github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9139] Intel libggml-cpu-alderlake.so and so on are built on arm64 architecture #31704

New Issue
Closed
opened 2026-04-22 12:25:10 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-22 12:25:10 -05:00
Owner
Copy Link

Originally created by @vt-alt on GitHub (Feb 15, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9139

Originally assigned to: @jmorganca on GitHub.

What is the issue?

Currently (on v0.5.11) Intel microarchitecture libs are built on arm64 too.
But this does not seem useful.

[00:00:07] -- CMAKE_SYSTEM_PROCESSOR: aarch64
[00:00:07] -- Including CPU backend
[00:00:07] -- ARM detected
[00:00:07] -- Performing Test GGML_COMPILER_SUPPORTS_FP16_FORMAT_I3E
[00:00:07] -- Performing Test GGML_COMPILER_SUPPORTS_FP16_FORMAT_I3E - Failed
[00:00:07] -- ARM feature FMA enabled
[00:00:07] -- Adding CPU backend variant ggml-cpu-sandybridge:  
[00:00:07] -- ARM detected
[00:00:07] -- ARM feature FMA enabled
[00:00:07] -- Adding CPU backend variant ggml-cpu-haswell:  
[00:00:07] -- ARM detected
[00:00:07] -- ARM feature FMA enabled
[00:00:07] -- Adding CPU backend variant ggml-cpu-skylakex:  
[00:00:07] -- ARM detected
[00:00:07] -- ARM feature FMA enabled
[00:00:07] -- Adding CPU backend variant ggml-cpu-icelake:  
[00:00:07] -- ARM detected
[00:00:07] -- ARM feature FMA enabled
[00:00:07] -- Adding CPU backend variant ggml-cpu-alderlake:  
[00:00:07] -- ARM detected
[00:00:07] -- ARM feature FMA enabled
[00:00:07] -- Adding CPU backend variant ggml-cpu-sapphirerapids:  
[00:00:07] -- Looking for a CUDA compiler
[00:00:07] -- Looking for a CUDA compiler - NOTFOUND
[00:00:07] -- Looking for a HIP compiler
[00:00:07] -- Looking for a HIP compiler - NOTFOUND
[00:00:07] -- Configuring done (1.3s)

I thought maybe this is my mistake but you build the same way too:

host-189:~$ wget https://github.com/ollama/ollama/releases/download/v0.5.11/ollama-linux-arm64.tgz
host-189:~$ less ollama-linux-arm64.tgz
...
-rwxr-xr-x runner/runner     594128 2025-02-14 06:57 lib/ollama/libggml-base.so
-rwxr-xr-x runner/runner     397056 2025-02-14 06:57 lib/ollama/libggml-cpu-alderlake.so
-rwxr-xr-x runner/runner     397056 2025-02-14 06:57 lib/ollama/libggml-cpu-haswell.so
-rwxr-xr-x runner/runner     397056 2025-02-14 06:57 lib/ollama/libggml-cpu-icelake.so
-rwxr-xr-x runner/runner     397056 2025-02-14 06:57 lib/ollama/libggml-cpu-sandybridge.so
-rwxr-xr-x runner/runner     397056 2025-02-14 06:57 lib/ollama/libggml-cpu-sapphirerapids.so
-rwxr-xr-x runner/runner     397056 2025-02-14 06:57 lib/ollama/libggml-cpu-skylakex.so

