XS6.2 SP1 introduces virtualized GPU technology, vGPU to XenDesktop platform, enabled by HDX 3D Pro and support this technology, new measures have been added to Citrix XenServer to access parameters available from NVIDIA GRID K1 and K2 cards via their libraries NVML
the new parameters are available :.
| Class | Name | Units | description | Enabled by default ? | condition for existence |
| host | gpu_memory_free_ | byte | Unallocated framebuffer memory | No | A supported GPU is installed on the host |
| host | gpu_memory_used_ | byte | memory framebuffer Allocated | No | a supported GPU is installed on the host |
| host | gpu_power_usage_ | mW | consumption power of this GPU | No | A supported GPU is installed on the host |
| host | gpu_temperature_ | C | Temperature of the GPU | No | A supported GPU is installed on the host |
| host | gpu_utilisation_compute_ | (fraction) | Proportion of time during the period of the last sample in which a or more cores was executed on the GPU | No | a GPU support is installed on the host |
| host | gpu_utilisation_memory_io_ | (fraction) | Proportion of time spent on the sample period when the global memory (of the device) was playing or writing on this GPU | No | A supported GPU is installed on the host |
We are limited to the available measures NVML and as such, users should be aware that the measures concern the physical GPU.
Those who seek to use these settings in the comparative analysis may also be interested in some of the previous blogs on GPU benchmarking on Citrix blogs such as:
- / blogs / 2013/12/17 / calibration-CAD-applications-to-xenserver-vgpu-and-nvidia-gate-how-long-is-a-piece-of-string /
- / blogs / 2013/12 / 24 / scripting automating-the-test-of-graphic-intensive-gpu-workloads /
- / blogs / 2013/12/16 / Control-xenserver to use turbo-mode-which-in -3-d-graphically intense-applications /
- / blogs / 2014/01 / 11 / coast to watch-out-for-dark-shadow-side comparison-de-hdx-3d-and-pro -vsga /
in recent versions of XenServer we 've been increasing the number of measurements available, and improving documentation and processes around these warning measures. These new measures complement those broadly documented in Chapter 9 of XenServer 6.2 Administration Guide. The documentation is extensive and covers:
- existing measures and their units, including IOPS per virtual disk, latency, CPU utilization, CPU P-states and many other
- metric details not enabled by default and how to allow them
- Documentation tools to generate the parameters in cSV
- How to activate and display metrics XenCenter
- How to configure alerts and automated emails based on threshold metrics
XenServer measures are available from the XenCenter GUI (details included in this blog), but are also directly accessible sys program. admins. and many are embedded in third party control products such as CA Nimsoft, EGInnovations, MonitorIT Goliath and many others. The availability of these new measures should help XenDesktop users looking to control and use the reference GPU and I hope that soon we will see the new measures incorporated into third party control products.
For developers and administrators seeking enthusiastic writing system monitoring products we also provide development guides on xenserver.org site, including sample code for collection measures. The parameters are available not only from the command line interface XenServer (xe CLI) but also via applets, Python, Java, C # and C API XenServer PowerShell.
0 Komentar