Boosting GPU Virtualization Performance with Hybrid Shadow Page Tables
نویسندگان
چکیده
The increasing adoption of Graphic Process Unit (GPU) to computation-intensive workloads has stimulated a new computing paradigm called GPU cloud (e.g., Amazon’s GPU Cloud), which necessitates the sharing of GPU resources to multiple tenants in a cloud. However, state-ofthe-art GPU virtualization techniques such as gVirt still suffer from non-trivial performance overhead for graphics memory-intensive workloads involving frequent page table updates. To understand such overhead, this paper first presents GMedia, a media benchmark, and uses it to analyze the causes of such overhead. Our analysis shows that frequent updates to guest VM’s page tables causes excessive updates to the shadow page table in the hypervisor, due to the need to guarantee the consistency between guest page table and shadow page table. To this end, this paper proposes gHyvi, an optimized GPU virtualization scheme based on gVirt, which uses adaptive hybrid page table shadowing that combines strict and relaxed page table schemes. By significantly reducing trap-and-emulation due to page table updates, gHyvi significantly improves gVirt’s performance for memory-intensive GPU workloads. Evaluation using GMedia shows that gHyvi can achieve up to 13x performance improvement compared to gVirt, and up to 85% native performance for multithread media transcoding.
منابع مشابه
Modeling and Abstraction of Memory Management in a Hypervisor
Hypervisors must isolate memories of guest operating systems. This paper is concerned with proving memory isolation properties about the virtualization of the memory management unit provided by a hypervisor through shadow page tables. We conduct the proofs using abstraction techniques between high-level and low-level descriptions of the system, based on techniques from previous work on formally...
متن کاملVerification of TLB Virtualization Implemented in C
Efficient TLB virtualization is a core component of modern hypervisors. Verifying such code is challenging; the code races with TLB virtualization code in other processors, with other guest threads, and with the hardware TLBs, and implements an abstract TLB that races with other abstract TLBs and guest threads. We give a general methodology for verifying virtual device implementations, and demo...
متن کاملUsing Shadow Page Cache to Improve Isolated Drivers Performance
With the advantage of the reusability property of the virtualization technology, users can reuse various types and versions of existing operating systems and drivers in a virtual machine, so as to customize their application environment. In order to prevent users' virtualization environments being impacted by driver faults in virtual machine, Chariot examines the correctness of driver's write o...
متن کاملQuest-V: A Virtualized Multikernel for High-Confidence Systems
This paper outlines the design of ‘Quest-V’, which is implemented as a collection of separate kernels operating together as a distributed system on a chip. Quest-V uses virtualization techniques to isolate kernels and prevent local faults from affecting remote kernels. This leads to a high-confidence multikernel approach, where failures of system subcomponents do not render the entire system in...
متن کاملgScale: Scaling up GPU Virtualization with Dynamic Sharing of Graphics Memory Space
With increasing GPU-intensive workloads deployed on cloud, the cloud service providers are seeking for practical and efficient GPU virtualization solutions. However, the cutting-edge GPU virtualization techniques such as gVirt still suffer from the restriction of scalability, which constrains the number of guest virtual GPU instances. This paper introduces gScale, a scalable GPU virtualization ...
متن کامل