Hackernews posts about FP64

1. 01
   

   new15 years of FP64 segmentation, and why the Blackwell Ultra breaks the patternnicolasdickenmann.com

   ▲ 202fp64enjoyer262d💬 80savesaved
2. 02
   

   newFP8 Is All You Need (Part 1): Debunking Hardware FP64 as the HPC Holy Grailarxiv.org

   ▲ 17matt_d152d💬 2savesaved
3. 03
   

   newNvidia Blackwell Ultra (GB300): -97% INT8/FP64,+50% FP4 Dense,+55% VRAM,+114% Atresources.nvidia.com

   ▲ 8Sweepi437d💬 3savesaved
4. 04
   

   newAMD Hints at Big FP64 Increases in MI430X GPU as Ozaki Underwhelmswww.hpcwire.com

   ▲ 5rbanffy238ddiscusssavesaved
5. 05
   

   newThe AI era is pulling FP64 hardware away from scientific HPCfortran-lang.discourse.group

   ▲ 3Bostonian143d💬 1savesaved
6. 06
   

   newFP64 Floating-Point Emulation in INT8arxiv.org

   ▲ 3matt_d403ddiscusssavesaved
7. 07
   

   newAMD's FP64 Boost with MI430X Is Even Bigger Than Expectedwww.hpcwire.com

   ▲ 2rbanffy95ddiscusssavesaved
8. 08
   

   newSC25: Estimating AMD's Upcoming MI430X's FP64 and the Discovery Supercomputerchipsandcheese.com

   ▲ 2rbanffy329ddiscusssavesaved
9. 09
   

   newSC25: Estimating AMD's Upcoming MI430X's FP64 and the Discovery Supercomputerchipsandcheese.com

   ▲ 2rbanffy331ddiscusssavesaved
10. 10
    

    newSC25: Estimating AMD's Upcoming MI430X's FP64 and the Discovery Supercomputerchipsandcheese.com

    ▲ 2rbanffy332ddiscusssavesaved
11. 11
    

    newShow HN: IEEE-754-Conformant FP64 on Metal (Apple Silicon)github.com

    ▲ 1guyfischman183d💬 1savesaved
12. 12
    

    newBolt Graphics Targets FP64 HPC Workloads with Zeus GPUwww.hpcwire.com

    ▲ 1rbanffy190ddiscusssavesaved
13. 13
    

    newShow HN: 1B Embeddings

    ▲ 4INVARIAN216d💬 2savesaved
14. 14
    

    newShow HN: HyperAI GPU Leaderboard – A benchmark comparison site for AI workloadshyper.ai

    ▲ 2Ada_trying287d💬 2savesaved
15. 15
    

    newResearch Acceleration Engine: The Arrival of Excellent GPU Rental Service

    ▲ 1chlor692ddiscusssavesaved
16. 16
    

    new4-bit floating point FP4www.johndcook.com

    ▲ 98chmaynard203d💬 79savesaved
17. 17
    

    newThriftAttention: Selective Mixed Precision for Long-Context FP4 Attentionarxiv.org

    ▲ 4joesharratt29166d💬 1savesaved
18. 18
    

    newSupermicro Super AI Station (GB300, 20 Pflops FP4, 748GB Coherent Memory)www.supermicro.com

    ▲ 3peter_d_sherman75d💬 2savesaved
19. 19
    

    newFlux runs 2x faster on 10GB VRAM with Nvidia FP4blackforestlabs.ai

    ▲ 3anjneymidha670ddiscusssavesaved
20. 20
    

    newCustom FP4 CUDA Kernel – 129 Tflops on DGX Spark with Pre-Quantized Weight Cacheforums.developer.nvidia.com

    ▲ 2vkaufmann255d💬 1savesaved
21. 21
    

    newNumber to GPU Float Converter: FP32, FP16, BF16, FP8, and FP4 Explainedwww.bestgpusforai.com

    ▲ 2javaeeeee114ddiscusssavesaved
22. 22
    

    newShow HN: realistic_blas: Exact infinite-precision LA, useful errs, f64 fast pathgithub.com

    ▲ 2timschmidt195ddiscusssavesaved
23. 23
    

    newSol-RL: FP4 Explore, BF16 Trainnvlabs.github.io

    ▲ 2gmays209ddiscusssavesaved
24. 24
    

    newTesting Nvidia's FP4: Running 70B LLMs on a Single RTX 5090 with Real Benchmarksai.gopubby.com

    ▲ 2Aedelon242ddiscusssavesaved
25. 25
    

    newSageAttention3: Microscaling FP4 Attention. 5x Speed uparxiv.org

    ▲ 20xjunhao507ddiscusssavesaved
26. 26
    

    newLarge Language Model Training Using FP4 Quantizationarxiv.org

    ▲ 2t55647ddiscusssavesaved
27. 27
    

    newTflops Gap: Why FP4 Moe Kernel Engineering Matters on Blackwellhuggingface.co

    ▲ 1apsys306ddiscusssavesaved
28. 28
    

    newHuawei unveils Atlas 950 SuperCluster – promises 1 ZettaFLOPS FP4 performancewww.tomshardware.com

    ▲ 1buyucu414ddiscusssavesaved
29. 29
    

    newNvidia's moat is not what it used to be

    ▲ 6florianleibert209d💬 11savesaved
30. 30
    

    newShow HN: Mumble Dictation – local dictation that learns your vocabulary

    ▲ 6tarafans7107d💬 1savesaved

· edit