On: Intel Extension for TensorFlow

[make3]

yes, but Intel now has mainstream GPUs that are reasonably (~ RTX 3060) performant, which is what I assume this is for

[Roark66 → make3]

Are you talking about Intel Arc? I'm yet to see any ML relevant benchmarks on Intel Arc. If you are aware of any please let me know.

[throwaway1851 → Roark66]

Same here. At 16GB of VRAM and only $349, it could fill a really nice slot for DL.

[tttttt5ts → Roark66]

You won't for some time as ARC is only supported on Linux kernel 6+. And honestly, their hardware is pretty bad for ML given you can pick up a used 3090 for $600. I just don't think there will be an ML push for ARC with current gen.

[learndeeply → tttttt5ts]

> And honestly, their hardware is pretty bad for ML given you can pick up a used 3090 for $600.

How do you know it's bad? Do you have benchmarks?

Just checked, used 3090s are going for $700-900 on eBay.

[ShamelessC → learndeeply]

It won’t be able to make use of CUDA, which is more than enough reason.

[doublepg23 → learndeeply]

hardwareswap.reddit.com is the best way to buy used computer hardware.

[fulafel → tttttt5ts]

I think a recent Linux version requirement shouldn't hamper benchmarking if you're a somewhat technical user.

You can get the latest Linux pacakges for eg Ubuntu here: https://wiki.ubuntu.com/Kernel/MainlineBuilds - probably other distributions have similar easy ways since it's needed so often in bug reporting for users to tell if a hw support bug is in a distro specific kernel change or not.

[gnufx → make3]

It presumably has to support Aurora urgently: https://www.alcf.anl.gov/aurora