NVIDIA Data Center’s Post

View organization page for NVIDIA Data Center

236,048 followers

📣 We just redefined what's possible in graph processing. NVIDIA and CoreWeave delivered a record-breaking 410 trillion traversed edges per second (TEPS) on the Graph500 benchmark, ranking #1. Tapping into our full-stack compute, networking, and software technologies — including the NVIDIA #CUDA platform, Spectrum-X networking, #NVIDIAHopper GPUs and a new active messaging library — we're able to push the boundaries of performance while minimizing the hardware footprint. Read the blog now. ➡️ https://nvda.ws/44nFRep #HPC

  • No alternative text description for this image

Impressive milestone by NVIDIA and CoreWeave! Achieving 410 trillion TEPS on Graph500 is a massive leap in graph processing performance. The combination of CUDA, Spectrum-X networking, Hopper GPUs, and the new active messaging library clearly shows how full-stack innovation can set new industry standards. 🚀 #AI #GreyMatterz

Like
Reply

Whoa, 410 trillion TEPS? That’s next-level graph processing! NVIDIA + CoreWeave crushing the Graph500, total performance goals unlocked. 🚀

Like
Reply
Orlando Emmanuelli

Follow me on LinkedIn id: emmanuelliny MS Health Informatics, SME: Health Care Operations, Revenue Cycle, Urology Practice

47m

and to think i'm having issues building a 1000 node graph on my 5yo laptop

Like
Reply

This is amazing 😯

Like
Reply
Lakshmana Prasad Sreekakulam

Senior Manager Software Engineering (Associated Director) at UnitedHealth Group

14h

This looks like a new benchmark in graph processing. Achieving 410 trillion TEPS is truly next level performance.

Like
Reply
Trevor Sullivan

🚀 Content Creator | ⌨️ Dev Advocate | 💻 Open Source | 🔃 Automation | ✝️ James 1:17

21h

410 trillion graph traversals every single second? That is impossible to comprehend.

Like
Reply
See more comments

To view or add a comment, sign in

Explore content categories