GPUs - Moozonian Search

news.ycombinator.com• Aug 23, 2024• 1 min read

Launch HN: Moonglow (YC S24) – Serverless Jupyter NotebooksHi Hacker News! We’re Leila and Trevor from Moonglow (https://moonglow.ai). We let you run local Jupyter notebooks on remote cloud GPUs. Here’s a quick demo video: https://www.youtube.com/watch?v=Bf-xTsDT5FQ. If you want to try it out directly, there are instructions below.With Moonglow, you can start and stop pre-configured remote cloud machines within VSCode, and it makes those servers appear to VSCode like normal Jupyter kernels that you can connect your notebook to.We built this because we learned from talking to data scientists and ML researchers that scaling up experiments is hard. Most researchers like to start in a Jupyter notebook, but they rapidly hit a wall when they need to scale up to more powerful compute resources. To do so, they need to spin up a remote machine and then also start a Jupyter server on it so they can access it from their laptop. To avoid wasting compute resources, they might end up setting this up and tearing it down multiple times a day.When Trevor used

news.ycombinator.com• Aug 21, 2024• 1 min read

Launch HN: Outerport (YC S24) – Instant hot-swapping for AI model weightsHi HN! We’re Towaki and Allen, and we’re building Outerport (https://outerport.com), a distribution network for AI model weights that enables ‘hot-swapping’ of AI models to save on GPU costs.‘Hot-swapping’ lets you serve different models on the same GPU machine with only ~2 second swap times (~150x faster than baseline). You can see this in action through a live demo where you try the same prompts on different open source large language models at https://hotswap.outerport.com and see the docs here https://docs.outerport.com.Running AI models on the cloud is expensive. Outerport came from our own experience working on AI services ourselves and struggling with the cost.Cloud GPUs are charged by the amount of time used. A long start-up time (from loading models into GPU memory) means that to serve requests quickly, we need to acquire extra GPUs with models pre-loaded for spare capacity (i.e. ‘overprovision’). The time spent on loading models also adds to the cost. Both lead to inefficient

auction.voltagepark.com• May 7, 2024• 1 min read

Show HN: Voltage Park – H100 GPU OrderbookHi HN -Voltage Park is a GPU cloud powered by Nvidia H100s. We own and manage the infrastructure and currently have over 7,000 H100s live. We will be launching an additional 17,000 H100s in the next 2 months.Today, we are launching our H100 Auction Platform with 1,000+ GPUs live. This platform gives users on-demand access to compute for AI training, fine-tuning, and inference. We are growing the platform to over 5,000 H100s in the next month.We created the Auction Platform to make the process of procuring GPUs more transparent and simple. It allows you to see the supply and price of GPUs in real-time. Place an order for the number of GPUs you need and get near-instant access to the compute.We have been running this on-demand auction platform for the past 2 months in Beta. Early customers include Luma AI, 273 Ventures and Mirelo.ai. One great customer comment so far: “I love using Voltage Park auction. It's so empowering to know we can spin up high-perf GPU nodes if we need them at affo

tv.algora.io• May 2, 2024• 1 min read

Show HN: I built vector search for COSS podcasts & livestreamsHey HN! I built COSSgpt using videos from the Open Source Founder Podcast [1] and livestreams from COSS Office Hours [2][3]I transcribed the VODs using Whisper and vectorized fixed-size segments from the transcripts with MPNet on Replicate GPUs. I made these segments overlap a little to prevent semantic meaning being lost inbetween segmentsThen I indexed the vectors using HNSWLib in-memory vectorstore [4] and persisted the entire vectorstore into Tigris object storage [5] to cache multimedia and vectors across all Fly.io regionsI built the app in Elixir, almost entirely server-side rendered with minimal diffs sent to the client over WebSockets using Phoenix LiveView. I also used Livebook [6] a ton when I was building the multimedia processing & ML pipeline. I'm super bullish on Elixir for building webapps and/or MLops!Let me know what you think :) If you're curious you can find the code at https://github.com/algora-io/tv[1]: https://algora.io/podcast [2]: https://tv.algora.io/peerrich

bing.com• Dec 14, 2023• 1 min read

Confidential computing in Microsoft Azure gets a boostHardware-backed confidential computing in Microsoft Azure now includes protected environments for VMs, containers, and GPUs, without the need to write specialized code. One of the biggest challenges ...

