github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 05:05:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #6233] Strange! Each request consumes an additional 2 seconds when I used /api/embed #81578

New Issue
Closed
opened 2026-05-09 11:28:34 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-05-09 11:28:34 -05:00
Owner
Copy Link

Originally created by @AlbertXu233 on GitHub (Aug 7, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6233

What is the issue?

image OS: win11 device: intel i5 1335U with only iGPU embedding model: shaw/dmeta-embedding-zh-q4 I test with very simple input: ["你好"]

OS

Windows

GPU

Intel

CPU

Intel

Ollama version

0.3.4

Originally created by @AlbertXu233 on GitHub (Aug 7, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6233 ### What is the issue? <img width="809" alt="image" src="https://github.com/user-attachments/assets/d649cc9e-7b95-4146-aeef-4eb564be7047"> OS: win11 device: intel i5 1335U with only iGPU embedding model: shaw/dmeta-embedding-zh-q4 I test with very simple input: ["你好"] ### OS Windows ### GPU Intel ### CPU Intel ### Ollama version 0.3.4
GiteaMirror added the networkingperformance labels 2026-05-09 11:28:34 -05:00
GiteaMirror commented 2026-05-09 11:28:37 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 7, 2024):

What's making the API call? Making a direct API call in linux doesn't show any delays:

$ curl -s localhost:11434/api/version
{"version":"0.3.4"}

$ for i in {1..5} ; do curl -s localhost:11434/api/embed -d '{"model":"shaw/dmeta-embedding-zh-q4","input":"你好"}' | jq -c 'del(.embeddings[0][])' ; done
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":138950724,"load_duration":1587447,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":142275262,"load_duration":1525043,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145625093,"load_duration":1402358,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":148817738,"load_duration":1032026,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145985135,"load_duration":1396395,"prompt_eval_count":4}

$ docker compose logs ollama | grep /api/embed
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  139.237667ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  142.622972ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  145.953288ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  149.176403ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  146.312404ms |   192.168.192.1 | POST     "/api/embed"
<!-- gh-comment-id:2273987726 --> @rick-github commented on GitHub (Aug 7, 2024): What's making the API call? Making a direct API call in linux doesn't show any delays: ``` $ curl -s localhost:11434/api/version {"version":"0.3.4"} $ for i in {1..5} ; do curl -s localhost:11434/api/embed -d '{"model":"shaw/dmeta-embedding-zh-q4","input":"你好"}' | jq -c 'del(.embeddings[0][])' ; done {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":138950724,"load_duration":1587447,"prompt_eval_count":4} {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":142275262,"load_duration":1525043,"prompt_eval_count":4} {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145625093,"load_duration":1402358,"prompt_eval_count":4} {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":148817738,"load_duration":1032026,"prompt_eval_count":4} {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145985135,"load_duration":1396395,"prompt_eval_count":4} $ docker compose logs ollama | grep /api/embed ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 139.237667ms | 192.168.192.1 | POST "/api/embed" ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 142.622972ms | 192.168.192.1 | POST "/api/embed" ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 145.953288ms | 192.168.192.1 | POST "/api/embed" ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 149.176403ms | 192.168.192.1 | POST "/api/embed" ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 146.312404ms | 192.168.192.1 | POST "/api/embed" ```
GiteaMirror commented 2026-05-09 11:28:37 -05:00
Author
Owner
Copy Link

@AlbertXu233 commented on GitHub (Aug 8, 2024):

What's making the API call? Making a direct API call in linux doesn't show any delays:

$ curl -s localhost:11434/api/version
{"version":"0.3.4"}

$ for i in {1..5} ; do curl -s localhost:11434/api/embed -d '{"model":"shaw/dmeta-embedding-zh-q4","input":"你好"}' | jq -c 'del(.embeddings[0][])' ; done
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":138950724,"load_duration":1587447,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":142275262,"load_duration":1525043,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145625093,"load_duration":1402358,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":148817738,"load_duration":1032026,"prompt_eval_count":4}
{"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145985135,"load_duration":1396395,"prompt_eval_count":4}

$ docker compose logs ollama | grep /api/embed
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  139.237667ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  142.622972ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  145.953288ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  149.176403ms |   192.168.192.1 | POST     "/api/embed"
ollama  | [GIN] 2024/08/07 - 17:34:32 | 200 |  146.312404ms |   192.168.192.1 | POST     "/api/embed"