Relevant log output

OS

Linux

GPU

Other

CPU

ARM64

Ollama version

v0.5.11

Originally created by @vt-alt on GitHub (Feb 15, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9139 Originally assigned to: @jmorganca on GitHub. ### What is the issue? Currently (on v0.5.11) Intel microarchitecture libs are built on arm64 too. But this does not seem useful. ``` [00:00:07] -- CMAKE_SYSTEM_PROCESSOR: aarch64 [00:00:07] -- Including CPU backend [00:00:07] -- ARM detected [00:00:07] -- Performing Test GGML_COMPILER_SUPPORTS_FP16_FORMAT_I3E [00:00:07] -- Performing Test GGML_COMPILER_SUPPORTS_FP16_FORMAT_I3E - Failed [00:00:07] -- ARM feature FMA enabled [00:00:07] -- Adding CPU backend variant ggml-cpu-sandybridge: [00:00:07] -- ARM detected [00:00:07] -- ARM feature FMA enabled [00:00:07] -- Adding CPU backend variant ggml-cpu-haswell: [00:00:07] -- ARM detected [00:00:07] -- ARM feature FMA enabled [00:00:07] -- Adding CPU backend variant ggml-cpu-skylakex: [00:00:07] -- ARM detected [00:00:07] -- ARM feature FMA enabled [00:00:07] -- Adding CPU backend variant ggml-cpu-icelake: [00:00:07] -- ARM detected [00:00:07] -- ARM feature FMA enabled [00:00:07] -- Adding CPU backend variant ggml-cpu-alderlake: [00:00:07] -- ARM detected [00:00:07] -- ARM feature FMA enabled [00:00:07] -- Adding CPU backend variant ggml-cpu-sapphirerapids: [00:00:07] -- Looking for a CUDA compiler [00:00:07] -- Looking for a CUDA compiler - NOTFOUND [00:00:07] -- Looking for a HIP compiler [00:00:07] -- Looking for a HIP compiler - NOTFOUND [00:00:07] -- Configuring done (1.3s) ``` I thought maybe this is my mistake but you build the same way too: ``` host-189:~$ wget https://github.com/ollama/ollama/releases/download/v0.5.11/ollama-linux-arm64.tgz host-189:~$ less ollama-linux-arm64.tgz ... -rwxr-xr-x runner/runner 594128 2025-02-14 06:57 lib/ollama/libggml-base.so -rwxr-xr-x runner/runner 397056 2025-02-14 06:57 lib/ollama/libggml-cpu-alderlake.so -rwxr-xr-x runner/runner 397056 2025-02-14 06:57 lib/ollama/libggml-cpu-haswell.so -rwxr-xr-x runner/runner 397056 2025-02-14 06:57 lib/ollama/libggml-cpu-icelake.so -rwxr-xr-x runner/runner 397056 2025-02-14 06:57 lib/ollama/libggml-cpu-sandybridge.so -rwxr-xr-x runner/runner 397056 2025-02-14 06:57 lib/ollama/libggml-cpu-sapphirerapids.so -rwxr-xr-x runner/runner 397056 2025-02-14 06:57 lib/ollama/libggml-cpu-skylakex.so ``` ### Relevant log output ```shell ``` ### OS Linux ### GPU Other ### CPU ARM64 ### Ollama version v0.5.11
GiteaMirror added the bug label 2026-04-22 12:25:10 -05:00
GiteaMirror commented 2026-04-22 12:25:12 -05:00
Author
Owner
Copy Link

@vt-alt commented on GitHub (Feb 15, 2025):

I think the problem is in this statement:

if((NOT CMAKE_OSX_ARCHITECTURES MATCHES "arm64")
    OR (NOT CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_SYSTEM_PROCESSOR MATCHES "arm|aarch64|ARM64|ARMv[0-9]+"))
    set(GGML_CPU_ALL_VARIANTS ON)
endif()

If CMAKE_OSX_ARCHITECTURES is undefined (perhaps when this is not OSX) due to NOT and OR this immediately results in TRUE and GGML_CPU_ALL_VARIANTS is enabled. (Negative logic is sometimes hard to follow and error prone.)

<!-- gh-comment-id:2660987170 --> @vt-alt commented on GitHub (Feb 15, 2025): I think the problem is in this statement: ```cmake if((NOT CMAKE_OSX_ARCHITECTURES MATCHES "arm64") OR (NOT CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_SYSTEM_PROCESSOR MATCHES "arm|aarch64|ARM64|ARMv[0-9]+")) set(GGML_CPU_ALL_VARIANTS ON) endif() ``` If `CMAKE_OSX_ARCHITECTURES` is undefined (perhaps when this is not OSX) due to `NOT` and `OR` this immediately results in `TRUE` and `GGML_CPU_ALL_VARIANTS` is enabled. (Negative logic is sometimes hard to follow and error prone.)
GiteaMirror commented 2026-04-22 12:25:12 -05:00
Author
Owner
Copy Link

@vt-alt commented on GitHub (Feb 18, 2025):

Thanks for the fix!

