T
traeai
Sign in

产品

Blackwell

别名:Blackwell GPU

NVIDIA 推出的用于高性能计算和人工智能的芯片架构。

已跟踪 6 条高相关材料

TraeAI 观察

相关材料

已收录 6 条与 Blackwell 相关的内容,按评分排序。

英伟达重新思考AI TCO:为何每Token成本才是唯一重要的指标

NVIDIA advocates for cost per token as the core economic metric for AI infrastructure, replacing traditional measures like compute cost or FLOPS per dollar, emphasizing full-stack optimization to reduce inference costs and enhance business value.

入选理由:每Token成本是衡量AI基础设施经济效益的核心指标,直接反映实际产出效率。

FeaturedArticle#NVIDIA#AI TCO#Inference Optimization#Cost Per Token中文
英伟达版「MacBook Pro」曝光:老黄自研了CPU!

NVIDIA’s ‘MacBook Pro’ Revealed: Huang Built Its Own CPU!

量子位1426 字 (约 6 分钟)
85

NVIDIA is set to launch the N1X chip-based AI-native laptop, targeting MacBook Pro users with ARM architecture + Blackwell GPU (6144 CUDA cores) and 128GB LPDDR5X shared memory — ideal for local AI inference and agent automation, but unsuitable for gaming due to bandwidth limits.

入选理由:N1X芯片含20核ARM CPU + Blackwell GPU(6144 CUDA单元),共享128GB LPDDR5X内存

FeaturedArticle#NVIDIA#N1X#ARM Architecture#AI PC#DGX Spark中文
This NVIDIA remains the strongest platform for large-model inference at scale. Prefill/decode disagg...

NVIDIA platform, through various optimization techniques, becomes the best platform for large-scale model inference, significantly reducing service costs and improving performance.

入选理由:NVIDIA 平台通过预填充/解码分离、Blackwell 原生量化、自定义内核和机架级 NVLink 提高了大规模模型推理的性能。

FeaturedTweet#NVIDIA#Large-scale Model Inference#Optimization Techniques中文
Open source 🤝 NVIDIA

Open source 🤝 NVIDIA

cohere(@cohere)56 字 (约 1 分钟)
75

Cohere与NVIDIA合作,推出优化的Command A+模型,专为NVIDIA Blackwell设计,利用NVIDIA CUDA-X库进行训练。这一合作展示了开源与专有技术的结合,为AI基础设施带来了新的可能性。

入选理由:Cohere与NVIDIA的合作展示了开源与专有技术的结合。

FeaturedTweet#Cohere#NVIDIA#AI#Command A+#Blackwell#CUDA-X中文
TokenSpeed is a brand new inference engine purpose built for speed-of-light agentic workloads.  

Re...

TokenSpeed is a new open-source LLM inference engine optimized for agentic workloads, featuring advanced KV caching, an efficient scheduler, and a modular kernel architecture with multi-silicon support.

入选理由:TokenSpeed 实现了媲美 TensorRT-LLM 的性能与接近 vLLM 的易用性。

FeaturedTweet#LLM Inference#NVIDIA#Open Source#KV Cache#Attention Mechanism中英混合

跨材料问答 · Blackwell

回答基于:Blackwell 相关 6 条材料
    0 / 500

    AI may generate inaccurate information. Please verify important content.