Christian Binow Lassen

The next KVM in a PowerVM LPAR technology blog is now available. Check it out! This second blog in the series dives into deeper technical details about how CPU & memory management work. Let us know what you think and what questions you have for future blogs. #IBM #PowerVM #KVM #Power #Linux

109

2 Kommentarer

Check out how the world’s first #generativeAI models for #OpenUSD development will be available as NVIDIA NIM microservices.

76

Spotlight on AI Innovation in Norway: Stelia and AQ Compute - Data Centers Power NVIDIA GPU Cloud Providers AI has forced data mobility into the spotlight and the ability to move unprecedented data volumes with speed and simplicity is the key to unlocking innovation and growth. That is why Stelia has partnered with NVIDIA and AQ Compute - Data Centers to revolutionise the way GPU cloud providers in Norway harness the power of AI. Stelia’s data mobility platform enables lightning-fast data transfer, intelligent orchestration, and seamless scalability for AI workloads, and is integrated with NVIDIA's GPU infrastructure to further enhance performance. This synergy empowers NVIDIA cloud partners in Norway, or elsewhere for that matter, to tackle the most demanding AI challenges and deliver transformative business outcomes for their clients. With our platform hosted at AQ Compute in Norway, we're bringing the combined strength of Stelia and NVIDIA right to the heart of the Nordic region's AI ecosystem. NVIDIA Cloud Partners can now leverage this modern platform to drive innovation, reduce time-to-results, and scale their AI offerings without boundaries Norway's AI landscape is evolving rapidly, and we're proud to be at the forefront of this transformation alongside NVIDIA. Together, we're enabling AI innovation and revolutionising the way businesses harness the power of data and AI to drive meaningful, lasting impact. Learn more: [email protected] #DataMobility #AIInnovation #NVIDIA #NVIDIACloudPartners #NorwayAI #AQCompute

29

Oh yes ! x86 getting lazy.. A large number of vulnerabilities in x86 VMWare environments…. Heavily cost increasing VMware virtualization packages… It‘s really time for thinking about a valid more secure, performant, sustainable and very cost efficient alternative, especially for business critical SAP workloads: IBM Power Servers! … combined with, as well, secure, sustainable and cost efficient IBM FlashStorage with FCM4 modules with Ransomware detection built in - a very winning combination which should be really taken into consideration…

25

#Nvidia CEO Jensen #Huang has advanced the idea of remote #datacenters, because he thinks they're where the heavy lifting for #AI can be done with low energy costs and environmental impact. As he explained during #June's #Computex exhibition, the world has enough #energy to run the infrastructure that builds #AI models – it's just not available in densely populated areas. https://lnkd.in/ef6G3dhS

1

Is optics within the datacenter finally here for mainstream deployment?... We all understood the limitations of electrons in network connectivity and the benefits of fiber communication using light. Integrating optics in the datacenter has been challenging due to latencies and power issues with transceivers. However, co-packaged optics (CPO) offers a solution by building transceivers on chips. Notably, companies like Broadcom are advancing CPO technology, but a recent presentation by Intel Corporation at #HotChips24 showcased significant progress. Intel unveiled a 4TBps optical chiplet for XPU-to-XPU connectivity, indicating a promising future for datacenter optics. Check out the details here: [Intel 4TBps Optical Chiplet for XPU-to-XPU Connectivity](https://lnkd.in/gTHNhKtW) #iamintel

9

2 Kommentarer

💡 How can UPMEM PIM transform data-intensive applications? In this insightful session, Sylvan Brocard, Technical Leader on SDK and Applications at UPMEM, dives into the high-level hardware architecture of UPMEM’s PIM-DRAM, programming models, and new SDK features. Learn how these advancements are shaping the future of data processing. 🎥 Watch the full video here: https://lnkd.in/dr7jUBAs 📺 Check out the full ABUMPIMP 2024 Symposium playlist: https://lnkd.in/d6ydf5fZ #UPMEMPIM #PIMDRAM #ProcessingInMemory #PIM #NearMemoryComputing #DataIntensiveApplications #HardwareArchitecture #Programming #SDK #PIMTechnology #TechInnovation #HighPerformanceComputing

