Hackernews posts about MI200

1. 01
   

   newVibeSys Builds a Qwen3.5-397B Engine on MI300A, 2.3× Tuned SGLangsyfi.cs.washington.edu

   ▲ 2matt_d6ddiscusssavesaved
2. 02
   

   newDeepSeek V4 Flash on a Single AMD MI300Xgithub.com

   ▲ 382zhoutong73d💬 109savesaved
3. 03
   

   newThe AMD Radeon Instinct MI300A's Giant Memory Subsystemchipsandcheese.com

   ▲ 212pella636d💬 99savesaved
4. 04
   

   newCES 2026: Taking the Lids Off AMD's Venice and MI400 SoCschipsandcheese.com

   ▲ 136rbanffy283d💬 82savesaved
5. 05
   

   newBringing Up DeepSeek-V4-Flash on AMD MI300Xfergusfinn.com

   ▲ 120kkm136d💬 25savesaved
6. 06
   

   newExploring inference memory saturation effect: H100 vs. MI300xdstack.ai

   ▲ 57latchkey680d💬 12savesaved
7. 07
   

   newBoosting Computational Fluid Dynamics Performance with AMD MI300Xrocm.blogs.amd.com

   ▲ 47latchkey639d💬 45savesaved
8. 08
   

   newWhy isn't AMD's MI300X competitive?newsletter.semianalysis.com

   ▲ 46colonCapitalDee171d💬 36savesaved
9. 09
   

   newAn EPYC Exclusive for Azure: AMD's MI300C – By George Cozmachipsandcheese.com

   ▲ 41rbanffy682d💬 14savesaved
10. 10
    

    newAttention is NOT all you need: Qwerky-72B trained using only 8 AMD MI300X GPUssubstack.recursal.ai

    ▲ 20jtatarchuk563d💬 3savesaved
11. 11
    

    newAMD Helios MI400 System Architecture at Hot Chips 2026www.servethehome.com

    ▲ 10pella52d💬 2savesaved
12. 12
    

    newCross-vendor byte-identical inference for a 72B LLM (AMD MI300X vs. Nvidia H100)zenodo.org

    ▲ 9ashsyngh52ddiscusssavesaved
13. 13
    

    newUsing AMD MI300X for High-Throughput, Low-Cost LLM Inferencewww.herdora.com

    ▲ 8technoabsurdist461ddiscusssavesaved
14. 14
    

    newMI300X vs. H100 vs. H200 Benchmark Part 1: Trainingnewsletter.semianalysis.com

    ▲ 5nl313ddiscusssavesaved
15. 15
    

    newAMD Says "Helios" Racks and MI400 Series GPUs on Track for 2H 2026www.nextplatform.com

    ▲ 4rbanffy234ddiscusssavesaved
16. 16
    

    newInter-APU Communication on AMD MI300A Systems via Infinity Fabric: A Deep Divearxiv.org

    ▲ 4matt_d421ddiscusssavesaved
17. 17
    

    newDeepSeek R1 inference performance: MI300X vs. H200dstack.ai

    ▲ 4latchkey577ddiscusssavesaved
18. 18
    

    newSizing Up MI300A's GPUchipsandcheese.com

    ▲ 4klelatti634ddiscusssavesaved
19. 19
    

    newMicrosoft beat H200 Deepseek inference with MI300techcommunity.microsoft.com

    ▲ 3galeos540d💬 1savesaved
20. 20
    

    newAMD MI400 GPU at Hot Chips 2026www.servethehome.com

    ▲ 3pella52ddiscusssavesaved
21. 21
    

    newA Fast Attention Kernel for MI300X, Written in Hip, Not Assemblymoonmath.ai

    ▲ 3latchkey117ddiscusssavesaved
22. 22
    

    newRepomind – 256K context coding agent on a single AMD MI300X (FP8)github.com

    ▲ 3sardor_r1161ddiscusssavesaved
23. 23
    

    newHPE Launches AMD EPYC Venice Instinct MI400 and Nvidia Vera Rubin Compute Bladeswww.servethehome.com

    ▲ 3PaulHoule327ddiscusssavesaved
24. 24
    

    newMy First Multi-GPU Kernel: Writing All-to-All for AMD MI300Xgau-nernst.github.io

    ▲ 3mfiguiere347ddiscusssavesaved
25. 25
    

    newDissecting CPU-GPU Unified Physical Memory on AMD MI300A APUsarxiv.org

    ▲ 3matt_d423ddiscusssavesaved
26. 26
    

    newThe Microsoft Azure HBv5 and AMD MI300Cwww.servethehome.com

    ▲ 3ashvardanian695ddiscusssavesaved
27. 27
    

    newGigabyte G383-R80-AAP1 AMD Instinct MI300A Server Review – ServeTheHomewww.servethehome.com

    ▲ 2rbanffy502d💬 1savesaved
28. 28
    

    newReal world(ish) DeepSeek V4 Flash performance on a single MI300Xmatthusby.github.io

    ▲ 2mhusby56ddiscusssavesaved
29. 29
    

    newAMD doubles rack size for 2027 with Verano CPUs and 144 MI500 GPUswww.techradar.com

    ▲ 2ibobev341ddiscusssavesaved
30. 30
    

    newCreating custom kernels for the AMD MI300huggingface.co

    ▲ 2skidrow377ddiscusssavesaved
31. 31
    

    newCreating custom kernels for the AMD MI300huggingface.co

    ▲ 2skidrow379ddiscusssavesaved

· edit