I used python requests in windows
Here is the code:

    def request_ollama(query, host="localhost:11434", ollama_model_name="shaw/dmeta-embedding-zh"):
        """
        使用ollama的embedding策略
        """
        url = f"http://{host}/api/embeddings"

        payload = json.dumps(
                {
                    "model": ollama_model_name,
                    "prompt": query,
                }
            )
        headers = {"Content-Type": "application/json"}

        response = requests.request(
                "POST", url, headers=headers, data=payload, stream=False
            )
        embeddings = json.loads(response.text)
        return [embeddings["embedding"]]
    for i in range(10):
        print(i)
        print(request_ollama("你好"))
<!-- gh-comment-id:2274814845 --> @AlbertXu233 commented on GitHub (Aug 8, 2024): > What's making the API call? Making a direct API call in linux doesn't show any delays: > > ``` > $ curl -s localhost:11434/api/version > {"version":"0.3.4"} > > $ for i in {1..5} ; do curl -s localhost:11434/api/embed -d '{"model":"shaw/dmeta-embedding-zh-q4","input":"你好"}' | jq -c 'del(.embeddings[0][])' ; done > {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":138950724,"load_duration":1587447,"prompt_eval_count":4} > {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":142275262,"load_duration":1525043,"prompt_eval_count":4} > {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145625093,"load_duration":1402358,"prompt_eval_count":4} > {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":148817738,"load_duration":1032026,"prompt_eval_count":4} > {"model":"shaw/dmeta-embedding-zh-q4","embeddings":[[]],"total_duration":145985135,"load_duration":1396395,"prompt_eval_count":4} > > $ docker compose logs ollama | grep /api/embed > ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 139.237667ms | 192.168.192.1 | POST "/api/embed" > ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 142.622972ms | 192.168.192.1 | POST "/api/embed" > ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 145.953288ms | 192.168.192.1 | POST "/api/embed" > ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 149.176403ms | 192.168.192.1 | POST "/api/embed" > ollama | [GIN] 2024/08/07 - 17:34:32 | 200 | 146.312404ms | 192.168.192.1 | POST "/api/embed" > ``` I used python requests in windows Here is the code: ```python def request_ollama(query, host="localhost:11434", ollama_model_name="shaw/dmeta-embedding-zh"): """ 使用ollama的embedding策略 """ url = f"http://{host}/api/embeddings" payload = json.dumps( { "model": ollama_model_name, "prompt": query, } ) headers = {"Content-Type": "application/json"} response = requests.request( "POST", url, headers=headers, data=payload, stream=False ) embeddings = json.loads(response.text) return [embeddings["embedding"]] for i in range(10): print(i) print(request_ollama("你好")) ```
GiteaMirror commented 2026-05-09 11:28:38 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 8, 2024):

If you change localhost to 127.0.0.1 does it run faster?

<!-- gh-comment-id:2275535009 --> @rick-github commented on GitHub (Aug 8, 2024): If you change `localhost` to `127.0.0.1` does it run faster?
GiteaMirror commented 2026-05-09 11:28:38 -05:00
Author
Owner
Copy Link

@AlbertXu233 commented on GitHub (Aug 13, 2024):

If you change localhost to 127.0.0.1 does it run faster?

Amazing! It works! Thanks!

<!-- gh-comment-id:2285710883 --> @AlbertXu233 commented on GitHub (Aug 13, 2024): > If you change `localhost` to `127.0.0.1` does it run faster? Amazing! It works! Thanks!
GiteaMirror commented 2026-05-09 11:28:39 -05:00
Author
Owner
Copy Link

@royjhan commented on GitHub (Aug 13, 2024):

@AlbertXu233 does the issue persist when using localhost? I'm unable to recreate it with your script.

<!-- gh-comment-id:2286838016 --> @royjhan commented on GitHub (Aug 13, 2024): @AlbertXu233 does the issue persist when using localhost? I'm unable to recreate it with your script.
GiteaMirror commented 2026-05-09 11:28:40 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Sep 5, 2024):

By switching to the IP you're likely bypassing hostname resolution and/or proxy configurations which are likely what is adding the latency.

<!-- gh-comment-id:2332423643 --> @dhiltgen commented on GitHub (Sep 5, 2024): By switching to the IP you're likely bypassing hostname resolution and/or proxy configurations which are likely what is adding the latency.
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 networking performance
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#81578
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?