wonnx
cutlass
Our great sponsors
wonnx | cutlass | |
---|---|---|
18 | 16 | |
1,478 | 4,496 | |
6.2% | 5.5% | |
6.5 | 8.8 | |
20 days ago | 4 days ago | |
Rust | C++ | |
GNU General Public License v3.0 or later | GNU General Public License v3.0 or later |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
wonnx
-
Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'
The two I know of are IREE and Kompute[1]. I'm not sure how much momentum the latter has, I don't see it referenced much. There's also a growing body of work that uses Vulkan indirectly through WebGPU. This is currently lagging in performance due to lack of subgroups and cooperative matrix mult, but I see that gap closing. There I think wonnx[2] has the most momentum, but I am aware of other efforts.
[1]: https://kompute.cc/
-
VkFFT: Vulkan/CUDA/Hip/OpenCL/Level Zero/Metal Fast Fourier Transform Library
To a first approximation, Kompute[1] is that. It doesn't seem to be catching on, I'm seeing more buzz around WebGPU solutions, including wonnx[2] and more hand-rolled approaches, and IREE[3], the latter of which has a Vulkan back-end.
[1]: https://kompute.cc/
-
Onnx Runtime: “Cross-Platform Accelerated Machine Learning”
There's also a third-party WebGPU implementation: https://github.com/webonnx/wonnx
-
Are there any ML crates that would compile to WASM?
By experimental I meant e.g. using WGPU to run compute shaders like wonnx, which is working fine but only on a very restricted set of devices and browsers.
- WebGPU ONNX inference runtime written in Rust
-
PyTorch Primitives in WebGPU for the Browser
https://news.ycombinator.com/item?id=35696031 ... TIL about wonnx: https://github.com/webonnx/wonnx#in-the-browser-using-webgpu...
microsoft/onnxruntime: https://github.com/microsoft/onnxruntime
Apache/arrow has language-portable Tensors for cpp: https://arrow.apache.org/docs/cpp/api/tensor.html and rust: https://docs.rs/arrow/latest/arrow/tensor/struct.Tensor.html and Python: https://arrow.apache.org/docs/python/api/tables.html#tensors https://arrow.apache.org/docs/python/generated/pyarrow.Tenso...
Fwiw it looks like the llama.cpp Tensor is from ggml, for which there are CUDA and OpenCL implementations (but not yet ROCm, or a WebGPU shim for use with emscripten transpilation to WASM): https://github.com/ggerganov/llama.cpp/blob/master/ggml.h
Are the recommendable ways to cast e.g. arrow Tensors to pytorch/tensorflow?
FWIU, Rust has a better compilation to WASM; and that's probably faster than already-compiled-to-JS/ES TensorFlow + WebGPU.
What's a fair benchmark?
-
rustformers/llm: Run inference for Large Language Models on CPU, with Rust 🦀🚀🦙
wonnx has done some fantastic work in this regard, so that's where we plan to start once we get there. In terms of general discussion of alternate backends, see this issue.
-
I want to talk about WebGPU
> GPU in other ways, such as training ML models and then using them via an inference engine all powered by your local GPU?
Have a look at wonnix https://github.com/webonnx/wonnx
A WebGPU-accelerated ONNX inference run-time written 100% in Rust, ready for native and the web
-
Chrome Ships WebGPU
Looking forward to your WebGPU ML runtime! Also, why not contribute back to WONNX? (https://github.com/webonnx/wonnx)
-
OpenXLA Is Available Now
You can indeed perform inference using WebGPU (see e.g. [1] for GPU-accelerated inference of ONNX models on WebGPU; I am one of the authors).
The point made above is that WebGPU can only be used for GPU's and not really for other types of 'neural accelerators' (like e.g. the ANE on Apple devices).
cutlass
-
Optimization Techniques for GPU Programming [pdf]
I would recommend the course from Oxford (https://people.maths.ox.ac.uk/gilesm/cuda/). Also explore the tutorial section of cutlass (https://github.com/NVIDIA/cutlass/blob/main/media/docs/cute/...) if you want to learn more about high performance gemm.
-
Want to understand INT8 better
The latter (and I guess you were asking about this one) is designed to accelerate NN inference in reduced precision. It is possible to use Tensor Cores for you own purposes, mainly through CUTLASS. But because Tensor Cores are designed to execute matrix multiplications, it can be hard to adapt your problem to them. The performance with them is insane (IIRC 32x the performance of the INT32 pipeline), but only for matrix multiplication…
-
How do I deal with tensor core and cuda core with different precision?
If you want to learn about controlling Tensor Cores, the main way is through the CUTLASS library, that wraps the complexity of Tensor Cores into higher level abstractions. You can also look for mma/wmma instructions in the PTX specification, or for the WMMA API in CUDA.
-
AI’s compute fragmentation: what matrix multiplication teaches us
> we used tensor cores and managed to get back fp32 accuracy with 3 rounds of the things
Hey are you referring to 3xTF32 (https://github.com/NVIDIA/cutlass/tree/master/examples/28_am...)? IMO this is a perfect example where proper abstraction could save engineers non-trivial amount of time - imagine a compiler stack which allows 3xTF32 as a normal dtype and subsequent analysis compatible with this special dtype :-)
- With LLVM and MLIR, is manual cuda optimizing still important?
- CUTLASS 3.0 is now available
-
How to Optimize a CUDA Matmul Kernel for CuBLAS-Like Performance: A Worklog
This is a great post for people who are new to optimizing GPU code.
It is interesting to see that the author got this far without interchanging the innermost loop over k to the outermost loop, as is done in CUTLASS (https://github.com/NVIDIA/cutlass).
As you can see in this blog post the code ends up with a lot of compile-time constants (e.g. BLOCKSIZE, BM, BN, BK, TM, TN) one way to optimize this code further is to use an auto-tuner to find the optimal value for all of these parameters for your GPU and problem size, for example Kernel Tuner (https://github.com/KernelTuner/kernel_tuner)
-
pytorch example to actually see anything near 83 TFLOP/s on a RTX 4090?
Some examples here have a benchmark: https://github.com/NVIDIA/cutlass/blob/master/examples/24_gemm_grouped/gemm_grouped.cu
-
Create a bare CMake for Nvidia CUTLASS
I would like to make a minimum CMakeLists to use the CUDA CUTLASS library in another project. The build system is CMake, however I have little experience with CMake.
-
[D] What are some good resources to learn CUDA programming?
If you already know some C++, the Nvidia devblog is a great resource. Going further, Cub and Cutlass provide examples of efficient implementations for key operations at all hardware levels. Finally, this is more anecdotal but I always start my lectures on Cuda programming with the pictures in this doc page, to provide some intuition on the different memory layers that you can leverage to speed up a program. In any case, good luck :-)
What are some alternatives?
stablehlo - Backward compatible ML compute opset inspired by HLO/MHLO
TensorRT - PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT
onnx - Open standard for machine learning interoperability
iree - A retargetable MLIR-based machine learning compiler and runtime toolkit.
tract - Tiny, no-nonsense, self-contained, Tensorflow and ONNX inference
GPU-Puzzles - Solve puzzles. Learn CUDA.
triton - Development repository for the Triton language and compiler
burn - Burn is a new comprehensive dynamic Deep Learning Framework built using Rust with extreme flexibility, compute efficiency and portability as its primary goals.
Chess_BinaryNeuralNetwork - Training and Code Emitting Library for Binary Neural Networks
blaze - A Rustified OpenCL Experience
shark-samples