Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

High-Velocity GPU Kernel Authoring with CUTLASS Python Join Guray Ozen from NVIDIA at PyTorch Conference North America 2026, where he will present recent improvements to CUTLASS for Python. CUTLASS for Python is the technology stack used to write speed-of-light kernels for the latest GPU generations, powering today's most critical... show more

PyTorch

516,698 subscribers

11,242 views • 5 days ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

3 Comments

NVIDIA AI5 days ago

@nvidia Looking forward to it!

Eli5 days ago

@nvidia Bridging Python ergonomic wrappers directly down to CUTLASS warp-specialized GEMM templates without leaving JIT compile overhead is the holy grail.

Wei佳5 days ago

@nvidia I still don't write custom kernels because the debug loop scares me off. Static checks before launch is the feature I'd actually show up for.

Related Videos

ERRC: Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication PyTorch Foundation Ambassador Abdulsalam Bande will present a poster on ERRC (Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication) at PyTorch Conference North America 2026. ERRC focuses on optimizing GPU data exchange during LLM inference. By compressing inter-GPU communication and using the saved bandwidth to mitigate quantization errors, ERRC makes multi-GPU inference significantly more efficient. Register and join us to San Jose, October 20-21: #PyTorchCon Abdulsalam Bande

ERRC: Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication PyTorch Foundation Ambassador Abdulsalam Bande will present a poster on ERRC (Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication) at PyTorch Conference North America 2026. ERRC focuses on optimizing GPU data exchange during LLM inference. By compressing inter-GPU communication and using the saved bandwidth to mitigate quantization errors, ERRC makes multi-GPU inference significantly more efficient. Register and join us to San Jose, October 20-21: #PyTorchCon Abdulsalam Bande

PyTorch

16,886 views • 11 days ago

Involved in the AI ecosystem? There's something for everyone at PyTorch Conference North America. Steve Pousty of Red Hat explains why you should attend from the new intro track added this year to poster sessions, booths, and colocated events. Steve is giving a talk on the use of PyTorch and PyTorch-based libraries from the ecosystem to do remote sensing for satellite imagery. Join us in San Jose, October 20-21: #PyTorchCon

Involved in the AI ecosystem? There's something for everyone at PyTorch Conference North America. Steve Pousty of Red Hat explains why you should attend from the new intro track added this year to poster sessions, booths, and colocated events. Steve is giving a talk on the use of PyTorch and PyTorch-based libraries from the ecosystem to do remote sensing for satellite imagery. Join us in San Jose, October 20-21: #PyTorchCon

PyTorch

12,529 views • 7 days ago

SGLang-Diffusion is a high-performance serving framework for diffusion models, designed for both large-scale offline generation and latency-sensitive real-time inference. At PyTorch Conference North America 2026, Yihao Wang and Kevin Mi of RadixArk will present the design of SGLang-Diffusion and the PyTorch-based inference optimizations behind it, including efficient request scheduling, memory management, batching strategies, and execution-path optimizations for diffusion pipelines. Join us in San Jose, October 20-21: #PyTorchCon

SGLang-Diffusion is a high-performance serving framework for diffusion models, designed for both large-scale offline generation and latency-sensitive real-time inference. At PyTorch Conference North America 2026, Yihao Wang and Kevin Mi of RadixArk will present the design of SGLang-Diffusion and the PyTorch-based inference optimizations behind it, including efficient request scheduling, memory management, batching strategies, and execution-path optimizations for diffusion pipelines. Join us in San Jose, October 20-21: #PyTorchCon

PyTorch

17,035 views • 5 days ago

PyTorch Conference is where the open source Al community gathers to talk about what we are all building together. Priya Nagpurkar, VP of AI Native Systems at IBM Research will give a keynote at #PyTorchCon North America on IBM & Red Hat's contributions to the PyTorch ecosystem, including exciting developments on infrastructure for out-of-tree accelerators, enablement of IBM's Spyre Accelerator, and work with Ilm-d for distributed inference. Come to the talk, meet the community, and hear what's next in open source Al. Join us in San Jose, October 20-21:

