摘要
将GPU高并发处理的优秀能力应用到云计算中,结合利用云计算的低成本和弹性扩展。文章主要讨论通过VT-D硬件技术以及KVM/QEMU实现PCI设备透传,在Openstack开源云平台实现,从而在虚机里使用GPU,可以获得几乎原生的性能。
As 3D rendering, scientific computing and large data fields on the requirements of high-performance computing, The concurrent processing capacity of GPU catch more and more attention and promotion of the industry.At the same time as the popularity of cloud computing, its low-cost and flexible expansion is also sought after by the enterprise and individual users. So how to use of low-cost cloud computing and high-performance GPU in combination has become a very high scientific and commercial value. This paper mainly discusses the use of VT-D hardware technology and KVM / QEMU PCI device pass-through so as to use GPU in the virtual machine.
作者
关昆仑
GUAN Kun-lun(Information Technology Institute, Peking University袁Beijing 100871, China)
出处
《电脑与信息技术》
2016年第6期11-13,共3页
Computer and Information Technology