# 关于AI芯片架构的争论，GPU好还是TPU好？

2017-05-31 唐杉

## 具体到AI应用，是不是可以说TPU就比GPU好，或者反之呢？

“This tight coupling of TPU2 accelerators to processors is much different than the 4:1 to 6:1 ratios typical for GPU accelerators in deep learning training tasks. The low 2:1 ratio suggests that Google kept the design philosophy used in the original TPU: “the TPU is closer in spirit to an FPU (floating-point unit) coprocessor than it is to a GPU.” The processor is still doing a lot of work in Google’s TPU2 architecture, but it is offloading all its matrix math to the TPU2.”

## 另一方面，硬件架构是取得竞争优势的门槛吗？

“Under The Hood Of Google’s TPU2 Machine Learning Clusters”，这篇文章最后这样说：

“There is not enough information yet about Google’s TPU2 stamp behavior to reliably compare it to merchant accelerator products like Nvidia’s new “Volta” generation. The architectures are simply too different to compare without benchmarking both architectures on the same task. Comparing peak FP16 performance is like comparing the performance of two PCs with different processor, memory, storage, and graphics options based solely on the frequency of the processor.”

“That said, we believe the real contest is not at the chip level. The challenge is scaling out compute accelerators to exascale proportions. Nvidia is taking its first steps with NVLink and pursuing greater accelerator independence from the processor. Nvidia is growing its software infrastructure and workload base up from single GPUs to clusters of GPUs.”

“Google chose to scale out its original TPU as a coprocessor directly linked to a processor. The TPU2 can also scale out as a direct 2:1 accelerator for processors. However, the TPU2 hyper-mesh programming model doesn’t appear to have a workload that can scale well. Yet. Google is looking for third-party help to find workloads that scale with TPU2 architecture.”


• 微信扫一扫
一键转发
• 最前沿的电子设计资讯
请关注“电子技术设计微信公众号”

• 视觉系统方案的有效整合，可使自动化生产线效率极大提高 本文介绍的三个应用案例展示了业界上先进的机器视觉软件和及其图像预处理技术如何促使2D和3D视觉检测的性能成倍提升。
• 继谷歌、甲骨文等之后，IBM中国研究院为什么关闭？下一个 1月23日，有微博爆料：IBM中国研究院正式关闭。此事成了今日热点，继谷歌、诺基亚、摩托罗拉、甲骨文等国际科技巨头接连关闭（大部分）退出中国市场后，IBM也要退出中国市场吗？IBM中国研究院关闭的原因是什么？很多人都在欢呼，“外企信仰”要终结了，国外科技不如国内了......，那么，IBM等的退出对中国科技是利还是弊？
• 或将用于HW 5.0，特斯拉与三星合作开发5nm纯自动驾驶芯 2019年，特斯拉推出了HW 3.0自动驾驶芯片，2020年8月，曾有报道称特斯拉与台积电合作开发 HW 4.0 自动驾驶芯片，采用 7nm 工艺；最近，又有媒体曝光特斯拉与三星合作开发5nm芯片，这次是用于HW 5.0吗？
• 30家国产AI芯片厂商调研分析报告（一） 《电子工程专辑》分析师团队对中国本土的AI芯片设计公司进行了第一手调查和网络汇编整理，从众多AI芯片设计厂商中挑选30家，从核心技术、代表产品、典型应用场景等多个维度进行了分析。无论云端训练和推理、边缘计算还是终端AI，AI都需要高能效的算力支持，而AI芯片无疑是输送算力的硬件保障……
• 低功耗蓝牙SoC利用达角(AoA)定位技术增强IoT资产管理 业界最低功耗的基于闪存的蓝牙低功耗无线电系统单芯片(SoC) RSL10提供Quuppa智能定位系统(Quuppa Intelligent Locating System)。 该方案以用户友好的CMSIS-Pack格式提供，让制造商能设计超低功耗室内资产跟踪应用，具备测向特性和先进的到达角(AoA)技术。

广告
近期热点
广告
广告
广告
广告
广告
广告
可能感兴趣的话题
广告