I think it's basically an NPU hanging off a USB connection, so it can be used as a compute device for local LLM inference instead of your existing CPU/GPU.
This entire page (except for the footer) is made with images. Is there any reason they would do this? Are they making these images by hand or are they generated somehow? Just seems like an unusual choice.
This device is selling for $1,999. In the product page on their website, they show the device plugged into a MacBook Pro.
A MacBook Pro already has excellent ability to run local LLMs, so why wouldn't someone just spend that $1,999 on maxing out their MacBook specs (extra RAM and GPU) instead?
Pretty decent but needs more details. Overall I give it a thumbs down. You can buy a 64GB Macbook Pro M1 Max in pretty good condition right now for a good bit less than 2K.
In practice you are better off:
- if youre in the mac ecosystem using 2k and upgrading to the ultra class chip next time you want to buy
- if youre into windows wait for the next ryzen ai max chip slated for next year.
- if you want desktop consider gx10 given that you need a computer to use this anyway
In retrospect it is crazy how good the m1 max series is given the amount of ram and memory bandwidth.
Compatible with Mac and Windows, but no Linux support? That's an odd choice. I would have picked Mac and Linux and de-prioritized Windows as a nice-to-have when there's time to get around to it.
Tell me how long it can run without heating up. It's a good step in the right direction. Hardware reliability and testing still needs to go a long way.
18 comments
[ 0.25 ms ] story [ 38.1 ms ] threadThis device is selling for $1,999. In the product page on their website, they show the device plugged into a MacBook Pro.
A MacBook Pro already has excellent ability to run local LLMs, so why wouldn't someone just spend that $1,999 on maxing out their MacBook specs (extra RAM and GPU) instead?
Perhaps they don't want a new/upgraded machine or prefer a more portable solution.
Because this device has 80 GB, which allows running larger models (which is the whole point of the device).
AFAIK, the next significant step in terms of quality above 32 GB is 256 GB, but 80 GB still buys convenience.
In practice you are better off:
- if youre in the mac ecosystem using 2k and upgrading to the ultra class chip next time you want to buy
- if youre into windows wait for the next ryzen ai max chip slated for next year.
- if you want desktop consider gx10 given that you need a computer to use this anyway
In retrospect it is crazy how good the m1 max series is given the amount of ram and memory bandwidth.
- CIX P1/CD8180 + 32 GB RAM with 30 TOPS integrated NPU
- Houmo M50 + 48 GB RAM with 160 TOPS INT8
Anyone know more?