> For the complete documentation index, see [llms.txt](https://docs.platform9.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.platform9.com/private-cloud-director/gpu/gpu-support-pcd/gpu-faqs.md).
# GPU FAQs
Discover the GPU configurations supported by PCD, including Passthrough and vGPU modes. Learn about GPU profiles, VM scheduling, and resource availability. Get insights on GPU management, limitations.
**Q1. What types of GPU configurations are supported in PCD?**
PCD supports:
* **GPU Passthrough** – Full GPU assigned to a VM
* **vGPU** – Fractional GPU assigned using vendor-defined profiles
***
**Q2. Can a host support both Passthrough and vGPU on the same GPU?**
No. A physical GPU can only operate in **one mode at a time**: either Passthrough or vGPU. space.vars.product\_acronym blocks conflicting configurations during setup.
***
**Q3. What if the host has multiple GPUs?**
You can configure each GPU on the host independently. Currently, we do not support a combination of multiple GPU modes on a single host. For example, all GPUs on a host can be passthrough or vGPU.
***
**Q4. Can I change the GPU mode of a cluster on VMs are already running?**
No. You must ensure no VM is using the GPU before changing its mode. The system will prevent mode changes if the GPU is actively assigned.
***
**Q5. What are vGPU profiles and who defines them?**
vGPU profiles are predefined by NVIDIA. Each profile determines how much GPU memory and what capabilities are given to a VM (e.g., `Nvidia-35`, `Nvidia-40`). Only one profile type can be used per GPU at a time.
***
**Q6. How are GPU-enabled VMs scheduled to GPU hosts?**
GPU-enabled flavors are mapped to GPU-capable host. space.vars.product\_acronym schedules VMs only to hosts with compatible GPU hardware and profiles.
***
**Q7. Can I migrate a VM with a GPU?**
* **Passthrough GPU:** Live migration is not supported for Passthrough gpu VMs.
* **vGPU:** Live migration of vGPU VMs requires newer host components For reference, the minimum versions are:\
libvirt - 8.6.0 or later\
QEMU - 8.1.0 or later\
Linux kernel - 5.18.0
| Host OS | vGPU live migration | vGPU cold migration |
| ---------------- | ------------------------------------------------------------------------ | ------------------- |
| Ubuntu 22.04 | Not available — base packages ship versions below the requirements above | Supported |
| Ubuntu 24.04 | Supported | Supported |
| Rocky Linux 10.2 | Supported | Supported |
***
**Q8. How do I know how much GPU memory is available?**
space.vars.product\_acronym will display GPU resource availability per host or per host aggregate, including free VRAM. This helps admins plan capacity and allocations.
***
**Q9. How do I know what GPU model is on a host?**
GPU Host details in the console will list GPU model, mode (Passthrough/vGPU), available profiles, and current usage.
***
**Q10. Can I add a non-GPU host to a GPU cluster?**
No, Currently, space.vars.product\_acronym does not support it.
***
**Q11. Can we passthrough two GPU cards to single VM?**
Yes, you can passthrough two or more GPU cards to single VM by configuring GPU flavor.
***
**Q12: Does pausing or suspending a GPU VM automatically release underlying GPUs and re-attach on resuming/power on?**
This depends on the host OS and vGPU configuration.
On Ubuntu 24 and Rocky Linux 10.2 hosts with SR-IOV enabled, yes. The vGPU is released when the VM is paused or suspended, and re-attached when the VM is resumed or powered on.
On Ubuntu 22 hosts, and on hosts without SR-IOV, the GPU is not released. It remains allocated to the paused or suspended VM and is not automatically re-attached to another GPU VM.
***
**Q13: Is VM HA supported for GPU-enabled clusters?**
No, VM HA is currently not supported for GPU-enabled clusters.
***
**Q14: Can we over provision GPU passthrough physical PCI resources?**
No, you can not over provision the physical GPU PCI passthrough hardware, unlike CPU,Memory over provision ratios.
***
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://docs.platform9.com/private-cloud-director/gpu/gpu-support-pcd/gpu-faqs.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.