Re:Intel to Develop Discrete GPUs, Hires AMD's Raja Koduri a...
by ironchef253 on Wednesday November 08, 2017 @10:27PM
(#15659594)
Intel needs a CEO that skates to where the puck is going to be, not where it was three goals ago.
Re:Intel to Develop Discrete GPUs, Hires AMD's Raja Koduri a...
by keganunderwood on Saturday November 11, 2017 @03:13AM
(#15675462)
That's scary of that's what is required of a ceo. You would either need to be an oracle and predict where the industry will go or you'd need to make the industry go the direction you're taking the company.
Re:Intel to Develop Discrete GPUs, Hires AMD's Raja Koduri a...
by dirtyaura on Thursday November 09, 2017 @01:44AM
(#15660380)
I think there is more to AI and ML than deep learning in which TPUs and GPUs are obvious choice.
But I bet that branching instructions (various variants of search) still play a big role when you go beyond classifiers to reinforcement learning etc. so there is need for other architectures beyond GPUs.
Re:Intel to Develop Discrete GPUs, Hires AMD's Raja Koduri a...
by stochastic_monk on Thursday November 09, 2017 @12:39AM
(#15660147)
I do love AVX512 and FMA instructions for CPU code with lots of branching. But that's not definitely going to cut it for the pure almost solely linear algebra workloads of most deep learning.
I do have high hopes for their memristor initiative, but that's got to be years out.
The GPU move is smart for Intel.
How many machine learning strategies is Intel going to try? Does it even know what it's doing? Spending billions of dollars left and right on totally different machine learning technologies kind of looks like it doesn't, and it's just hoping it will get lucky with one of them.
And even if you think that's not a terrible strategy to "see what works", there's still the issue that they need to have great software support for all of these platforms if they want developer adoption. The more different machine learning strategies it adopts, the harder that's going to be for Intel to achieve.