llama : accept a list of devices to use to offload a model #10497
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
slaren
added
the
breaking change
slaren
force-pushed
the
sl/llama-dev-selection
branch
from
8a793ce to
f4457cb
Compare
slaren
force-pushed
the
sl/llama-dev-selection
branch
from
a6fc1d9 to
d6cf918
Compare
github-actions
bot
added
examples
server
ggml
ggerganov
approved these changes
Nov 25, 2024
This was referenced Nov 25, 2024
arthw
pushed a commit
to arthw/llama.cpp
that referenced
this pull request
Dec 20, 2024
…10497) * llama : accept a list of devices to use to offload a model * accept `--dev none` to completely disable offloading * fix dev list with dl backends * rename env parameter to LLAMA_ARG_DEVICE for consistency
This was referenced Oct 22, 2025
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Outdated suggestions cannot be applied.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.
Adds parameters
--device(-dev) and--device-draft(-devd) to specify a comma-separated list of devices to use with the main and draft models. Use--list-devicesto see the list of available devices.--dev nonewill cause GPU usage to be disabled completely, including for large batches.API changes: adds
devicesparameter tollama_model_params. IfNULL, uses all the available devices (same as the current behavior).ggml_backend_reg_by_nameandggml_backend_dev_by_nameare now case-insensitive.