CoreWeave是第一个完全为NVIDIA RTX Pro 6000 Blackwell Server

IT Home在7月15日报告说，AI Cloud Enterprise的新兴公司Coreweave在7月9日当地时间宣布，它将成为第一个完全基于NVIDIA RTX PRO 6000 BLACKWELL SERVER版本数据中心GPU完全提供云实例的云计算平台。 This example is based on 8 RTX Pro 6000 graphics cards, 128 VCPU virtual nuclei based on the fifth generation Intel XEON processor (Emeral Rapids), 1 TB of system memory, 7.68 TB of local NVME storage and 7.68TB with a 100 GBS network performance, and is integrated with the ai -cloud cloud platform AI AI level AI level AI level A AI level ai.在新闻稿中，NVIDIA RTX Pro 6000 Blackwell Server版本的速度比上一代的LLM推理速度快5.6倍，比上一代的LLM产品快3.5倍。足以推断70B参数模型，并且是高性能的GPU组。 “我们已经看到了很多未来的经验e," said Dave Salvator, director of Accelerated Computer Products in Nvidia. The NVIDIARTXPRO 6000 GPU represents an advance in AI and graphics performance, providing sophisticated and profitable solutions for all industries. As the first company to offer the RTX Pro 6000, Coreweave shows how fast the power of Blackwell architecture can be carried out to the market, allowing companies to accelerate innovation and achieve transformative results.他说，与NVIDIA密切相关的Coreweave也于本月初宣布，这是第一个实施其GB300 NVL72系统的IA云服务提供商。