search

qdrant / quantization

Apache License 2.0
5 stars 4 forks source link

Build error when cross compiling quantization for arm #22

Open chiru091096 opened 8 months ago

chiru091096 commented 8 months ago

I'm getting below error when cross-compiling quantization for arm64.

warning: quantization@0.1.0: tools/arm-linux-gnueabi-gcc-9.2/bin/arm-linux-gnueabi-ar: 
'app/target/arm-unknown-linux-gnueabi/debug/build/quantization-84a087b7ec779edc/out/libsimd_utils.a': No such file
error: failed to run custom build command for quantization v0.1.0 (https://github.com/qdrant/quantization.git#9395db62

How can I fix this

timvisee commented 8 months ago

Is it possible that you've configured cross-compilation incorrectly? We're doing cross compilation in Qdrant as well targeting ARM64 without issues.

A bunch of architecture specific code is gated by specific platform attributes. It looks like something is going wrong here.

Could you share the command you use for cross compilation?

chiru091096 commented 8 months ago

Is it possible that you've configured cross-compilation incorrectly? We're doing cross compilation in Qdrant as well targeting ARM64 without issues.

A bunch of architecture specific code is gated by specific platform attributes. It looks like something is going wrong here.

Could you share the command you use for cross compilation?

@timvisee , thank you for the reply. did you try building qdrant for ARM(not ARM 64) specifically? I didn't see any condition here for arm https://github.com/qdrant/quantization/blob/939fdb627a8edcf92fd71e3c79017156690850e9/quantization/build.rs#L18-L38

jay11ca39 commented 3 months ago

@timvisee I am able to do cross compilation for aarch64 but for arm it is failing with same as mentioned on this issue. Can you please help on this?

jay11ca39 commented 3 months ago

Hello @IvanPleshkov , Any suggestion for this issue. Is ARM build is possible for quantization ?

IvanPleshkov commented 3 months ago

Thanks for the issue! I gonna remove all C code from repo and this issue will be fixed (draft PR https://github.com/qdrant/quantization/pull/26)