PyTorch Conference is where the open source Al community gathers to talk about what we are all building together. Priya Nagpurkar, VP of AI Native Systems at IBM Research will give a keynote at #PyTorchCon North America on IBM & Red Hat's contributions to the PyTorch ecosystem, including exciting developments on infrastructure for out-of-tree accelerators, enablement of IBM's Spyre Accelerator, and work with Ilm-d for distributed inference. Come to the talk, meet the community, and hear what's next in open source Al. Join us in San Jose, October 20-21:

PyTorch

16,181 views • 9 days ago

Interested in optimizing Mixture of Experts LLM inference on ARM CPUs? In his talk at PyTorch Conference North America, Maajid Khan from Fujitsu Research India will explore efficient MoE LLM inference using vLLM and OpenVINO, sharing practical strategies for running these models effectively on ARM architectures. Register and join us in San Jose, October 20-21: #PyTorchCon

Interested in optimizing Mixture of Experts LLM inference on ARM CPUs? In his talk at PyTorch Conference North America, Maajid Khan from Fujitsu Research India will explore efficient MoE LLM inference using vLLM and OpenVINO, sharing practical strategies for running these models effectively on ARM architectures. Register and join us in San Jose, October 20-21: #PyTorchCon

PyTorch

16,448 views • 19 days ago

Learn how PyTorch can extend language modeling beyond classification to practical time-to-event prediction. Witold Czubala of UBS will share a PyTorch-based transformer model that predicts six-month attrition from longitudinal advisor–client email communications for wealth management during PyTorch Conference North America in San Jose, CA. Join us October 20-21: #PyTorchCon

Learn how PyTorch can extend language modeling beyond classification to practical time-to-event prediction. Witold Czubala of UBS will share a PyTorch-based transformer model that predicts six-month attrition from longitudinal advisor–client email communications for wealth management during PyTorch Conference North America in San Jose, CA. Join us October 20-21: #PyTorchCon

PyTorch

17,124 views • 1 month ago

At #PyTorchCon North America 2026, Liz Li and Shekhar of AMD will discuss their work enabling scalable MXFP8 pretraining on MI355X GPUs. Liz and Shekhar will cover optimization of key MXFP8 operators in TorchAO and how those kernels were integrated with TorchTitan to build an end-to-end upstream PyTorch training stack for large language models. Their session, “Scaling MXFP8 Pretraining on 1K+ AMD Instinct MI355X: TorchAO Kernels and TorchTitan Training,” will also cover Triton and FlyDSL implementations and lessons from numerical validation, convergence, and performance tuning. Join us in San Jose on October 20-21. Register today:

At #PyTorchCon North America 2026, Liz Li and Shekhar of AMD will discuss their work enabling scalable MXFP8 pretraining on MI355X GPUs. Liz and Shekhar will cover optimization of key MXFP8 operators in TorchAO and how those kernels were integrated with TorchTitan to build an end-to-end upstream PyTorch training stack for large language models. Their session, “Scaling MXFP8 Pretraining on 1K+ AMD Instinct MI355X: TorchAO Kernels and TorchTitan Training,” will also cover Triton and FlyDSL implementations and lessons from numerical validation, convergence, and performance tuning. Join us in San Jose on October 20-21. Register today:

PyTorch

16,137 views • 20 days ago

The open source AI community thrives on collaboration, choice, and open contribution. As Joseph Groenenboom from Red Hat highlights, empowering developers to freely contribute and build options is essential for growing the entire AI ecosystem. PyTorch Conference North America 2026 is where that community comes together in person. Join us to collaborate with open source pioneers, expand your deployment capabilities, and directly contribute to shaping the future of machine learning. Register for PyTorch Conference North America today: #PyTorchCon

The open source AI community thrives on collaboration, choice, and open contribution. As Joseph Groenenboom from Red Hat highlights, empowering developers to freely contribute and build options is essential for growing the entire AI ecosystem. PyTorch Conference North America 2026 is where that community comes together in person. Join us to collaborate with open source pioneers, expand your deployment capabilities, and directly contribute to shaping the future of machine learning. Register for PyTorch Conference North America today: #PyTorchCon

