAM to CPU RAM was slow!! 
  const float3 lbm_force = lbm.calculate_force_on_object(TYPE_S|TYPE_X); // slow CPU-multithreaded summation
  const float3 lbm_torque = lbm.calculate_torque_on_object(lbm_com, TYPE_S|TYPE_X); // slow CPU-multithreaded summation

  now:

  lbm.voxelize_mesh_on_device(mesh, TYPE_S|TYPE_X);
  const float3 lbm_com = lbm.object_center_of_mass(TYPE_S|TYPE_X);
  // ...
  const float3 lbm_force = lbm.object_force(TYPE_S|TYPE_X); // fast GPU-accelerated summation, copy only result to CPU
  const float3 lbm_torque = lbm.object_torque(lbm_com, TYPE_S|TYPE_X); // fast GPU-accelerated summation, copy only result to CPU

• improved coloring in VIS_FIELD/ray_grid_traverse_sum()
• updated OpenCL-Wrapper now compiles OpenCL C code with -cl-std=CL3.0 if available

Bug fixes

• fixed compiling on macOS with new OpenCL headers

Have fun with the software!
-- Moritz

Here a showcase of the improved coloring in VIS_FIELD/ray_grid_traverse_sum():

FluidX3D v3.1 (more bug fixes)

Thank you for using FluidX3D! Update v3.1 brings two critical bug fixes/workarounds and various small improvements under the hood:

Improvements

• faster enqueueReadBuffer() on modern CPUs with 64-Byte-aligned host_buffer
• hardened ray intersection functions against planar ray edge case
• updated OpenCL headers
• better OpenCL device specs detection using vendor ID and Nvidia compute capability
• better VRAM capacity reporting correction for Intel dGPUs
• improved styling of performance mermaid gantt chart in Readme
• added multi-GPU performance mermaid gantt chart in Readme
• updated driver install guides

Bug fixes

• fixed voxelization being broken on some GPUs
• added workaround for compiler bug in Intel CPU Runtime for OpenCL that causes Q-criterion isosurface rendering corruption
• fixed TFlops estimate for Intel Battlemage GPUs
• fixed wrong device name reporting for AMD GPUs (unlike every sane GPU vendor they don't report device name as CL_DEVICE_NAME but need CL_DEVICE_BOARD_NAME_AMD extension instead)

Have fun with the software!
-- Moritz

FluidX3D v3.0 (larger CPU/iGPU simulations)

A little gift to you all: FluidX3D v3.0 enables 31% larger grid resolution when running on CPUs or iGPUs!

Improvements

• reduced memory footprint on CPUs and iGPU from 72 to 55 Bytes/cell (fused OpenCL host+device buffers for rho/u/flags), allowing 31% higher resolution in the same RAM capacity
• faster hardware-supported and faster fallback emulation atomic floating-point addition for PARTICLES extension
• hardened calculate_f_eq() against bad user input for D2Q9

Bug fixes

• fixed velocity voxelization for overlapping geometry with different velocity
• fixed Remaining Time printout during paused simulation
• fixed CPU/GPU memory printout for CPU/iGPU simulations
• fixed bug that default_filename() would fail if there was a . in the file path

Have fun with the software!
-- Moritz

PS: Here's a little demo of what FluidX3D v3.0 is capable of:

FluidX3D v2.19 (camera splines)

Thank you for using FluidX3D! Update v2.19 adds Catmull-Rom splines for smooth camera movement, and bug fixes:

Improvements

• the camera can now fly along a smooth path through a list of provided keyframe camera placements, using Catmull-Rom splines
• more accurate remaining runtime estimation that includes time spent on rendering
• enabled FP16S memory compression by default
• printed camera placement using key G is now formatted for easier copy/paste
• added benchmark chart in Readme using mermaid gantt chart
• placed memory allocation info during simulation startup at better location

Bug fixes

• fixed threading conflict between INTERACTIVE_GRAPHICS and lbm.graphics.write_frame();
• fixed maximum buffer allocation size limit for AMD GPUs and in Intel CPU Runtime for OpenCL
• fixed wrong Re<Re_max info printout for 2D simulations
• minor fix in bandwidth_bytes_per_cell_device()

Have fun with the software!
-- Moritz

FluidX3D v2.18 (more bug fixes)

Thank you for using FluidX3D! Update v2.18 brings support for high refresh rate monitors on Linux and bug fixes:

Improvements

• added support for high refresh rate monitors on Linux
• more compact OpenCL Runtime installation scripts in Documentation
• driver/runtime installation instructions will now be printed to console if no OpenCL devices are available
• added domain information to LBM::write_status()
• added LBM::index function for uint3 input parameter

