Puces & HW

Xiaomi announced a prototype for their Xiaomi AI Cube. 3 chip system: - Xiaomi Xuanjie O3 - Xiaomi Xuanjie O100 - Xiaomi Xuanjie D100 The specs are impressive, but a bit confusing. The D100 chip (originally for their EVs) supports up to 160GB of RAM, but O100 has the 1.22TB/s memory bandwidth. Perhaps the 1.22TB/s figure is for SRAM? Hard to say definitively. Source:

定义国产推理算力新范式 赋能万卡级AI推理算力规模化落地

The five largest GPU neoclouds now run on very different models. CoreWeave and Nebius report to the SEC; Lambda and Crusoe are private and heading toward IPOs; Groq rebuilt itself as an inference cloud after licensing its LPU technology to NVIDIA. This comparison checks each provider's live rate card, Q2 2026 financials, active and contracted gigawatts, anchor contracts, and SemiAnalysis ClusterMAX tier. Nebius posts the lowest H100 rate and the only published B300 price, Lambda has the cheapest

评论区普遍认可小米新芯片性能亮眼，但质疑功耗与能效比，认为单核仅追平苹果去年产品，多核靠核心数取胜；但也有人认为竞争利好消费者。

Credit to Twitter Post

As the memory shortage continues to cause trouble for hardware makers, Amazon says it is now being forced to pass on the costs to its consumers.

Nvidia is investing $1 billion in Poolside and paying $6 billion to license its technology and hire most of its engineers. Over 100 Poolside staff will move to Nvidia to work on Nemotron. Good news for us!

Me and my team made Atomic Dynamic GGUF quants for Qwen 3.8 27B, so we wanted to see the difference between them by giving each quant the same voxel island creation task First of all we were surprised at how well Qwen 3.8 27B handled the 3D scenes in general, though part of that is probably because all the scenes were voxels quant size top-1 vs BF16 mean KLD decode, RTX PRO 6000 AD-Q4_K_M 17.1 GB 95.6% 0.0113 67 tok/s AD-Q5_K_M 20.2 GB 97.3% 0.0042 57 tok/s AD-Q6_K 25.0 GB 98.7% 0.0011 49 tok/s

Welcome back to TechCrunch Mobility — your central hub for news and insights on the future of transportation.

Here's a brand new Blackwell-native, prefill-optimized 4-bit quant that runs 50% faster on compatible hardware than a Q4 quant of the same memory footprint. And it runs 4-7% faster than other NVFP4 quants as benchmarked on RTX 5090 32GB. Quant Benchmark Speed NVFP4 pp2048 6250 t/s unsloth NVFP4 pp2048 6010 t/s Q4_0 pp2048 4130 t/s Q6_K pp2048 3210 t/s This GGUF also includes a quantized MTP draft head for a good measure. Check it out for all details and specifically recommended settings for 15%

Nvidia continues to pour money into data center development — just as AI data centers bring lots of money into Nvidia.

the Memory crunch continues - Moore’s Law reversed to 2007 levels

Four Tesla V100s from 2017 matched my RTX 5090 on single-request Qwen 3.8 decode. Repo: The 5090 was not being held back. It ran NInfer , a specialist engine built to make this exact model as fast as possible on that GPU. (love this guys work) The V100s ran Qwen3.8's published mixed FP4/FP8 weights unchanged. This should be impossible . NVFP4 was built for Blackwell. The RTX 5090 has native silicon for FP4 and FP8; V100 has none of these advantages. And yet via software I wrote a translator fast

Cerebras CS-4 宣称推理速度比 GPU 快 30 倍，评论区认可性能但质疑功耗、价格与可及性。

评论区普遍认可Cerebras CS-4推理性能强劲，但质疑其功耗、价格、模型更新及可及性；也有人认为其将挑战英伟达。

作者用几何学与 CUDA 编程对随机岛屿照片进行地理定位，全程未使用 LLM 生成。

TerraPower's nuclear power plant possesses a strategic advantage over competitors, especially when chasing after data center deals.

Jane Street has installed Etched's first shipped AI cluster system, and was so impressed, it led another massive round, the startup says.