Hackernews posts about HB100

  1. DiffusionGemma: 1k tok/s on an H100, 43 tok/s on a Mac (astgl.com)
    2 points by Jmeg8r 14 days ago | discuss
  2. $2 H100s: How the GPU Rental Bubble Burst (www.latent.space)
    403 points by swyx over 1 year ago | 279 comments
  3. So you want to rent an NVIDIA H100 cluster? 2024 Consumer Guide (www.photoroom.com)
    297 points by ea016 almost 2 years ago | 139 comments
  4. Deploying DeepSeek on 96 H100 GPUs (lmsys.org)
    285 points by GabrielBianconi 10 months ago | 80 comments
  5. Processing Strings 109x Faster Than Nvidia on H100 (ashvardanian.com)
    216 points by ashvardanian 10 months ago | 26 comments
  6. Furiosa: 3.5x efficiency over H100s (furiosa.ai)
    212 points by written-beyond 6 months ago | 160 comments
  7. Huawei's Ascend 910C delivers 60% of Nvidia H100 inference performance (www.tomshardware.com)
    118 points by sien over 1 year ago | 61 comments
  8. Exploring inference memory saturation effect: H100 vs. MI300x (dstack.ai)
    57 points by latchkey over 1 year ago | 12 comments
  9. Colossus AI Supercluster with over 100k Nvidia H100 GPUs (twitter.com)
    47 points by DeathArrow over 1 year ago | 33 comments
  10. Processing Strings 109x Faster Than Nvidia on H100 (ashvardanian.com)
    34 points by samspenc 10 months ago | 3 comments
  11. How to train a model on 10k H100 GPUs? (soumith.ch)
    33 points by bobrenjc93 almost 2 years ago | discuss
  12. Dual RTX 5090 Beats $25,000 H100 in Real-World LLM Performance (www.hardware-corner.net)
    29 points by mdp2021 over 1 year ago | 12 comments
  13. XAI's Memphis Supercluster has gone live, with up to 100,000 Nvidia H100 GPUs (www.datacenterdynamics.com)
    21 points by belter almost 2 years ago | 9 comments
  14. China's Taichi-II Chip: First Optical AI Processor Outperforms Nvidia H100 (techovedas.com)
    13 points by yogrish almost 2 years ago | 4 comments
  15. AMD MI300X vs. Nvidia H100 LLM Benchmarks (blog.runpod.io)
    13 points by aadillpickle almost 2 years ago | 4 comments
  16. Show HN: H100cloud.com – Dutch Auction Cloud Exchange (h100cloud.com)
    11 points by zandrew almost 2 years ago | discuss
  17. Rent H100 SXMs for 0.99 (app.hyperbolic.xyz)
    10 points by hyperbolic_labs over 1 year ago | 5 comments
  18. Ask HN: What is the cheapest ($/hour) H100 renting service with quick spin-up?
    10 points by dtquad over 1 year ago | 4 comments
  19. Meta is using more than 100k Nvidia H100 AI GPUs to train Llama-4 (www.tomshardware.com)
    8 points by turtlegrids over 1 year ago | 1 comments
  20. xAI's Colossus (100k H100 cluster) has begun training (twitter.com)
    7 points by famouswaffles almost 2 years ago | 1 comments
  21. H100 prices are melting *UP (www.latent.space)
    7 points by swyx 3 months ago | discuss
  22. Processing Strings 109x Faster Than Nvidia on H100 (ashvardanian.com)
    6 points by binarymax 10 months ago | discuss
  23. Nvidia B200 vs. H100: Independent Training Performance Tests (www.lightly.ai)
    6 points by isusmelj about 1 year ago | discuss
  24. MI300X vs. H100 vs. H200 Benchmark Part 1: Training (newsletter.semianalysis.com)
    5 points by nl 7 months ago | discuss
  25. Show HN: A GPU group-buying layer to get enterprise rates (~50% off H100s) (trysaga.ai)
    5 points by DdotR over 1 year ago | discuss
  26. Faulty Nvidia H100 GPUs and HBM3 memory caused failures during Llama 3 training (www.tomshardware.com)
    4 points by surfingdino almost 2 years ago | 1 comments
  27. Cohere Open-Sources Command A+, a 218B Moe Model That Runs on Two H100s (firethering.com)
    4 points by steveharing1 about 2 months ago | discuss
  28. Outperforming cuBLAS on H100: A Worklog (cudaforfun.substack.com)
    4 points by skidrow over 1 year ago | discuss
  29. Elon Musk's xAI Colossus supercomputer with 100K H100 Nvidia GPUs comes online (www.tomshardware.com)
    4 points by Link_ke almost 2 years ago | discuss
  30. The Blackmarket for Nvidia H100, A100 AI GPUs Is Exploding (www.techradar.com)
    4 points by pwmtr almost 2 years ago | discuss
  31. The "setup tax" on AWS H100s is killing iterative research
    3 points by miyamotomusashi 6 months ago | 2 comments