Azure FPGA支持的范围？

本文介绍了Azure FPGA支持的范围？的处理方法，对大家解决问题具有一定的参考价值，需要的朋友们下面随着小编来一起学习吧！ 问题描述 首先，如果这是在错误的论坛上，请道歉。我不清楚这属于哪个领域。我在这里问这个问题，因为Azure网站的机器学习部分最多提到了FPGA。First off, apologies if this is on the wrong forum. It's not clear to me which domain this falls under. I'm asking this here since the Machine Learning section of the Azure site makes the most mention of FPGAs.我正在研究基于FPGA的神经网络加速器项目。我对简单地使模型快速工作的兴趣不如我在组织数据管道方面的兴趣，以最大限度地利用FPGA的能力，同时最大限度地重用像DSP这样的稀缺资源。因此，我对使用Project Brainwave网站上列出的一些预制机器学习框架并不感兴趣。I'm working on an FPGA based neural network accelerator project. I'm less interested in simply getting a model working quickly than I am on organizing the data pipeline in a way to best leverage the FPGA's capabilities while maximizing reuse of scarce resources like DSPs. As such, I'm not that interested in using some of the premade machine learning frameworks listed on the Project Brainwave site. Azure是否为寻求完全访问权限的人提供任何解决方案FPGA？我希望能够推动自己的比特流并通过PCIe实现与来自Azure实例的FPGA通信。 AWS以符合我需要的EC2 F1实例的形式提供，但它们仅限于Linux和Xilinx FPGA（并不可怕，但我也想权衡其他选项）。 Azure有任何竞争产品的机会吗？Does Azure offer any solutions to someone who is looking for full access to the FPGA? I'd like to be able to push my own bitstream and communicate with the FPGA over PCIe from an Azure instance. AWS offers in the form of EC2 F1 instances that do fit my needs, but they are restricted to Linux and Xilinx FPGAs (not terrible, but I want to weigh other options too). Any chance that Azure has any competing offering?推荐答案 对于FPGA的  Azure Project Brainwave  提供使用英特尔现场可编程门阵列设计的自定义硬件加速器（FPGA） - 特别是为了加速在Azure云中运行的AI。  For FPGA's Azure Project Brainwave provides custom hardware accelerator designed–using Intel field-programmable gate arrays (FPGAs)–specifically for the purpose of speeding up AI running in the Azure cloud. 如果您正在考虑使用像EC2 F1这样的天蓝色计算实例，那么 ND系列实例针对ML优化的GPU系列。 NDv2实例 正在预览中提供卓越的FP32和FP64性能HPC和AI工作负载，请检查预览 链接 并提供所需的信息。您还可以尝试在提供的Azure实例上安装FPGA开发工具 intel  。If you are looking at using an azure compute instance like EC2 F1 there are ND series instances of GPU family that are optimized for ML. NDv2 instance which is in preview offers excellent FP32 and FP64 performance for HPC and AI workloads, Please check the previewlink  and provide required information. You can also try to install FPGA development tools on Azure instances provided byintel from the marketplace. 这篇关于Azure FPGA支持的范围？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！