|
3 | 3 | The fastest and most memory efficient lattice Boltzmann CFD software, running on all GPUs via [OpenCL](https://github.com/ProjectPhysX/OpenCL-Wrapper "OpenCL-Wrapper"). Free for non-commercial use. |
4 | 4 |
|
5 | 5 | <a href="https://youtu.be/-MkRBeQkLk8"><img src="https://img.youtube.com/vi/o3TPN142HxM/maxresdefault.jpg" width="50%"></img></a><a href="https://youtu.be/oC6U1M0Fsug"><img src="https://img.youtube.com/vi/oC6U1M0Fsug/maxresdefault.jpg" width="50%"></img></a><br> |
6 | | -<a href="https://youtu.be/XOfXHgP4jnQ"><img src="https://img.youtube.com/vi/XOfXHgP4jnQ/maxresdefault.jpg" width="50%"></img></a><a href="https://youtu.be/BStzTRmLW7Q"><img src="https://img.youtube.com/vi/BStzTRmLW7Q/maxresdefault.jpg" width="50%"></img></a> |
| 6 | +<a href="https://youtu.be/XOfXHgP4jnQ"><img src="https://img.youtube.com/vi/XOfXHgP4jnQ/maxresdefault.jpg" width="50%"></img></a><a href="https://youtu.be/clAqgNtySow"><img src="https://img.youtube.com/vi/clAqgNtySow/maxresdefault.jpg" width="50%"></img></a> |
7 | 7 | (click on images to show videos on YouTube) |
8 | 8 |
|
9 | 9 | <details><summary>Update History</summary> |
@@ -220,7 +220,7 @@ $$f_j(i\\%2\\ ?\\ \vec{x}+\vec{e}_i\\ :\\ \vec{x},\\ t+\Delta t)=f_i^\textrm{tem |
220 | 220 | - FluidX3D (D3Q19) requires only 55 Bytes/cell with [Esoteric-Pull](https://doi.org/10.3390/computation10060092)+[FP16](https://www.researchgate.net/publication/362275548_Accuracy_and_performance_of_the_lattice_Boltzmann_method_with_64-bit_32-bit_and_customized_16-bit_number_formats)<br> |
221 | 221 | - 🟧🟧🟧🟧🟦🟦🟦🟦🟦🟦🟦🟦🟦🟦🟦🟦🟨🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩<br>(density 🟧, velocity 🟦, flags 🟨, DDFs 🟩; each square = 1 Byte) |
222 | 222 | - allows for 19 Million cells per 1 GB VRAM |
223 | | - - in-place streaming with [Esoteric-Pull](https://doi.org/10.3390/computation10060092): eliminates redundant copy `B` of density distribution functions (DDFs) in memory; almost cuts memory demand in half and slightly increases performance due to implicit bounce-back boundaries; offers optimal memory access patterns for single-cell in-place streaming |
| 223 | + - in-place streaming with [Esoteric-Pull](https://doi.org/10.3390/computation10060092): eliminates redundant copy of density distribution functions (DDFs) in memory; almost cuts memory demand in half and slightly increases performance due to implicit bounce-back boundaries; offers optimal memory access patterns for single-cell in-place streaming |
224 | 224 | - [decoupled arithmetic precision (FP32) and memory precision (FP32 or FP16S or FP16C)](https://www.researchgate.net/publication/362275548_Accuracy_and_performance_of_the_lattice_Boltzmann_method_with_64-bit_32-bit_and_customized_16-bit_number_formats): all arithmetic is done in FP32 for compatibility on all hardware, but DDFs in memory can be compressed to FP16S or FP16C: almost cuts memory demand in half again and almost doubles performance, without impacting overall accuracy for most setups |
225 | 225 | - <details><summary>only 8 flag bits per lattice point (can be used independently / at the same time)</summary> |
226 | 226 |
|
@@ -460,6 +460,7 @@ Colors: 🔴 AMD, 🔵 Intel, 🟢 Nvidia, ⚪ Apple, 🟡 ARM, 🟤 Glenfly |
460 | 460 | | 🔴 Radeon RX 5700 XT | 9.75 | 8 | 448 | 1368 (47%) | 3253 (56%) | 3049 (52%) | |
461 | 461 | | 🔴 Radeon RX 5600 XT | 6.73 | 6 | 288 | 1136 (60%) | 2214 (59%) | 2148 (57%) | |
462 | 462 | | 🔴 Radeon RX Vega 64 | 13.35 | 8 | 484 | 1875 (59%) | 2878 (46%) | 3227 (51%) | |
| 463 | +| 🔴 Radeon RX 590 | 5.53 | 8 | 256 | 1257 (75%) | 1573 (47%) | 1688 (51%) | |
463 | 464 | | 🔴 Radeon RX 580 4GB | 6.50 | 4 | 256 | 946 (57%) | 1848 (56%) | 1577 (47%) | |
464 | 465 | | 🔴 Radeon R9 390X | 5.91 | 8 | 384 | 1733 (69%) | 2217 (44%) | 1722 (35%) | |
465 | 466 | | 🔴 Radeon HD 7850 | 1.84 | 2 | 154 | 112 (11%) | 120 ( 6%) | 635 (32%) | |
|
0 commit comments