PyTorch

17,740 views • 1 month ago

From torch.profiler to Hardware Cycles: A Practical Profiling Playbook AWS Trainium: designed from the ground up for next generation AI training and inference is now a standard PyTorch device. When you need to optimize, the tools are built in. Esha Lakhotia and Pinak Panigrahi of Amazon Web Services will walk you from torch.profiler capture to cycle-level hardware debugging, with AI-assisted analysis that suggests fixes at PyTorch Conference North America. 📅 Oct 21 | 12:20 PM | LL21ABC Join us in San Jose, October 20-21: #PyTorchCon

From torch.profiler to Hardware Cycles: A Practical Profiling Playbook AWS Trainium: designed from the ground up for next generation AI training and inference is now a standard PyTorch device. When you need to optimize, the tools are built in. Esha Lakhotia and Pinak Panigrahi of Amazon Web Services will walk you from torch.profiler capture to cycle-level hardware debugging, with AI-assisted analysis that suggests fixes at PyTorch Conference North America. 📅 Oct 21 | 12:20 PM | LL21ABC Join us in San Jose, October 20-21: #PyTorchCon

PyTorch

14,523 views • 8 days ago

What You Cannot Profile, You Cannot Optimize: Learning to Read PyTorch Traces In his talk at PyTorch Conference North America 2026, Suvaditya Mukherjee from Hugging Face will share practical guidance on how to start using the PyTorch Profiler on your models to extract maximum performance efficiency. Whether you are aiming to identify performance bottlenecks or optimize model execution, this talk will help you get the most out of your hardware. Join the open source AI community in San Jose, October 20-21: #PyTorchCon Suvaditya Mukherjee

PyTorch

16,264 views • 14 days ago

"It's really great to feel the energy and see all the collaboration that's happening, lots of conversations. We have people here who are just starting to learn about PyTorch, and we have people who are maintainers. It's super nice to see that energy and see how together we're going to be able to solve some of these very difficult problems for humanity." Ankit Patel, NVIDIA, at PyTorch Conference Europe 2026 Register today to join the open source AI community at PyTorch Conference North America in San Jose, October 20–21: #PyTorchCon

"It's really great to feel the energy and see all the collaboration that's happening, lots of conversations. We have people here who are just starting to learn about PyTorch, and we have people who are maintainers. It's super nice to see that energy and see how together we're going to be able to solve some of these very difficult problems for humanity." Ankit Patel, NVIDIA, at PyTorch Conference Europe 2026 Register today to join the open source AI community at PyTorch Conference North America in San Jose, October 20–21: #PyTorchCon

PyTorch

17,000 views • 1 month ago

PyTorch Ambassador and #PyTorchCon North America 2026 program committee member Masahiro Hiramori (Masahiro Hiramori), 三菱電機｜Mitsubishi Electric, will be following work across on-device AI, efficient inference, voice agents, and machine learning compilers at this year’s conference. He's also looking forward to exchanging ideas with other developers and seeing what the broader PyTorch community is building. Register for PyTorch Conference North America 2026:

PyTorch Ambassador and #PyTorchCon North America 2026 program committee member Masahiro Hiramori (Masahiro Hiramori), 三菱電機｜Mitsubishi Electric, will be following work across on-device AI, efficient inference, voice agents, and machine learning compilers at this year’s conference. He's also looking forward to exchanging ideas with other developers and seeing what the broader PyTorch community is building. Register for PyTorch Conference North America 2026:

PyTorch

19,811 views • 3 days ago

At PyTorch Conference North America 2026, hear directly from and connect with the people working on PyTorch, vLLM, DeepSpeed, ray, Helion, and Safetensors, alongside experts working across the AI stack. At #PyTorchCon NA 2025, Richard Liaw (ray, Anyscale) spoke about the Ray team’s focus on building something “useful and critical to the AI infrastructure space” and about “how important Ray can be for the stack.” PyTorch conferences are the open source AI community’s town square, where what’s next gets decided. Register by September 4 (tomorrow) to save on your conference pass: #PyTorchCon

