@CuPy_Team

Drop-in replacement of NumPy and SciPy for GPU - Follow us for the latest updates! 🫶 Support us via GitHub Sponsors: https://xitter.cf/t.co/Xk2Syli21o

cupy.dev
Joined June 2015
  • Tweets483
  • Following19
  • Followers6.2K
18Media
CuPy retweeted
PyTorch
@PyTorchSep 8
Major ecosystem libraries like PyTorch and CuPy now build on the shared cuda.core foundation, reducing version conflicts and interop bugs for developers. With CUDA 13.3, @nvidia released CUDA Python 1.0, the libraries and tools that give you the full CUDA platform from Python. Using standard CUDA context, devices, streams, and memory are shared with the rest of the Python GPU ecosystem, so your own kernels can run against CuPy arrays or PyTorch tensors without copying data. Read the full blog post: bit.ly/4hcSu2C
2
17
3
144
17,735
CuPy retweeted
mohit@mohitwt_Mar 7

Flash Attention from Scratch in CUDA and CuPy

Article · @mohitwt_
8
106
1
969
112,420
📢 Just announced: nvmath-python brings NVIDIA accelerated math to #Python. ✅ Pythonic APIs for using CUDA math operations ✅ Works with your favorite Python packages ✅ #OpenSource Beta version is available now! #HPC Learn more now: nvda.ws/3VV2fGN
1
30
1
96
6,603
CuPy retweeted
AMD
@AMD18 Dec 2023
AMD ROCm 6.0 is out, release notes: rocm.docs.amd.com/en/latest/… 🧮 Improved perf for lower precision math and attention layers ✨ Sparsity and quantization with DeepSpeed, ONNX-RT, CuPy plus new hipSPARSELt lib 🔥 Frameworks and ML pipeline updates for PyTorch, TensorFlow, JAX
AMD
@AMD6 Dec 2023
Looking to unleash significantly greater performance and productivity for large language models? Say hello to ROCm 6 – our latest open software stack that accelerates AI performance and enables new capabilities for generative AI and LLMs.
5
53
1
215
51,705
I had the rare opportunity to present @CuPy_team in front of Guido van Rossum @gvanrossum. I've never thought I could have a chance to say "Thank you for making Python!" to him directly. 😊 Thanks @PyConJapan guys for hosting & inviting me to #GuidoMeetup!
8
35
3,536
Exciting news from the CuPy team! 🙌 Improved interpolation & signal processing capabilities are bridging the gap between @SciPy_team and @CuPy_Team. Performance gains are up 32x compared to SciPy's CPU implementations. CuPy v13.0 is approaching! 🚀 labs.quansight.org/blog/cupy…

Improving the interpolation and signal processing capabilities of CuPy | Quansight Labs

We are excited to spread the news about the improvements that have been taking place in CuPy, where 18 interpolation and more than 100 signal processing parallel GPU APIs are now available as part of...

labs.quansight.org
17
1
43
4,665
Are you familiar with @NumFOCUS? It's a non-profit that sponsors numpy, pandas, jupyter, Julia, scikit, matplotlib, and more. They've had decreased funding this year, let's help. Donate to @NumFOCUS today, and I will match all donations up to $10,000 within the next 48 hours!
13
113
8
390
119,769
🎉 @CuPy_Team has been approved for the Docker-Sponsored Open Source Program! A huge thank you to @Docker for providing a free Docker Team subscription, empowering our community to thrive! 🙌 hub.docker.com/r/cupy/cupy
1
5
19
8,355
The ex-Chainer team at @PreferredNet has contributed a pluggable memory allocator feature in @PyTorch required for our production use-case, and it comes with another bonus - making @RAPIDSai, @CuPy_Team, and many other CUDA-powered libraries interoperable with each other!
RAPIDS AI@RAPIDSai12 Jan 2023
We are kicking off an exciting year for RAPIDS with this great blog on using @PyTorch with RAPIDS Memory Manager! medium.com/rapids-ai/pytorch…
1
7
26
6,219
This is again a very impactful funding round. @cziscience has a major positive influence on open source scientific computing. From the EOSS grants, I'm particularly pleased to see @CuPy_Team and @Spyder_IDE getting their first ever grants, very much well-deserved! 🎉 1/3
biohub
@biohub10 Nov 2022
Replying to @biohub
We’re announcing 40 new grants to some of the most widely-used scientific #OpenSource software tools that are essential to science chanzuckerberg.com/eoss/prop…