github.com• Aug 18, 2023• 1 min read

Show HN: UForm v2 – tiny CLIP-like embeddings in 21 languages and Graphcore APII want to share the most recent model release we have prepared. It's a Vision-Language understanding Transformer.It has 40% fewer parameters than vanilla CLIP while performing much better on text-to-image retrieval, where it's also beneficial that our output embeddings have 2x fewer dimensions (256 vs. 512).Moreover, it supports 21 languages, including popular English, Hindi, Chinese, Arabic, and lower-resource languages like Ukrainian, Hebrew, and Armenian.We have packed the library into ONNX and CoreML, providing PyTorch inference code for CPUs and GPUs and PopTorch code for Graphcore IPUs.Demo: http://usearch-images.com/ Blog: https://www.unum.cloud/blog/2023-08-17-uform-graphcoreLooking forward to your feedback!

bing.com• Feb 11, 2020• 1 min read

Radeon RX 5600 XT vs. GeForce GTX 1060 6GB vs. GTX 1070: 32 Game BenchmarkToday we're running the Radeon RX 5600 XT through more than 30 games for the first time and for comparison we have the older Pascal-based GTX 1070 and 1060 GPUs. We've also thrown in the RTX 2060 ...

bing.com• Jul 16, 2018• 1 min read

PyTorch review: A deep learning framework built for speedPyTorch 1.0 shines for rapid prototyping with dynamic neural networks, auto-differentiation, deep Python integration, and strong support for GPUs Deep learning is an important part of the business of ...

news.ycombinator.com• Jul 4, 2016• 1 min read

Ask HN: Why it is seemingly hard to break duopolies in hardware?I noticed that for desktop Add In Board GPUs there are only nVidia, and AMD.x86 CPU, Intel and AMD.Ram chips, SK Hynix and Samsung (Micron and Elpida were tired in third/fourth place, when Elpida went bankrupt and got purchased by Micron, they temporarily overtook Hynix, but now their combined market share shrunk even further than before Elpida went bankrupt, cementing Hynix in second place).Flat Panels I remember there were only a couple manufacturers, and just by Sony and someone else decided to stop making CRTs they killed the entire CRT market at the same time.Internal Sound Cards, Creative and Asus Xonar.Many other stuff is divided between Asus (ASMedia + spinoffs, like ASRock) and Formosa Plastics (owners of Via, WonderMedia, S3, Centaur...)HDDs now are mostly WD and companies they purchased, or Seagate and the companies they purchased.and the list keeps going on.Many of these look ready to be disrupted by new competitors... but there are few attempts, for example PowerVR started

maximumpc.com• Oct 4, 2012• 1 min read

Maximum PC No BS Podcast 90: Intel Intel vs. Amd vs Nvidia, iPhone 5, and moreThis week it’s a full-house in the Podcast room as Deputy Editor Gordon Ung is back from vacation (and IDF) and is joined by Online Managing Editor Jimmy Thang, Editor Josh Norem and our new intern Chris Zele who was offered the gig despite previously working for the Geek Squad at Best Buy.Kicking things off we have Gordon dishing the dirt he uncovered at the Intel Developer’s Forum. He tells us all about all the gear he saw including Intel’s Ivy Bridge successor, code-named Haswell. He then tells us what he thinks about AMD’s current situation. We also talk about new GPUs from Nvidia and possibly AMD, how Borderlands 2 is awesome, the Dream Machine article which was (finally) posted online, everyone’s favorite Apple phone and Steam’s new Big Picture mode.