Yi Xu
Yi Xu
Quantized types are an experimental feature introduced in the [QuanTaichi paper](https://yuanming.taichi.graphics/publication/2021-quantaichi/quantaichi.pdf). With this useful feature, users can significantly save memory usage of their Taichi programs. The feature can also enable...
Issue: #6134
There's a mandatory `demote_operations` pass (https://github.com/taichi-dev/taichi/blob/master/taichi/transforms/demote_operations.cpp) which demotes high-level operations into low-level ones so that there's no need to handle these high-level ones in codegen of different backends. Those legacy...
In the release of v1.1.3, the workflow generates an incomplete changelog (https://github.com/taichi-dev/taichi/actions/runs/3087352313/jobs/4993540670): ``` Highlights: Full changelog: - [bug] [gui] Fix a bug of drawing mesh instacing that cpu/cuda objects have...
Sub-task of #5819. The following image shows the ideal state we would like to achieve: 
**Describe the bug** Running https://github.com/taichi-dev/taichi_elements/blob/2552c476838ee02071afd16ff2465613610c6f2f/demo/demo_quantized_simulation_letters.py with `use_bls=True` (L95) will get stuck. **To Reproduce** I managed to shorten the script from 1000+ lines to 100+ lines but I could not go...