Sustainable Metal Cloud (SMC)

ML Commons MLPerf® Training 4.0 benchmarks are now live. Click the link to read how we've proven to save up to 50% total energy whilst maintaining industry-leading benchmark training performance. ➡️ bit.ly/3KHl6Qs @MLCommons #Sustainable #AI #Llama #GPT3

By accelerating AI inference efficiently, we're helping businesses make smarter decisions faster—at a fraction of the usual energy footprint. Using MLPerf's Inference Benchmarks to measure process outputs, when compared to industry giants, SMC sits in the top 5% of performers.

🇦🇺 Australia, it’s nearly time. Good luck to all teams participating in Stage 1 of The Generative AI CodeFest 👍 Key Date - Stage 1: Friday, November 29, 2024 [09:00 AM - 12:30 PM] #Hackathon #AI #GPUcloud

Thrilled to be celebrating with Firmus Technologies—we're a finalist for the DCD Awards: Asia Pacific Data Center Project of the Year! We’re honoured to be recognised for our Sustainable HyperCube Deployment, in collaboration with ST Telemedia Global Data Centres. 🌱

Using our patented immersion cooling technology, we cut CO2 by 3.8 tonnes per H100 per year. 🌱 Learn about the benefits and types of immersion cooling, as well as an in-depth understanding of single-phase immersion cooling here: youtu.be/zzi3WoNstwQ

🔧Designed to keep your AI services optimized, resilient, and ready to scale, SMC’s support transforms AI infrastructure management, making it as seamless as it is sustainable. #SustainableAI #GPUCloud

Our systems are among the fastest at processing inputs and delivering real-time results using trained models — reviewed alongside industry giants like Dell, Microsoft, NVIDIA, Oracle, and Google.

AI enthusiasts in Australia, here's your chance to bring your concepts to life. 𝐇𝐚𝐜𝐤𝐚𝐭𝐡𝐨𝐧 𝐝𝐚𝐭𝐞𝐬: November 29 - December 06, 2024 𝐑𝐞𝐠𝐢𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧 𝐝𝐞𝐚𝐝𝐥𝐢𝐧𝐞: Tuesday, November 05, 2024 Learn more: bit.ly/3Nviuqt

Electrical & Controls Engineer, Ken Ng, talks about his experience so far: 🌱 Deploying the first Sustainable AI Factory in Singapore 🚀 Leveraging AI algorithms to further boost efficiency 🌏 A recent work rotation in Tasmania Read his interview here ➡️ bit.ly/3U98ire

NVIDIA-powered GPU clusters that are optimised for AI training and deployment with supporting applications like TensorFlow, PyTorch, RAPIDS, and more. Take your AI projects to the next level — sustainably. 🌱 #AI #GPUCloud #NVIDIA #TensorFlow #PyTorch #RAPIDS #SustainableAI

🤖 Calling all #AI enthusiasts in SEA! Join the #generativeAI Hackathon hosted by @OpenACCorg, NVIDIA and @SMC_FutureAI. Learn from NVIDIA experts and gain hands-on skills. Registration closes on Oct. 13. Register now! nvda.ws/4eaTZu9

Are you an AI enthusiast, keen to build some real-world impact? 𝐇𝐚𝐜𝐤𝐚𝐭𝐡𝐨𝐧 𝐝𝐚𝐭𝐞𝐬: October 29 - November 15, 2024 𝐑𝐞𝐠𝐢𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧 𝐝𝐞𝐚𝐝𝐥𝐢𝐧𝐞: October 13, 2024 (less than 2 weeks away!) ➡️ Learn more: bit.ly/3BpKqZT

Our Solutions Architect, Derek Ngo, discusses tackling deployment challenges, MLPerf results, and details of our immersion cooling technology. Read more: bit.ly/3NqCVox

At SMC, we’re not just moving fast—we’re moving with purpose. Hear from our team, sharing what they love about working at SMC. 💼 We're hiring. Visit smc.co/careers for more. #Sustainable #AI #GlobalImpact #SMC

Our GPU clusters are built for ultimate reliability, delivering enterprise-grade performance alongside strict compliance (SOC 2 compliant). With unmatched network capabilities and dedicated resources, security and flexibility go hand in hand.

🏮 Wishing everyone a bright and prosperous Mid-Autumn Festival from our Singapore team. 🌕

Our energy cost advantage allows Sustainable Metal Cloud to be globally price competitive, regardless of location. 🌏 #AI #CloudComputing #Sustainability #CostEfficiency

In the MLCommons-verified report released in July, SMC’s GPT-3 175B, 512 H100 Tensor Core GPUs submission consumed only 468 kWh of total energy when connected with NVIDIA Quantum-2 Infiniband networking. 🔗 Read more: bit.ly/47eJkMm 🧵 4/4

A question then arises, "Is there a way to determine power efficiency at a more granular level than what PUE offers?" This year’s MLPerf benchmarks not only added tests for two generative AI models but also invited contributions regarding energy consumption. 🧵 3/4

𝐈𝐭'𝐬 𝐭𝐢𝐦𝐞 𝐭𝐨 𝐫𝐞𝐭𝐡𝐢𝐧𝐤 𝐃𝐂 𝐞𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐜𝐲 𝐦𝐞𝐭𝐫𝐢𝐜𝐬. 𝐏𝐔𝐄 𝐬𝐢𝐦𝐩𝐥𝐲 𝐢𝐬𝐧'𝐭 𝐞𝐧𝐨𝐮𝐠𝐡 𝐚𝐧𝐲𝐦𝐨𝐫𝐞. As AI workloads grow, traditional metrics like Power Usage Effectiveness (PUE) are no longer sufficient to gauge true efficiency. 🧵 1/4