Bug fixes

• fixed that very large simulations sometimes wouldn't render properly by increasing maximum render distance from 10k to 2.1M
• fixed mouse input stuttering at high screen refresh rate on Linux
• fixed graphical artifacts in free surface raytracing on Intel CPU Runtime for OpenCL
• fixed runtime estimation printed in console for setups with multiple lbm.run(...) calls
• fixed density oscillations in sample setups (too large lbm_u)
• fixed minor graphical artifacts in raytrace_phi()
• fixed minor graphical artifacts in ray_grid_traverse_sum()
• fixed wrong printed time step count on raindrop sample setup

Have fun with the software!
-- Moritz

FluidX3D v2.17 (unlimited domain resolution)

Thank you for using FluidX3D! Update v2.17 removes the limit on 2³² cells per domain and adds new field visualization:

Improvements

• for GPUs/CPUs with >225 GB memory: domains are no longer limited to 4.29 billion (2³², 1624³) grid cells; if more are used, the OpenCL code will automatically compile with 64-bit indexing
• new, faster raytracing-based field visualization for single-GPU simulations (thanks @Snektron for the idea!)
• added GPU Driver and OpenCL Runtime installation instructions to documentation
• refactored INTERACTIVE_GRAPHICS_ASCII

Bug fixes

• fixed memory leak in destructors of floatN, floatNxN, doubleN, doubleNxN (all unused)
• made camera movement/rotation/zoom behavior independent of framerate
• fixed that smart_device_selection() would print a wrong warning if device reports 0 MHz clock speed

Have fun with the software!
-- Moritz

A glimpse of the new raytracing-based field visualization:

FluidX3D v2.16 (bug fixes)

I'm doing my part! With the v2.16 update I've put down all remaining known bugs for good. 🖖😎❌🪳❌
WOULD YOU LIKE TO KNOW MORE?

Bug fixes in this release:

• fixed that voxelization failed in Intel OpenCL CPU Runtime due to array out-of-bounds access
• fixed that voxelization did not always produce binary identical results in multi-GPU compared to single-GPU
• fixed that velocity voxelization failed for free surface simulations
• fixed terrible performance on ARM GPUs by macro-replacing fused-multiply-add (fma) with a*b+c
• fixed that Y/Z keys were incorrect for QWERTY keyboard layout in Linux
• fixed that free camera movement speed in help overlay was not updated in stationary image when scrolling
• fixed that cursor would sometimes flicker when scrolling on trackpads with Linux-X11 interactive graphics
• fixed flickering of interactive rendering with multi-GPU when camera is not moved
• fixed missing XInitThreads() call that could crash Linux interactive graphics on some systems
• fixed z-fighting between graphics_rasterize_phi() and graphics_flags_mc() kernels

Other improvements:

• simplified 10% faster marching-cubes implementation with 1D interpolation on edges instead of 3D interpolation, allowing to get rid of edge table
• added faster, simplified marching-cubes variant for solid surface rendering where edges are always halfway between grid cells
• refactoring in OpenCL rendering kernels

With GitHub I can track every bug from day it was discovered/fixed back to the day it was first introduced. This allows me to graph the number of open bugs over time, along with a curve weighted by their individual severity (minor 0.25, low 0.5, medium 1.0, high 2.0, showstopper 4.0):

Here is the distribution of days open, days till discovery and days till fix. I fixed 56% of bugs on the day of discovery. Notice the bimodal distribution of days open - a clear separation into "easy" and "nasty" bugs.

Lessons learned:

• Since release there was 63 bugs in FluidX3D in total, with at max 41 open bugs at one time. 🖖😱 Now there is 0, at least until I find more. 🖖😎 For reference: FluidX3D is 12.1k lines of code.
• Most bugs were a byproduct of big feature updates, like v2.0 (multi-GPU) and v2.1/v2.2 (voxelization). Of course at the time of introduction I didn't know that bugs slipped through, and I (or users) only discovered them later.
• Only 17% of bugs were found by users, all the others I found myself with rigorous testing. It takes continuous poking around in the code to find these often super rare bugs.
• 30% of bugs were actually bugs in the compiler, driver or operating system that needed a workaround on application side.
• The latest v2.16 release is the best FluidX3D has ever been. The worst, most bugged version by this metric is v2.2. 🖖🤠

Have fun with the software!
-- Moritz

PS: Here's an amusing FluidX3D video from @SLGY, he's doing his part too!