At PyTorch Conference North America 2026, hear directly from and connect with the people working on PyTorch, vLLM, DeepSpeed, ray, Helion, and Safetensors, alongside experts working across the AI stack. At #PyTorchCon NA 2025, Richard Liaw (ray, Anyscale) spoke about the Ray team’s focus on building something “useful and critical to the AI infrastructure space” and about “how important Ray can be for the stack.” PyTorch conferences are the open source AI community’s town square, where what’s next gets decided. Register by September 4 (tomorrow) to save on your conference pass: #PyTorchCon

PyTorch

16,342 views • 1 month ago

New chips are shipping faster than ever before, but the ecosystem is being held back by having to rewrite and then re-debug the same code over & over again. Chris Lattner CEO and Co-founder at Modular and EVP of Advanced AI Software & Platforms at Qualcomm, will deliver a keynote at PyTorch Conference North America about an open software platform for heterogeneous compute powered by Mojo and MAX. PyTorch has always been the place where the best models come together, and now there's a way to get those models onto all kinds of hardware. If you're interested in Al and compute, join us at the PyTorch Conference in San Jose, CA. Register now: #PyTorchCon

New chips are shipping faster than ever before, but the ecosystem is being held back by having to rewrite and then re-debug the same code over & over again. Chris Lattner CEO and Co-founder at Modular and EVP of Advanced AI Software & Platforms at Qualcomm, will deliver a keynote at PyTorch Conference North America about an open software platform for heterogeneous compute powered by Mojo and MAX. PyTorch has always been the place where the best models come together, and now there's a way to get those models onto all kinds of hardware. If you're interested in Al and compute, join us at the PyTorch Conference in San Jose, CA. Register now: #PyTorchCon

PyTorch

20,838 views • 29 days ago

Want to hear more about new vLLM and SGLang engines built on a new PyTorch-native TPU backend, TorchTPU? Find Qi Zhou of Google, Colin Taylor and Angela Yi of Meta at PyTorch Conference North America in San Jose where they will discuss how they focused on making the existing SGLang & vLLM infrastructure work natively on TPU, preserving the schedulers, batching systems, OpenAI-compatible APIs, and torch.compile workflows already familiar to GPU users. Because the serving engines remain upstream, new models and features can run on TPU without requiring separate TPU-specific implementations. Learn more about how it works across the stacks & what they learned running this in production, including compile time, device placement under tracing, KV cache layout, and cross-chip collectives. Register for #PyTorchCon NA now:

Want to hear more about new vLLM and SGLang engines built on a new PyTorch-native TPU backend, TorchTPU? Find Qi Zhou of Google, Colin Taylor and Angela Yi of Meta at PyTorch Conference North America in San Jose where they will discuss how they focused on making the existing SGLang & vLLM infrastructure work natively on TPU, preserving the schedulers, batching systems, OpenAI-compatible APIs, and torch.compile workflows already familiar to GPU users. Because the serving engines remain upstream, new models and features can run on TPU without requiring separate TPU-specific implementations. Learn more about how it works across the stacks & what they learned running this in production, including compile time, device placement under tracing, KV cache layout, and cross-chip collectives. Register for #PyTorchCon NA now:

PyTorch

16,806 views • 13 days ago

At PyTorch Conference North America 2026, hear directly from and connect with the people working on PyTorch, vLLM, DeepSpeed, ray, Helion, and Safetensors, alongside experts working across the AI stack. Keynote speaker Simon Mo (vLLM, Inferact) says vLLM’s goal is to become “the easiest to use and most efficient inference engine.” At #PyTorchCon NA 2025, he spoke about the difficulty and cost of using large language models and continuing to push inference efficiency forward. PyTorch conferences are the open source AI community’s town square, where what’s next gets decided. Register by September 4 to save on your conference pass:

At PyTorch Conference North America 2026, hear directly from and connect with the people working on PyTorch, vLLM, DeepSpeed, ray, Helion, and Safetensors, alongside experts working across the AI stack. Keynote speaker Simon Mo (vLLM, Inferact) says vLLM’s goal is to become “the easiest to use and most efficient inference engine.” At #PyTorchCon NA 2025, he spoke about the difficulty and cost of using large language models and continuing to push inference efficiency forward. PyTorch conferences are the open source AI community’s town square, where what’s next gets decided. Register by September 4 to save on your conference pass:

