Configuration and Network
- GPU
- NVIDIA Tesla M60
- CPU
- Intel Xeon E5-2697V4 2.3GHz (8 vCPUs, processor-to-memory ratio of 1:8)
- Host Memory
- 64 GiB
- Network
- SR-IOV网络增强功能,提供高带宽能力;最大带宽/基准带宽标注为“中”
- Delivery method
- Elastic Cloud Server (ECS) cloud server instance, GPU hardware passthrough, XEN virtualization type
- 节点地区
- Tianyi Cloud availability zones equipped with NVIDIA M60 GPU cards (specific regions not listed on the page)
Deployment and Delivery Instructions
The Graphics-Accelerated G2 series of Tianyi Cloud Elastic Cloud Servers is based on NVIDIA Tesla M60 hardware passthrough technology, providing graphics acceleration and single-precision computing capabilities. Taking the g2.2xlarge specification as an example, it is configured with 8 vCPUs, 64 GiB of memory, and 1×M60 GPUs, with 8 GiB of video memory per card, offering 2048 CUDA cores, and supporting DirectX, OpenGL, CUDA, and OpenCL, targeting graphics workloads such as media editing, 3D rendering, and transcoding. The instance uses XEN virtualization and features SR-IOV network enhancement capabilities. Note that this instance does not support specification changes, and when using GPU acceleration, you should avoid relying on the MSTSC remote desktop driver.
Activation Process
- 提交用途、配置、卡数与租期
- 与供方确认资源、网络和正式报价
- 签订服务约定并确认交付
- 验收环境、访问权限与服务指标
Editorial Analysis Inferred from source information
Based on the source, it can be determined that the M60 of this instance belongs to an earlier generation of graphics accelerator cards, focusing on DirectX/OpenGL graphics and single-precision computing, suitable for scenarios such as media editing, 3D rendering, and transcoding, rather than large-model training scenarios that require high video memory or high computing power (8 GiB of video memory per card is an obvious constraint). For deployment, three usage restrictions from the source should be noted: first, the instance does not support specification changes, so the specification should be confirmed before selection; second, accessing via MSTSC will cause the GPU acceleration capability to fail under the WDDM model, so third-party remote tools such as VNC should be used; third, when using a private image, the GRID driver and SR-IOV NIC driver must be pre-installed. Regarding billing, the source page is a specification document and does not provide pricing; quotations should be verified in real time against the official pricing page.
- The source page does not indicate the specific sales regions, so available cities cannot be confirmed
- The source page does not provide the price, currency, or billing unit for this instance
- The source page does not specify the minimum lease term, availability commitment, or SLA
- The source page does not provide storage configuration or inter-card interconnect specifications
- No information on current inventory or available card counts for this instance is provided