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 #8649] Short run response duration calculations are off #52114

New Issue
Closed
opened 2026-04-28 22:07:38 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-28 22:07:38 -05:00
Owner
Copy Link

Originally created by @NerdyShawn on GitHub (Jan 29, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8649

What is the issue?

Running the smaller deepseek-r1:1.5b model it seems like the very short time and duration how it is being calculated is off given the duration in the response. It seems since its close to zero it throws off how the time is measured.

Image

 date && time curl -s https://ollama.somecooldomain.lan/api/generate -d '{
  "model": "deepseek-r1:1.5b",
  "prompt": "What is the meaning of life?",
  "stream": false
}' | jq
Tue Jan 28 11:10:42 PM EST 2025
{
  "model": "deepseek-r1:1.5b",
  "created_at": "2025-01-29T04:10:42.569719236Z",
  "response": "<think>\n\n</think>\n\nI am sorry, I cannot answer that question.",
  "done": true,
  "done_reason": "stop",
  "context": [
    151644,
    3838,
    374,
    279,
    7290,
    315,
    2272,
    30,
    151645,
    151648,
    271,
    151649,
    271,
    40,
    1079,
    14589,
    11,
    358,
    4157,
    4226,
    429,
    3405,
    13
  ],
  "total_duration": 256103089,
  "load_duration": 144586102,
  "prompt_eval_count": 10,
  "prompt_eval_duration": 24000000,
  "eval_count": 15,
  "eval_duration": 86000000
}

real	0m0.345s
user	0m0.050s
sys	0m0.016s

The real time it took to execute was under a half second but the durations are all wrong.

OS

Docker

GPU

Nvidia

CPU

Intel

Ollama version

0.5.6-0-g2539f2d-dirty

Originally created by @NerdyShawn on GitHub (Jan 29, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8649 ### What is the issue? Running the smaller `deepseek-r1:1.5b` model it seems like the very short time and duration how it is being calculated is off given the duration in the response. It seems since its close to zero it throws off how the time is measured. --- ![Image](https://github.com/user-attachments/assets/aba179e4-7c51-4c44-8503-2f91e60809a2) --- ```json date && time curl -s https://ollama.somecooldomain.lan/api/generate -d '{ "model": "deepseek-r1:1.5b", "prompt": "What is the meaning of life?", "stream": false }' | jq Tue Jan 28 11:10:42 PM EST 2025 { "model": "deepseek-r1:1.5b", "created_at": "2025-01-29T04:10:42.569719236Z", "response": "<think>\n\n</think>\n\nI am sorry, I cannot answer that question.", "done": true, "done_reason": "stop", "context": [ 151644, 3838, 374, 279, 7290, 315, 2272, 30, 151645, 151648, 271, 151649, 271, 40, 1079, 14589, 11, 358, 4157, 4226, 429, 3405, 13 ], "total_duration": 256103089, "load_duration": 144586102, "prompt_eval_count": 10, "prompt_eval_duration": 24000000, "eval_count": 15, "eval_duration": 86000000 } real 0m0.345s user 0m0.050s sys 0m0.016s ``` The real time it took to execute was under a half second but the durations are all wrong. ### OS Docker ### GPU Nvidia ### CPU Intel ### Ollama version 0.5.6-0-g2539f2d-dirty
GiteaMirror added the bug label 2026-04-28 22:07:38 -05:00
GiteaMirror commented 2026-04-28 22:07:43 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 29, 2025):

prompt_eval_duration + eval_duration = 0.11 seconds
load_duration = .144586102 seconds
eval + load = .254586102 seconds
total_duration = .256103089
difference = .001516987 or 0.6%, which seems like reasonable overhead for doing stuff that's not load or eval related.

<!-- gh-comment-id:2621355215 --> @rick-github commented on GitHub (Jan 29, 2025): prompt_eval_duration + eval_duration = 0.11 seconds load_duration = .144586102 seconds eval + load = .254586102 seconds total_duration = .256103089 difference = .001516987 or 0.6%, which seems like reasonable overhead for doing stuff that's not load or eval related.
GiteaMirror commented 2026-04-28 22:07:43 -05:00
Author
Owner
Copy Link

@NerdyShawn commented on GitHub (Jan 29, 2025):

ahh ok, I didn't see any decimal places or units on the values so I assumed seconds and thought things were way off. Good to see the math is close though.

<!-- gh-comment-id:2621734395 --> @NerdyShawn commented on GitHub (Jan 29, 2025): ahh ok, I didn't see any decimal places or units on the values so I assumed seconds and thought things were way off. Good to see the math is close though.
GiteaMirror commented 2026-04-28 22:07:44 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 29, 2025):

Yep, nanoseconds.

<!-- gh-comment-id:2621742108 --> @rick-github commented on GitHub (Jan 29, 2025): Yep, nanoseconds.
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 bug
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#52114
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?