Phil Sun

It's unfortunately a fairly big commitment for us to test and release binaries on MacOS, so we're hesitant to accept this change. Based off this PR though, I removed the...

Hey thanks for reporting this bug! Can you provide the CPU model and operating system you're using? This kind of crash generally comes when the CPU tries to execute some...

I've been trying to reproduce this issue but I haven't been able to. I'm also using Ubuntu 20.04.2 LTS with an Ivy Bridge-era CPU (AVX but not AVX2 support). Are...

I installed Conda with Python 3.7 and I also couldn't reproduce. I installed Anaconda 2021.05 from [here](https://repo.anaconda.com/archive/Anaconda3-2021.05-Linux-x86_64.sh) and then did ``` conda create -n py37 python=3.7 anaconda conda activate py37...

Thanks for debugging--I think the issue is that `vpbroadcastq` is an AVX2 instruction, which Sandy Bridge doesn't support. We will look into compiling the ScaNN wheels the next release without...

Here's a related Dockerfile that might help; it compiles a version of TensorFlow Serving linked against ScaNN: https://github.com/google-research/google-research/blob/master/scann/tf_serving/Dockerfile.devel What problems have you encountered with old versions of Bazel?

ScaNN 1.2.8 was recently released and doesn't assume AVX2 support; we now compile with `-mavx` rather than `-mavx2`, and do runtime dispatch to AVX2, when supported, for the important routines....

It's unfortunately not possible to build ScaNN without TensorFlow, because we use a number of internal functions provided by the TensorFlow library (for example, TF [Status](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/platform/status.h) and [StatusOr](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/platform/statusor.h)).

I spun up a Ubuntu 20.04 VM and installed pip packages until I got to a superset of your `pip freeze`; mine is ``` absl-py==1.2.0 astunparse==1.6.3 attrs==19.3.0 Automat==0.8.0 blinker==1.4 cachetools==5.2.0...

My hypothesis (from [here](https://github.com/tensorflow/recommenders/issues/328#issuecomment-1214314586)) is that ScaNN is using AVX2 instructions due to it being compiled with the `-mavx2` flag, and your processor doesn't support AVX2. We will look into...