Rod Burns的动态

?? Are you attending #SC24? Don’t miss Chenle YU's pre-recorded booth talk at Booth 3015! He’ll be presenting one of the key new features in #OpenMP API version 6.0: #Taskgraphs. ?? If you cannot attend... watch the presentation online: https://lnkd.in/dbTfXbxY This feature introduces a graph execution model for OpenMP task-based programs, reducing contention and overhead while ensuring compatibility with other graph-based paradigms like CUDA graphs. ?? Curious for more details? Check out his paper: https://lnkd.in/dKxh2SV4 ?? #HiPART is excited to leverage this new feature to enhance #OpenMP60 with better efficiency and interoperability!

Great conversation with SiliconANGLE & theCUBE today at #SC24. We discuss the challenges of data management when deploying AI at scale. #IBMStorage Scale has new unique features to support the most difficult challenges.

?? SC24 Bound? Let’s Connect! ?? ? Faster Data Access ? Reduced GPU/CPU Loads ? 90% Fewer File Transfers How MetadataHub Makes Data AI-Ready: 1.) Connect Storage to Applications - Link all storage types (SMB, NFS, S3) directly to users and apps. - Automate pipelines to platforms like Snowflake & Databricks. - Self-service data access + integration with Hammerspace, iRODS, and more. 2.) Prepare Unstructured Data - Capture and organize metadata for AI-readiness. - Boost data quality and ensure provenance. 3.) Reduce Resource Load - Act as a “proxy,” cutting direct file access by 90%. - Lower GPU/CPU demand, optimize storage & performance. Attending SC24? Scan the QR Code to book a meeting with us. #AIReady #MetadataHub #DataManagement #DataAutomation #HPC #SC24

If you thought #storage was no longer relevant for #AI workloads then #think again. It underpins #RAG and #resilience for new #enterprise data which is created at such a rate of change that old techniques like making multiple copies can no longer keep up with. Listen to this conversation at #SC24 to hear how #IBMStorage continues to #innovate and deliver #performant, #economic, and #sustainable solutions that work seamlessly with power-hungry and expensive #GPUs to keep them highly utilised and working efficiently.

Midweek at SC24, and it’s time to face the truth - are your workloads outpacing your storage? Here’s what you need to know today: - Scaling doesn’t have to mean more complexity - find solutions that grow with you - Performance matters - don’t let bottlenecks like NFS hold you back - Reliability is key - your infrastructure should work as hard as you do Jimmy the skeleton is still at Booth #714, waiting on NFS (and maybe some company). Don’t keep him waiting - stop by and see how to move beyond outdated storage. #SC24 #HPC #HPCStorage #Quobyte #AI #Storage

#Cloud providers like Microsoft Azure can benefit from leveraging IBM #Storage Scale to pick up Enterprise AI Models and advanced ML. Leveraging a global #data platform, which will cache on-prem data into the cloud, providers can leverage their dominance in the compute world for these new applications and deliver unparalleled value!

?? Live from SC24, VAST Data Head of Product Tomer Hagay joins Solidigm Account Executive Vaibhav N. to discuss his vision for the groundbreaking possibilities unlocked by Solidigm’s new 122 TB D5-P5336 drive. As a leading innovator in the storage space, VAST offers customers running data intensive workloads access to unprecedented tools like its Insight Engine, which gathers insights on data stored in VAST clusters without having to move or write data. The combination of VAST’s industry-leading tools with Solidigm’s new industry-first drive unlocks entirely new possibilities in data storage and manipulation. Check out the video below, and be sure to follow along with TechArena’s full coverage for an inside look at SC24, and what attendees will do with 122: https://lnkd.in/gvfPw8he #VASTData #Solidigm #D5P5336 #SC24 #Supercomputing #HPC

Do you need to do patent landscaping analysis to understand your chance in AI or any other sector? We can help. Yes, AI is a patent hot spot. But there have been concerns about copyright infringement and misinformation from such AI synthesis. Perhaps we need to change our mindset to open our eyes and think about how to solve our problems without using AI, and to make money by copyright licensing, instead of patent licensing. AI is ONLY a tool, NOT God, use it, but NOT worship it. Yes. AI can do many things, but NOT everything, at least NOT what we are doing. Our AI initiative is: "With our intellectual property (IP), a copyrighted multilingual metadata, we can do what ChatGPT, an icon of AI, can't do in data analytics." Do you or any of your contacts need our expertise and our IP? We're selling, NOT talking. Thanks. P.S. We wrote this statement ourselves, without using any AI tool.

Showing off the first d-Matrix Corsair inference server at #SC24! 60,000 tokens/sec at 1 ms/token latency for Llama3 8B in a single server. Ultra-low latency batched inference throughput for highly interactive LLMs, reasoning agents, and video generation. What will you create? #dmatrix #SC24

#EPPC24 Session - "The Ultimate Guide to What to Use When in #PowerApps: #DataSources and #UserInterfaces" ?? Understand the key differences and when a straightforward flat list or a more complex, feature-rich #ModelDrivenApp is the way to go with Lisa Crosbie, MVP and ??Luise Freese, MVP. Find out more: https://lnkd.in/eEEy67XE