INTEL NORMALLY THROWS AROUND cores and clock speeds when it comes to demonstrating it's a big fish in the chip world, but the chipmaker's latest willy-waving comes courtesy of deep learning performance.
Intel claims to have set a new deep learning record when it comes to artificial intelligence (AI) image classification using the ResNet-50 neural network model.
The chip giant noted that firing deep learning image classification tasks at its latest generation of scalable Xeon processors, it managed to chew through 7,878 images per second.
That apparently beats Nvidia's Tesla V100 score of 7,844, which will be a bit of a kick in the teeth for Team Green which is pretty good at AI workloads thanks to the parallel processing chops of its GPUs.
Intel is touting this new ResNet-50 record as an example of the flexibility of its Xeon processors at handling different workloads, which should appeal to businesses and developers who don't want to set up dedicated hardware for machine learning, due to costs and other fiddly bits.
"Accelerators are appropriate for certain user scenarios, where dedicated hardware is economically justified. Intel is also developing deep learning accelerators for both inference and training," Intel said.
"However, having the CPU with high deep learning capabilities gives AI customers the flexibility to manage their compute infrastructure uniformly and cost-effectively."
It's worth noting that this Nvidia-beating result was just in ResNet-50, and given Nvidia's tech is used in all manner of machine learning applications, underpinning a lot of self-driving cars for example, Team Green still has plenty in its back pocket to make its data centre-grade GPUs appealing.
But having more companies work on making the running of AI workloads faster is only a good thing, as once the tech trickles down, it means the average Jeremiah, like you and us, could see smarter and more effective features creep into our everyday gadgets. Hopefully, leading to smarter virtual assistants that when asked to book a massage doesn't send us to a place that insists on delivering happy endings. µ
'Some of us like the misery'
That'll surely affect its credit score