PyTorch

22,334 views • 1 month ago

Think DeepSpeed is only good for ZeRO and its usability, not training throughput? At #PyTorchCon North America 2026, Masahiro Tanaka will cover tensor, sequence, and expert parallelism in DeepSpeed and walk through benchmarks. His session, “DeepSpeed Is Not Just ZeRO: Tensor, Sequence, and Expert Parallelism with ZeRO-Level Usability,” will cover AutoTP, AutoSP, and AutoEP, how these capabilities can be combined with ZeRO, and where each parallelism strategy fits. Register for PyTorch Conference North America 2026:

Think DeepSpeed is only good for ZeRO and its usability, not training throughput? At #PyTorchCon North America 2026, Masahiro Tanaka will cover tensor, sequence, and expert parallelism in DeepSpeed and walk through benchmarks. His session, “DeepSpeed Is Not Just ZeRO: Tensor, Sequence, and Expert Parallelism with ZeRO-Level Usability,” will cover AutoTP, AutoSP, and AutoEP, how these capabilities can be combined with ZeRO, and where each parallelism strategy fits. Register for PyTorch Conference North America 2026:

PyTorch

15,911 views • 8 days ago

Get more out of your AI infrastructure. Carlos Costa, Distinguished Engineer at IBM Research and co-founder of the llm-d project will share some exciting news about llm-d: how the project changed the way the open source AI community scales inference at PyTorch Conference North America. A little over a year after the project was founded, llm-d is reaching a new level of maturity, delivering state-of-the-art performance for generative serving, while also enabling the capabilities needed for day two operations, including multi-tenant quality of service, zero downtime operations, and all of that without sacrificing performance in production. Join us in San Jose, October 20-21:

Get more out of your AI infrastructure. Carlos Costa, Distinguished Engineer at IBM Research and co-founder of the llm-d project will share some exciting news about llm-d: how the project changed the way the open source AI community scales inference at PyTorch Conference North America. A little over a year after the project was founded, llm-d is reaching a new level of maturity, delivering state-of-the-art performance for generative serving, while also enabling the capabilities needed for day two operations, including multi-tenant quality of service, zero downtime operations, and all of that without sacrificing performance in production. Join us in San Jose, October 20-21:

PyTorch

11,305 views • 2 days ago

PyTorch 2.13 brings FlexAttention to Apple Silicon, cuts peak memory by up to 4× for large-vocabulary models with nn.LinearCrossEntropyLoss, and updates distributed training, compilation, profiling, and on-device inference. Our live Q&A examined CUDA version support, CuTeDSL in TorchInductor, Python 3.15, ExecuTorch, torch.compile, and plans for PyTorch 2.14. Andrey Talman (Meta), albanD (Meta), and Piotr Bialecki (NVIDIA) joined moderator Chris Gottbrath (Gottbrath Technologies). This clip explains how FlexAttention compiles kernels specialized for the requested masking pattern, allowing them to skip unnecessary computation. It also covers speedups of up to approximately 12× over SDPA in tested sparse configurations and a deterministic backward path on CUDA that removes one source of nondeterminism in gradient computation. 🔗 Watch the full PyTorch 2.13 Release Live Q&A:

PyTorch 2.13 brings FlexAttention to Apple Silicon, cuts peak memory by up to 4× for large-vocabulary models with nn.LinearCrossEntropyLoss, and updates distributed training, compilation, profiling, and on-device inference. Our live Q&A examined CUDA version support, CuTeDSL in TorchInductor, Python 3.15, ExecuTorch, torch.compile, and plans for PyTorch 2.14. Andrey Talman (Meta), albanD (Meta), and Piotr Bialecki (NVIDIA) joined moderator Chris Gottbrath (Gottbrath Technologies). This clip explains how FlexAttention compiles kernels specialized for the requested masking pattern, allowing them to skip unnecessary computation. It also covers speedups of up to approximately 12× over SDPA in tested sparse configurations and a deterministic backward path on CUDA that removes one source of nondeterminism in gradient computation. 🔗 Watch the full PyTorch 2.13 Release Live Q&A:

PyTorch

18,591 views • 2 months ago