jon-chuang

feat(storage): split duties of sstable_store `load_table` (load blocks) and `sstable` (meta only), disambiguate `TableHolder::{CachedMeta, Owned}`, do not store blocks in `meta_cache`

2

I hereby agree to the terms of the [Singularity Data, Inc. Contributor License Agreement](https://gist.github.com/skyzh/0663682a70b0edde7ae991492f2314cb#file-s9y_cla). ## What's changed and what's your intention? Remove abuse of sstable_store meta_cache to store data blocks....

opt(source): Source performance does not seem to scale linearly with cores

2

Source performance seems not to scale linearly with 6 threads v.s. 12 threads with materialized view turned off. It could be due to SMT as the bench was run on...

feat(executor): concurrent fetch of `JoinStateEntry`s in streaming hash join

5

## What's changed and what's your intention? Fetch all `JoinStateEntry`s async concurrently, grouping updates in stream chunk by join key. TODO: - [ ] Add more comments - [ ]...

There are essentially 2 interrelated parts to this issue. (1) Prefetching and asynchronous reads/writes is essentially the same thing. We simply want to use more threads to reduce load on...

Unable to reset environments asynchronously

4

When I run environments in parallel, only one of them gets reset while the rest wait for the first to be reset again. This can be recreated with the following...

Reason for abandonment

2

Did this project lose steam due to the rapid iteration of zk-SNARK schemes that convinced the development team that it is much wiser to iterate on CPU-based prototypes rather than...

**Description** Add `MetaDrive` simulator. Runs at 30 FPS on 1928 X 1208 inputs on an iGPU (Ryzen 6850H - Laptop). - Faster (30 FPS), just the openpilot screen: https://www.youtube.com/watch?v=paVD-laxbcY -...

`supercombo.onnx` causes segmentation fault in onnxruntime==1.14.0

2

### Describe the bug ``` python3 onnx_runner.py "../models/supercombo.onnx" ['onnx_runner.py', '../models/supercombo.onnx'] Onnx available providers: ['CPUExecutionProvider'] Onnx selected provider: ['CPUExecutionProvider'] Segmentation fault (core dumped) ``` navmodel.onnx runs fine ``` python3 onnx_runner.py "../models/navmodel.onnx"...

feat: extract LoRA for arbitrary models

2

It's straightforward, using [low-rank approximation](https://en.wikipedia.org/wiki/Low-rank_approximation), [low-rank matrix factorization](https://www.sciencedirect.com/science/article/abs/pii/S0925231217315710#:~:text=The%20purpose%20of%20low%2Drank,the%20interactions%20between%20two%20entries.). Given a model and its fine-tuning, and a target rank $k$, extract the "best" low-rank approximation to each difference in the model...

perf: AVX2/AVX routines for tall and skinny matmul - up to 15X speedup

13

Benchmark results: ```c++ sizey=sizez=N,sizex=K,n_threads=8 K=8,N=8192,AVX2,FLOPS/us=27148.97 K=8,N=8192,AVX,FLOPS/us=15193.96 K=8,N=8192,default,FLOPS/us=1781.05 K=16,N=8192,AVX2,FLOPS/us=20128.26 K=16,N=8192,AVX,FLOPS/us=8224.13 K=16,N=8192,default,FLOPS/us=3540.52 K=32,N=8192,AVX2,FLOPS/us=13127.55 K=32,N=8192,AVX,FLOPS/us=9397.48 K=32,N=8192,default,FLOPS/us=6386.55 K=48,N=8192,AVX2,FLOPS/us=13206.16 K=48,N=8192,AVX,FLOPS/us=5801.21 K=48,N=8192,default,FLOPS/us=8199.44 K=64,N=8192,AVX2,FLOPS/us=10505.51 K=64,N=8192,AVX,FLOPS/us=6353.32 K=64,N=8192,default,FLOPS/us=13024.33 ``` We choose the K cutoff point to be 32 for...