Comments (6)
Yeah, I think previously the mlx-swift side didn't include the accelerate back end:
https://github.com/ml-explore/mlx-swift/blob/main/Package.swift#L78
(these are files not to build)
// opt-out of these backends (using metal)
"mlx/mlx/backend/no_metal",
"mlx/mlx/backend/accelerate",
When I prepared the build for v0.14.0 I had to revamp the build quite a bit to deal with the JIT and I reconciled what CMake does with the swiftpm build.
So you are right, it has probably been like this for a while but we were not using it and so had not seen this.
from mlx.
@davidkoski can you share some steps for how to reproduce this? I'm building mlx-swift main using Xcode 15.3 and macOS 14.5 and the build is fine. Do you only get the error on 14.4? Or do I need to build something specific to see the issue?
from mlx.
The easiest is to try building this branch for iOS:
You will to change various vecLib
includes:
- #include <vecLib/vDSP.h>
+ #include <Accelerate.h>
but those are trivial
from mlx.
I am using Xcode 15.3 on macOS 14.4 but I don't think it is environment dependent -- the key is building for iOS.
from mlx.
Thanks I can reproduce it now. I'm a little confused by what's going on here. The code that is failing has been there for a while now.. so any idea what changed that broke the compilation?
from mlx.
This is safe to close I believe.
from mlx.
Related Issues (20)
- Large ops with multiprocess fail with internal error HOT 3
- [Feature Request] support dtype in mlx.core's module initialization HOT 1
- [BUG] NaN when using max reduction + compile HOT 1
- [Feature] Metal inverse (`mx.linalg.inv`) HOT 4
- [BUG] Missing types HOT 6
- [BUG] Typo: traditinoal -> traditional
- [Feature] cross product HOT 1
- Sample packing using mx.fast.scaled_dot_product_attention? HOT 1
- [Feature] Make APIs non-blocking HOT 3
- [Feature] Enable Metal argsort, sort for > 2M elements along an axis
- [Feature] searchsorted HOT 4
- [Feature Request] Support convolution backward for groups > 1. HOT 1
- [BUG] 'mlx.core.linalg.qr' doesn't work on gpu, there's an abort. HOT 1
- [BUG] HOT 8
- [BUG] type stubs are missing for python 3.9 and 3.10 distributions
- [Feature Request] Adding option to follow PyTorch API in GELU initalization HOT 1
- [BUG] Segmentation fault while running custom operations HOT 4
- Memory Leakage Issue in MLX 0.16 HOT 9
- [PERFORMANCE] grads for bitwise ops + indexing HOT 1
- [Performance] Linear Layer Benchmark HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from mlx.