NVIDIA Developer | Scale AI Applications to the Data Center and Cloud with NVIDIA Nsight Systems @NVIDIADeveloper | Uploaded March 2024 | Updated October 2024, 4 days ago.
Applications are harnessing generative AI, and scaling out to the GPU-accelerated data center and cloud. To help you build software in today's large-scale computing environment, NVIDIA Nsight Systems introduces new features that support application developers using microservices.
Key features include:
Multi-node analysis with scripts called recipes that visualize your application’s execution with a Jupyter Notebook integration. Recipes guide you in optimizing GPU use, CUDA kernel time, and more. New recipes are available for key networking metrics like Infiniband throughput, and overlapped compute and networking. You can now see how inter-node communications relate to compute activity, to diagnose issues and boost performance.
Profiling support for popular container systems like Kubernetes, and support for cloud service providers Kubernetes services, including Amazon Web Services, Azure, Google, and Oracle.
A remote user interface streaming container enables you to read performance reports remotely to inspect them and make optimizations.
Optimized Python for AI and deep learning integrates with Jupyter Lab. Profile Python and other languages supported directly in Jupyter, including detailed analysis with the full Nsight Systems user interface.
Download Nsight Systems: developer.nvidia.com/nsight-systems
Get the NVIDIA Nsight Tools Jupyter Lab Extension: pypi.org/project/jupyterlab-nvidia-nsight
Check out the #GTC24 session scaling AI applications to the data center and cloud: nvidia.com/gtc/session-catalog/?tab.allsessions=1700692987788001F1cG&search=&search.h1topich1pppacceleratedcomputingtoolstechniquesp=1699467437927004CJGa#/session/1696026798429001xqAU
Get started with Nsight Developer tools for CUDA: youtube.com/playlist?list=PL5B692fm6--ukF8S7ul5NmceZhXLRv_lR
Explore more NVIDIA Nsight Developer Tools: developer.nvidia.com/tools-overview
Learn more: developer.nvidia.com
0:00 - Introduction
0:30 - Multi-node analysis
0:57 - Network profiling
1:27 - Profiling Kubernetes
1:40 - Remote server development
1:52 - Jupyter Lab integration
2:14 - Conclusion
#AI #GenerativeAI #Datacenter #Cloud #Nsight #CUDA
Applications are harnessing generative AI, and scaling out to the GPU-accelerated data center and cloud. To help you build software in today's large-scale computing environment, NVIDIA Nsight Systems introduces new features that support application developers using microservices.
Key features include:
Multi-node analysis with scripts called recipes that visualize your application’s execution with a Jupyter Notebook integration. Recipes guide you in optimizing GPU use, CUDA kernel time, and more. New recipes are available for key networking metrics like Infiniband throughput, and overlapped compute and networking. You can now see how inter-node communications relate to compute activity, to diagnose issues and boost performance.
Profiling support for popular container systems like Kubernetes, and support for cloud service providers Kubernetes services, including Amazon Web Services, Azure, Google, and Oracle.
A remote user interface streaming container enables you to read performance reports remotely to inspect them and make optimizations.
Optimized Python for AI and deep learning integrates with Jupyter Lab. Profile Python and other languages supported directly in Jupyter, including detailed analysis with the full Nsight Systems user interface.
Download Nsight Systems: developer.nvidia.com/nsight-systems
Get the NVIDIA Nsight Tools Jupyter Lab Extension: pypi.org/project/jupyterlab-nvidia-nsight
Check out the #GTC24 session scaling AI applications to the data center and cloud: nvidia.com/gtc/session-catalog/?tab.allsessions=1700692987788001F1cG&search=&search.h1topich1pppacceleratedcomputingtoolstechniquesp=1699467437927004CJGa#/session/1696026798429001xqAU
Get started with Nsight Developer tools for CUDA: youtube.com/playlist?list=PL5B692fm6--ukF8S7ul5NmceZhXLRv_lR
Explore more NVIDIA Nsight Developer Tools: developer.nvidia.com/tools-overview
Learn more: developer.nvidia.com
0:00 - Introduction
0:30 - Multi-node analysis
0:57 - Network profiling
1:27 - Profiling Kubernetes
1:40 - Remote server development
1:52 - Jupyter Lab integration
2:14 - Conclusion
#AI #GenerativeAI #Datacenter #Cloud #Nsight #CUDA
