We have said it before, and we will say it again right here: If you can make a matrix math engine that runs the PyTorch framework and the Llama large language model, both of which are open source and both of which come out of Meta Platforms and both of which will be widely adopted by enterprises, then you can sell that matrix math engine. …
When Intel announced its “Falcon Shores” project to build a hybrid CPU-GPU compute engine back in February 2022 that allowed the independent scaling of CPU and GPU capacity within a single socket, it looked like the chip maker was preparing to take on rivals Nvidia and AMD head on with hybrid compute motors, which Intel calls XPUs, AMD calls APUs, and Nvidia doesn’t really have if you want to be strict about what its “superchips” are and what they are not. …
Nvidia is not the only company that has created specialized compute units that are good at the matrix math and tensor processing that underpins AI training and that can be repurposed to run AI inference. …
Intel has made another big move toward its ambition to dominate the artificial intelligence space in the datacenter, acquiring Israeli AI chipmaker Habana Labs for $2 billion. …
All Content Copyright The Next Platform