github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9246] question about ＂ollama num parallel＂ #6025

New Issue
Closed
opened 2026-04-12 17:21:54 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 17:21:54 -05:00
Owner
Copy Link

Originally created by @A3shTnT on GitHub (Feb 20, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9246

How does"ollama num parallel"work,and does the default value of 0 mean there is no limit?
I have eight L20 GPUs and want to run the 202GB DeepSeek-Q2 model. Should I adjust the OLLAMA_NUM_PARALLEL to a more reasonable value?

Originally created by @A3shTnT on GitHub (Feb 20, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9246 How does"ollama num parallel"work,and does the default value of 0 mean there is no limit? I have eight L20 GPUs and want to run the 202GB DeepSeek-Q2 model. Should I adjust the OLLAMA_NUM_PARALLEL to a more reasonable value?
GiteaMirror commented 2026-04-12 17:21:54 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 20, 2025):

OLLAMA_NUM_PARALLEL creates a context for each parallel completion. So if you set num_ctx=4096 and OLLAMA_NUM_PARALLEL=4, ollama allocates enough VRAM to hold 16384 tokens. If it is unset, ollama picks a value based on the available memory: if you have lots, the default is 4, if not, ollama uses a value of 1.

<!-- gh-comment-id:2670939272 --> @rick-github commented on GitHub (Feb 20, 2025): `OLLAMA_NUM_PARALLEL` creates a context for each parallel completion. So if you set `num_ctx=4096` and `OLLAMA_NUM_PARALLEL=4`, ollama allocates enough VRAM to hold 16384 tokens. If it is unset, ollama picks a value based on the available memory: if you have lots, the default is 4, if not, ollama uses a value of 1.
GiteaMirror commented 2026-04-12 17:21:55 -05:00
Author
Owner
Copy Link

@A3shTnT commented on GitHub (Feb 20, 2025):

OLLAMA_NUM_PARALLEL creates a context for each parallel completion. So if you set num_ctx=4096 and OLLAMA_NUM_PARALLEL=4, ollama allocates enough VRAM to hold 16384 tokens. If it is unset, ollama picks a value based on the available memory: if you have lots, the default is 4, if not, ollama uses a value of 1.

got it, thank you.😘

<!-- gh-comment-id:2671011094 --> @A3shTnT commented on GitHub (Feb 20, 2025): > `OLLAMA_NUM_PARALLEL` creates a context for each parallel completion. So if you set `num_ctx=4096` and `OLLAMA_NUM_PARALLEL=4`, ollama allocates enough VRAM to hold 16384 tokens. If it is unset, ollama picks a value based on the available memory: if you have lots, the default is 4, if not, ollama uses a value of 1. got it, thank you.😘
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#6025
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?