Vasily Aristarkhov’s Post

CONGATEC conga-TC700 COM Express Compact Computer-on-Modules are based on the Intel Core Ultra processors (code named Meteor Lake). Providing a combination of heterogeneous compute engines, including CPU, GPU and NPU, the modules are suitable to run demanding AI workloads at the edge. Up to 6 P-Cores, up to 8 E-Cores, and 2 Low Power E-Cores support up to 22 threads, making it possible to consolidate distributed devices onto a single platform for the lowest total cost of ownership. The SoC-integrated Intel Arc GPU with up to 8 Xe Cores and up to 128 EUs can handle stunning graphics up to 2x 8K resolution and ultra-fast GPGPU-based vision data (pre)processing. The integrated NPU Intel AI Boost executes machine learning algorithms and AI inferences particularly efficient. Up to 96 GB DDR SO-DIMM with in-band ECC at 5600 MT/s contributes to power-efficient high data throughput and low latency.

👉 Intel's Power Play: New CPUs and AI Accelerator to Rival NVIDIA ✍Intel is stepping up its game in the AI and high-performance computing (HPC) arena with the launch of its new 𝐗𝐞𝐨𝐧 𝟔 𝐂𝐏𝐔 and  𝐆𝐚𝐮𝐝𝐢 𝟑 𝐀𝐈 𝐚𝐜𝐜𝐞𝐥𝐞𝐫𝐚𝐭𝐨𝐫. ➧Xeon 6 CPU: This powerhouse offers double the performance of its predecessor, making it a formidable choice for AI and HPC workloads. ➧Gaudi 3 AI Accelerator: This cutting-edge chip is specifically designed to handle large-scale generative AI applications, such as creating text or images. With these new offerings, Intel is clearly aiming to compete with NVIDIA, especially amidst rumors of a potential takeover. These moves demonstrate Intel's commitment to staying at the forefront of AI technology. #Intel #AI #HPC #Technology #Innovation #Xeon6 #Gaudi3

3XS #AI workstations make starting your AI journey easy! Powered by INTEL Core i9 Processors and NVIDIA GPUs, they provide data scientists with a cost-effective platform for developing AI models, prior to scaling up to datacentre-grade training hardware which is also available from Scan. Find out more here - https://bit.ly/3J5pLuJ Find out more about INTEL powered Workstations - https://bit.ly/3XVblFY #intelpowered Get in touch with the Scan AI team to discuss your requirements - [email protected] / 01204 474210 #deeplearning #datascience

📢💻Michael Kanellos's (Marvell Technology) latest blog in EE Times | Electronic Engineering Times "Chip Odyssey 2006 - The Start of the Modern Chip Era" dives deep into the seismic shifts in semiconductor technology that took off in 2006, from GPUs and chiplets to custom XPUs. It’s always satisfying to see 650 Group, LLC's assertions referenced in print, especially around the potential of XPUs to reduce power consumption by 20% or more over GPUs. The future of computing leans heavily on specialized chips, and Michael’s piece captures just how pivotal 2006 was in steering the industry toward this new era. 👉Check it out: https://lnkd.in/gskNPpf6 #Semiconductors #ChipInnovation #XPU #Efficiency #TechAnalysis

Is AMD Instinct™ MI300X affected by CVE-2023-4968 (GPU memory leak). AMD has the answer. Official announcement on May 7, 2024.   This article was published on May 21, 2024.   Preface: When I see the vulnerability it shows the date far away from now. Sometimes I lose interest. Maybe I'm missing a major technical detail. AMD officially released CVE-2023-4869 on March 7, 2024. It happened to wake me up! Although today is May 21, 2024, it seems that my study is not late!   Background: Is MI300X better than H100? While both GPUs are capable, the MI300X has the edge in memory-intensive tasks like rendering large scenes and simulations. In comparison, the H100 excels in its AI-enhanced workflow and ray-traced rendering performance. AMD Instinct™ MI300X accelerators are designed to deliver leadership performance for Generative AI workloads and HPC applications.   Vulnerability details: Insufficient clearing of GPU memory could allow a compromised GPU kernel to read local memory values from another kernel across user or application boundaries leading to loss of confidentiality.   Official announcement: Please refer to the link for details – https://lnkd.in/geEH_jSp

Intel Sees ‘Huge’ AI Opportunities For Xeon—With And Without Nvidia: https://lnkd.in/evZHZSWS Intel Corporation explains why its newly launched Xeon 6900P processors, which scale up to 128 cores and 8,800 megatransfers per second in memory speed, are a big deal for AI computing, whether it’s for CPU-based inferencing or serving as the host CPU for NVIDIA-accelerated systems. Intel said its latest Xeon processors present “huge” opportunities for channel partners in the AI computing space, whether the CPUs are used for inferencing or as the head node for systems accelerated by expensive, energy-guzzling chips like Nvidia’s GPUs. #semiconductor #manufacturing #technology #innovation #semiconductormanufacturing #chips #ai

