Support for per-vGPU mode profiling #166
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
For profiling counters in per-vGPU mode, this menu requests for vgpu_id and hw_id of each vGPU when vGPUs are created Signed-off-by: Pengyuan Jiao <pengyuan.jiao@intel.com>
When a vGPU is selected on the menu, the related hw_id is passed to client-c through webui, then per-vgpu filtering by hw_id is achieved in gputop-client-c.c in userspace. Signed-off-by: Pengyuan Jiao <pengyuan.jiao@intel.com>
add new option "--vgpu" for gputop-csv to print counter data in per-vGPU mode. Signed-off-by: Pengyuan Jiao <pengyuan.jiao@intel.com>
Collaborator
|
The new UI has gained the ability to list processes associated to a hw_id (requires kernel 4.15+). |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
These 3 commits are about per virtual GPU counter profiling, which are usage for graphics workloads running inside virtual machines. User can select a specific vGPU on the web ui or through gputop-csv to observe each vGPU performance.
Let me briefly introduce my idea. If we create 2 vGPUs, there are 3 states. ctx_mode = ["Global", "vGPU1", "vGPU2"], and corresponding hw_id = ["0", hw_id1, hw_id2]. The "Global" state means the current state, without applying my patch, that is, it shows the whole counter values for any context id. All of the vgpu_id and hw_id starts from non-zero value. Hence I define the hw_id as 0 to map the "Global" state. So in the beginning without selecting any hw id, the current_hw_id is 0, it works in the "Global" state.