Depends what AI flavour you want. For LLMs you are either short on Memory (GPUs) or on Memory BW (CPUs). For other ML applications situation might be more nuanced.Not a problem since you can offload AI calculations onto your GPU
That is the idea. I am looking forward to the official reveal but I am afraid they will try to outcompete Apple on higher price and not higher performance... Especially I would prefer miniPC to a laptop for my use case.
It's weird that the quantized score is the least bad compared to the desktop CPU. The other two test types get hit bad by the lack of something, probably cache.
I honestly don't get the scoring for this one, which numbers do we care about and why?
The Quantized one matters more. It's the best balance between speed and accuracy and most likely to be the one used in real world applications. So it's good enough there
Seems kinda meh, maybe? Also I cannot select GPU for anything.The Quantized one matters more. It's the best balance between speed and accuracy and most likely to be the one used in real world applications. So it's good enough there
And your OpenVINO score is going to be much better I think.
Probably AMD driver issue.
Facebook
Twitter