CZI – Essential Open Source Software for Science

CZI’s Essential Open Source Software for Science program supports software maintenance, growth, development, and community engagement for open source tools critical to science.

chanzuckerberg.com
1
6
1
37
New blog post 🚨: Read from Zarr straight to GPU-backed xarray objects! Read all about it ⬇️
伟吉@weiji14_30 Aug 2022
💥 #GPU-native analytics with @xarray_dev + @RAPIDSai kvikIO! ⏩ Go from @Zarr_dev to @CuPy_Team arrays using @NVIDIA GPU Direct Storage without going through CPU RAM 🤯! Read @cherian_deepak's blog post xarray.dev/blog/xarray-kviki… and accelerate your @pangeo_data #ML research 🚀🚀🚀

Enabling GPU-native analytics with Xarray and kvikIO

An experiment with direct-to-GPU reads from a Zarr store using Xarray.

xarray.dev
1
10
3
35
Learn how an advanced feature of #CuPy can increase computational efficiency – and create visually interesting fractals too. accntu.re/3vbFkJU
1
7
Check out accelerated topic modeling using @RAPIDSai, @numba_jit, @CuPy_Team, @huggingface, and @PyTorch for text processing, #deeplearning-based embedding creation, dimensionality reduction, and clustering on @NVIDIAAI #GPUs. medium.com/rapids-ai/acceler…

Accelerating Topic modeling with RAPIDS and BERT models

Topic modeling for text processing, Deep Learning based embedding creation, reduction, and clustering on GPUs.

medium.com
2
11
2
40
Good news! #Julialang can recognize #cupy via PyCall.jl.
7
19
Best CPU vs Best GPU: 50x NumPy = CuPy! @unum_cloud @ashvardanian #cuda #python #numpy #gpu unum.cloud/post/2022-01-25-c…

Unum | The IO Layer For AI

Unum builds the Search, Storage, and Networking engines powering the modern data stack. Deployed on 100M+ devices, our technology serves DBMS unicorns and frontier AI labs.

unum.cloud
1
2
11
Hi @CuPy_Team and @napari_imaging fans, I just made another #GPUaccelerated #BioImageAnalysis plugin and would love to hear if the installation works on other's computers. Please try it out and let me know! 🔬🖥️🚀 github.com/haesleinhuepf/nap…
2
15
56
Interested in ML Frameworks Interoperability? @koalalesque and I have been working on a set of blogs covering this topic. The first part is already online. I hope you enjoy it! developer.nvidia.com/blog/ma… @nvidia @RAPIDSai @CuPy_Team @PyTorch

Machine Learning Frameworks Interoperability, Part 1: Memory Layouts and Memory Pools | NVIDIA...

Learn about the pros and cons of distinct memory layouts, as well as memory pools for asynchronous memory allocation to enable zero-copy functionality.

developer.nvidia.com
13
1
32
Today, we open up our GitHub repository for AlphaPept, our brand new framework for open, fast, #Python-based next generation of #massspec #proteomics analysis. Check it out here: github.com/MannLabs/alphapep…. @numba_jit, @nvidia, @streamlit, @CuPy_Team, @Omic_Era, @jeremyphoward

GitHub - MannLabs/alphapept: A modular, python-based framework for mass spectrometry. Powered by...

A modular, python-based framework for mass spectrometry. Powered by nbdev. - MannLabs/alphapept

github.com
1
100
4
321
"Python bottleneck optimization progression from lists to cupy arrays" by Yair Beer from Pycon Israel 2021 youtube.com/watch?v=Zb2PXGGv…
2
12
Daniel’s going to talk (@SciPyConf) about what he did to accelerate the code 20x porting it to @nvidia GPUs using CuPy and Numba, and keeping the code Python just like DESI likes. @CuPy_Team @numba_jit
2
13
Load more