13

Breaking news this morning is around NVIDIA and Mistral AI partnering to deliver Mistral-NeMo, a 12B parameter model focused on delivering frictionless AI to the enterprise. - Delivered as a base model and tuned - Uses Apache 2.0 license - 128K token context - FP-8 for performance, scale - Trained on DGX Cloud with Megatron-LM There is a lot to like about this partnership. This is a model that can be easily downloaded by an organization and tuned to run entirely on prem. I like how NVIDIA is so focused on making AI simple for its customers. This is a big deal as complexity and cost are the two biggest challenges I hear from IT folks over and over again. Coverage coming soon, but until then - here's the NVIDIA blog 🔻 🔻 🔻 https://lnkd.in/gKs2C_sb

2

🚀 Introducing the new #LFEdge project Ocre. Ocre is a tiny #OpenSource container runtime for #embedded systems that leverages #WebAssembly and #ZephyrRTOS. Read the launch blog from Atym CEO Jason Shepherd who outlines the mission of the project and initial collaboration with The Zephyr Project and Margo communities, Emerson and Midokura. Project GitHub and Wiki are linked through the blog. 👉 Discover Project Ocre: https://lnkd.in/g9y2gHZJ #RTOS #WASM #WASI #firmware #embeddedsystems #devices #microcontroller #MCU #cloud #edge #cloudnative #edgenative #containers #iot #iiot #connectedproducts #smartcameras #AI #ML #edgeAI #tinyML #computervision #connectivity #sensors #appliances #robotics #drones #autonomousvehicles #consumer #industrial #manufacturing #energy #retail #smartbuildings #agriculture #medical #fed #smartcities #automotive

45

2 Kommentarer

🎉 Big Congratulations to team Atym! 🎉 We're thrilled to see Atym officially launch Project Ocre under the LF Edge umbrella. This groundbreaking project is set to revolutionize how applications are developed, deployed, and managed for billions of resource-constrained edge devices. With Ocre, cloud-native principles are breaking past the "Linux barrier," enabling greater innovation in traditionally embedded devices like sensors, smart cameras, and robots. A huge shoutout to some of the incredible partners who have joined forces with Atym to make this possible: Midokura (a division of Sony), with CTO Dan Mihai Dumitriu, pioneering app containerization for embedded devices using WebAssembly. Emerson, with Director of Engineering Marshall Meier, exploring how Ocre can benefit industrial sensors. The Zephyr Project, led by the The Linux Foundation, whose robust RTOS is the foundation for Ocre’s reference design. Together, we are driving significant advancements in IoT, making development as seamless and efficient as building cloud-native applications. The future of edge computing is here, and it's incredibly exciting to be a part of this journey. Let's continue to push the boundaries of what's possible! 🚀 #EdgeComputing #IoT #CloudNative #Innovation #OmniVentureLabs

9

1 Kommentar

I felt some of you might be interested in learning more about GPUs after my infographic post from yesterday that compared CPUs with GPUs. So for today I felt a more technical look at GPU architectures is warranted, and found this excellent resource in the Chip Letter substack. It's a fair bit more technical but I found it extremely useful. https://lnkd.in/gt5ufVU8 #cpu #gpu #computerarchitecture #ai

13

3 Kommentarer

We have updated the Tiers structure which now also allows anybody to make use of our high performance compute infrastructure 25 minutes for free per week. A complete overview of the available Tiers and Features can be found on the MQS webpage: https://mqs.dk You can now also connect a S3-compatible object storage to MQS Cebule. This keeps you in control of your calculation data. Even if you make use of the object storage provided by us, we do not analyse any of your input nor of your calculation data. In the MQS Dashboard (https://lnkd.in/ga5tBWd8) you can provide the information to the object storage of your choice. See this list of available object storage providers in the EU, EEA or EFTA: https://lnkd.in/e_Qu22Gf

16