<!-- gh-comment-id:2667050444 --> @vt-alt commented on GitHub (Feb 18, 2025): Thanks for the fix!
GiteaMirror commented 2026-04-22 12:25:12 -05:00
Author
Owner
Copy Link

@mgiessing commented on GitHub (Feb 19, 2025):

@jmorganca @mxyng

This happened to me as well working on ppc64le architecture

[...]
-- Warning: ccache not found - consider installing it for faster compilation or disable this warning with GGML_CCACHE=OFF
-- CMAKE_SYSTEM_PROCESSOR: ppc64le
-- Including CPU backend
-- PowerPC detected
-- Adding CPU backend variant ggml-cpu-sandybridge: -mcpu=power10 
-- PowerPC detected
-- Adding CPU backend variant ggml-cpu-haswell: -mcpu=power10 
-- PowerPC detected
-- Adding CPU backend variant ggml-cpu-skylakex: -mcpu=power10 
-- PowerPC detected
-- Adding CPU backend variant ggml-cpu-icelake: -mcpu=power10 
-- PowerPC detected
-- Adding CPU backend variant ggml-cpu-alderlake: -mcpu=power10 
-- PowerPC detected
-- Adding CPU backend variant ggml-cpu-sapphirerapids: -mcpu=power10 
-- Looking for a CUDA compiler
-- Looking for a CUDA compiler - NOTFOUND
-- Looking for a HIP compiler
-- Looking for a HIP compiler - NOTFOUND
-- Configuring done (0.7s)
-- Generating done (0.0s)

Could we instead change the conditional check if we are on an x86 system like this?

if((CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_OSX_ARCHITECTURES MATCHES "arm64")
    OR (NOT CMAKE_OSX_ARCHITECTURES AND CMAKE_SYSTEM_PROCESSOR MATCHES "x86_64|i[3-6]86|x86|AMD64|Win64"))
    set(GGML_CPU_ALL_VARIANTS ON)
endif()

Thank you!

Edit: Forgot the already merged part

<!-- gh-comment-id:2668668291 --> @mgiessing commented on GitHub (Feb 19, 2025): @jmorganca @mxyng This happened to me as well working on ppc64le architecture ```bash [...] -- Warning: ccache not found - consider installing it for faster compilation or disable this warning with GGML_CCACHE=OFF -- CMAKE_SYSTEM_PROCESSOR: ppc64le -- Including CPU backend -- PowerPC detected -- Adding CPU backend variant ggml-cpu-sandybridge: -mcpu=power10 -- PowerPC detected -- Adding CPU backend variant ggml-cpu-haswell: -mcpu=power10 -- PowerPC detected -- Adding CPU backend variant ggml-cpu-skylakex: -mcpu=power10 -- PowerPC detected -- Adding CPU backend variant ggml-cpu-icelake: -mcpu=power10 -- PowerPC detected -- Adding CPU backend variant ggml-cpu-alderlake: -mcpu=power10 -- PowerPC detected -- Adding CPU backend variant ggml-cpu-sapphirerapids: -mcpu=power10 -- Looking for a CUDA compiler -- Looking for a CUDA compiler - NOTFOUND -- Looking for a HIP compiler -- Looking for a HIP compiler - NOTFOUND -- Configuring done (0.7s) -- Generating done (0.0s) ``` Could we instead change the conditional check if we are on an x86 system like this? ```bash if((CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_OSX_ARCHITECTURES MATCHES "arm64") OR (NOT CMAKE_OSX_ARCHITECTURES AND CMAKE_SYSTEM_PROCESSOR MATCHES "x86_64|i[3-6]86|x86|AMD64|Win64")) set(GGML_CPU_ALL_VARIANTS ON) endif() ``` Thank you! Edit: Forgot the already merged part
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#31704
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?