AMD reveals core specs for Instinct MI355X CDNA4 AI accelerator — slated for shipping in the second half of 2025 🤩 AMD Instinct MI325X has officially launched #AMD provided more details on its upcoming Instinct MI350 CDNA4 #AI accelerator and data center GPU today, formally announcing the Instinct MI355X. It also provided additional details on the now-shipping MI325X, which apparently received a slight trim on #memory capacity since the last time AMD discussed it. MI355X is slated to begin shipping in the second half of 2025, so it's still a ways off. However, AMD has seen massive adoption of its AI accelerators in recent years, with the MI300 series being the fastest product ramp in AMD's history, so like NVIDIA, it is now on a yearly cadence for product launches. AMD is presenting the MI355X as a "preview" of what will come, that means some of the final specifications could change. It will support up to 288GB of HBM3E memory, presumably across eight stacks. AMD said it will feature 10 "#compute elements" per #GPU, which really doesn't tell us much about the potential on its own, but AMD did provide some other initial specifications. A big thank you to Jarred Walton and Tom's Hardware for the full article with more background and insights via the link below 💡🙏👇 https://lnkd.in/e6z3c_RF #semiconductorindustry #semiconductors #semiconductor #chip #it #datacenter #tsmc #chips #innovation #technology #chiplet #tech #technology #computer #server #computer #taiwan #usa #china #ic

When a CPU TDP is 125W, it means that 125W heat is generated under "full load". To run a CPU of 125W, you will also need to remove 125W heat from the data center. Cooling+operating a single CPU machine will cost at least 350W. H100 Nvidia chips required for LLM take up to 700W. That means at least 1600W per chip https://lnkd.in/gb4Hinmd Close to what a 2-Ton AC consumes. Further, removing the heat from earth isn't that easy. Replacing humans have never been this costly in any industrial revolution. https://lnkd.in/g4m3PtWC

@Xi Delivers 6.3 PetaFLOPS AI Supercomputer to UW ARCC By the end of September 2024 the Advanced Research Computing Center of the University of Wyoming (ARCC) has completed the operational deployment of a 6.3 PetaFLOP Supercomputer provided by @Xi Computer Corp. This new UW ARCC Supercomputer consists for a mix of CPU and GPU nodes. There are 25 1U Xi BladeRAIDer for a total of 50 AMD EPYC 9454 48-Core CPU nodes, each capable of 259,105 MOps/Sec. Floating Point Math. Also there are 30 4U Xi NetRAIDer Cluster nodes each carrying 8 GPU, with a total of 48 NVIDIA L40S GPU each capable of 91.6TF FP32 operations, 64 NVIDIA A30 GPU each capable of 5.2TF FP64 operations & 48 NVIDIA H100 SXM each capable of 34TF FP64 operations. Nodes are connected via a 200MHz NVIDIA InfiniBand network and 200 Gbps Ethernet network. Not counting the CPU nodes, this new Supercomputer power reaches up to a remarkable 6.3 PetaFLOPS FP64/32 peak performance. This new Xi Supercomputer is dedicated to substantially increase the computational capabilities of UW ARCC supporting the most advanced Scientific and Artificial Intelligence research projects. Switches and network cables are also part of the awarded contract to enable a full integration with the existing infrastructure. This is one of many client success stories that demonstrate the ability of @Xi to efficiently configure, optimize and deliver nationwide, HPC turnkey Supercomputer Clusters with the highest level of quality, reliability and support, maximizing our clients ROI. Learn More: https://bit.ly/4fixioj

🤩Explore our G593-ZD1: Powering the Future of Intelligence.💫 The 技嘉 AI flagship G593 family welcomes a new member today!🎉 The G593 family features an upper 1U designed to support dual AMD EPYC 9004 processors, memory, and other components, while the lower 4U space is dedicated to the highest-density NVIDIA H100 SXM5 8-GPU modules. 💡The G593-ZD1 doubles the number of root ports from 4 to 8, enhancing the connections between CPUs and GPUs with PCIe lanes. It also adds more FHHL expansion slots for NVIDIA BlueField®-3 DPU through PCIe switches, accelerating AI and HPC data movement, as well as ensuring security isolation.🚀 Read more 😎 https://gct.pse.is/5se6cz GIGABYTE remains highly flexible, continuously striving to offer systems that best suit the evolving needs of the market. Business Inquiry 📬 https://gct.pse.is/5r744r #GIGABYTE #GIGABYTEgroup#GigaComputing #GIGABYTEServer #serversolution #AIserver #AI #MachineLearning #NLP #inference #HPC #HGX #H100