Axel Dittmann’s Post

Hi everyone! This weekend, I decided to dive into AI by setting up Pinokio on my laptop, a system for running AI locally. I wanted to use AI right from my own computer, without needing the cloud. I've got a Ryzen 9 and a GeForce RTX 4070, so I thought I'd have no trouble with anything. Even though these are pretty powerful, I ran into a snag because of not having enough RAM (such a surprise! haha). I only have 16GB, and it looks like I need way more, maybe even 4 times that, especially for running something as demanding as Stable Cascade. It just didn’t work out because of the memory issue, but ComfyUI did run pretty well, which was interesting to see how different AI programs need different things. Despite this, I'm not giving up. I'm planning to test out NVIDIA's "Chat with RTX" next. It sounds cool, and I'm curious to see how it goes with my setup. #AI #ArtificialIntelligence #MachineLearning #LocalAI #TechChallenges #NVIDIA #Ryzen #GeForceRTX #SoftwareEngineering #TechInnovation #ComfyUI #StableCascade #Pinokio

Check out some interesting facts about datacenter grade GPUs vs Consumer grade GPUs👇

In the past month, I had an amazing experience working with a cutting-edge Grace Hopper system from #nvidia on #vultr Cloud. The system's capabilities were truly inspiring as I delved into real-time analytics using GPUs and discovered the immense potential for harnessing high-speed processing power, breaking the boundaries of the limit of GPU memory that limited the volume of data or the performance on traditional systems. After working with the system, I realized the potential of having a high-speed system memory accessible from the GPU and the current limit of our software. After improving the memory management system and optimizing the engine, I could set up and run a dashboard using the AIS data from the last six years. The dashboard is processing around 20 billion records blazingly fast and interactively. The lesson learned is that upgrading to a more powerful system is not a reason to stop improving the speed of your software. On the contrary, it should be an opportunity to enhance it. #nvidia #vutlr #heavyai #extremeanalytics

Hello Connections! Today I brought a very interesting comparison of datacer grade GPUs vs Consumer Grade GPUs on cloud. The idea behind Xtreme Cloud Services is quite simple "CONSUMER GRADE GPUs ON CLOUD". We have noticed that their are companies struggling due to low performing T4 GPUs and if they want to upgrade these 26₹/hr GPUs for 24x7 machines ,the very next GPUs on cloud is A40 that is directly 50-60₹/hr. There are no GPUs between these price range. Here we at Xtreme Cloud Services solves this problem. Even though the T4 has more VRAM, the RTX 3060 really outshines it when it comes to AI inference. It has more CUDA cores and faster memory, which means it can handle complex models more efficiently. Plus, Xtreme Cloud Services offers FOUR DIFFERENT CONSUMER GPU VARIANTS that can easily replace those lower-performing T4s.These Consumer Grade GPUs have advanced features like real-time ray tracing, it’s a fantastic choice for developers looking for top performance WITHOUT BREAKING THE BANK. Find a interesting comparison below👇 #gpu #cloudgpu #AI #LLMs

Struggling to access high-performance GPUs for your projects? Watch how effortlessly you can activate a NVIDIA A100 80GB SXM4 instance with ORIGIN. Our intuitive web-based portal offers seamless and scalable access to a suite of high-performance computing tools and resources, all from a single, easy-to-navigate dashboard. Dive into the future of computing today and unlock the full potential of your projects with ORIGIN. https://lnkd.in/gW6HT2tv 🔗 Get your instance today: https://lnkd.in/dtpAVZBg ⚡ A100s available now from $2.60/GPU/HR with no data egress fees. #cloud #cloudcomputing #nvidia #machinelearning #hpc #ai #deeplearning #bigdata

I spoke with a company in the AI image creation space last week. 40% of their expenditure is going towards the cloud costs associated with model deployment. Mystic leverages MIG to fractionalize GPU power. This customer will now be able to reduce the number of GPUs required from 90 to 45, by deploying SDXL twice on the same A100 80GB. Both models can still be ran in parallel with no reduction in performance. Managing compute costs will be one of the key challenges for companies in the Gen AI space. GPU fractionalization is one of the best techniques to ensure maximum optimization of your ML infra. https://lnkd.in/gsUsqdzP

True availability seems to be a major roadblock in the Cloud GPU space. We are constantly seeing multiple providers promoting low prices masked by hidden fees with no true availability. If you are in the #generativeai space and looking for high performance GPUs let me know, I know some people who can help! #ai #hpc #machinelearning #gpus #cloud #cloudcomputing

CloudWatch Container Insights now delivers observability for NVIDIA GPUs on EKS https://ift.tt/xbu4VrD Amazon CloudWatch Container Insights with Enhanced Observability for EKS now auto-discovers critical health and performance metrics from your NVIDIA GPUs and delivers them in automatic dashboards to enable faster problem isolation and troubleshooting for your AI/ML workloads. Container Insights with Enhanced Observability delivers you out-of-the-box trends and patterns on your infrastructure health and removes the overhead of manual dashboard and alarm set-ups saving you time and effort. via Recent Announcements https://ift.tt/N8TepWL March 11, 2024 at 08:01PM #aws #cloudcomputing

NVIDIA just released Chat with RTX, a personalized AI chatbot that you can run right on your Windows PC! 🤖 🔍 What’s unique? - It’s local: No more cloud servers! Chat with RTX harnesses the power of your NVIDIA GeForce RTX GPU. - Customizable: Personalize your chatbot with your content. - Fast: Lightning-quick responses, thanks to local processing. - Secure: Your data stays on your device—no third parties are involved. #AI #NVIDIA #Chatbot #GeForceRTX

Amazon CloudWatch Container Insights with Enhanced Observability for EKS now auto-discovers critical health and performance metrics from your NVIDIA GPUs and delivers them in automatic dashboards to enable faster problem isolation and troubleshooting for your AI/ML workloads. Container Insights with Enhanced Observability delivers you out-of-the-box trends and patterns on your infrastructure health and removes the overhead of manual dashboard and alarm set-ups saving you time and effort.