Jim Keller Reportedly Joins Intel After AMD, Tesla Stint

[coder]

If only someone talented from NVIDIA would move over to AMD to help with their GPU's. Things would be dandy, however I suspect NVIDIA would have some NASTY contracts that basically prohibit you from working elsewhere or something.

The state of California doesn't allow non-compete agreements. That's why the tech scene is so incestuous, there.

[schmidtbag]

There are these things called creativity and experience, engineers and technical managers are often hired for these properties. Shocking!

It seems you have completely missed my points...
Outside of the text you cropped out, I also stated how Intel doesn't seem to change their products at all based on what these people could bring to the table. I sure as hell don't see any creativity coming from Intel in years. If what you said was true, it seems Intel is taking the ideas of these creative and experienced people with a grain of salt. Again, I'm referring to everyone Intel has hired who used to work for AMD.

[microcode]

It seems you have completely missed my points...
Outside of the text you cropped out, I also stated how Intel doesn't seem to change their products at all based on what these people could bring to the table. I sure as hell don't see any creativity coming from Intel in years. If what you said was true, it seems Intel is taking the ideas of these creative and experienced people with a grain of salt. Again, I'm referring to everyone Intel has hired who used to work for AMD.

Well, I left the whole sentence intact. Do you mean to say that you didn't mean what you said?

If what you said was true, it seems Intel is taking the ideas of these creative and experienced people with a grain of salt.

I meant to say that the idea is probably not to take patented ideas they developed at AMD, but to employ the engineers so that future ideas they have can be used by Intel. This is the purpose of hiring engineers.

It would be cheaper for them to simply license the patents, than to hire the whole engineer. In reality, they'd have to license the patents anyway if they wanted to use them!

[starshipeleven]

It seems a bit suspicious to me how so many of AMD's employees (not just Koduri or Keller) end up at Intel, despite the fact that Intel's products never seem to change due to their presence. Gets me to think that Intel doesn't really want them as engineers, they just want these people contractually obligated to stop supporting AMD. Intel knows they can't really do anything directly to AMD without igniting a lawsuit, but there's nothing illegal about offering better pay to AMD's employees. As long as AMD doesn't have good engineers, they can't threaten Intel. Intel is overdue for a new architecture, but it wouldn't surprise me if they know that silicon transistors are pretty much at their limit. Intel might not have a worthy replacement, so they just need to slow down AMD's progress until they can figure out what to do.

This would make so much more sense if Intel didn't dominate the market with like 99% of laptops and 90% of desktops.

It's more likely that Intel is just scrambling to actually produce something that people sitting on their 5+ years old Intel PCs will feel interested to upgrade to, and just failing to get decent results.

[schmidtbag]

Well, I left the whole sentence intact. Do you mean to say that you didn't mean what you said?

No... I meant I said more than just that 1 sentence. My sentences were related to each other. Just because a sentence ends, that doesn't mean the context is completely dropped.

I meant to say that the idea is probably not to take patented ideas they developed at AMD, but to employ the engineers so that future ideas they have can be used by Intel. This is the purpose of hiring engineers.

I never suggested they were trying to take patented ideas, not that Intel needs to do that anyway. For the most part, any features that one architecture uses, the competitors manage to figure out within a year anyway. Whether that be instruction sets, pipeline length, cache design, bus communications, etc, Intel and AMD have found ways to replicate what the other was doing and I'm pretty sure they didn't ask for the designs. This even applies to GPUs.
So yes, it is reasonable to expect that Intel hired these engineers for future ideas, but don't you find it a little oddly coincidental that they keep hiring people who worked at AMD? This is kinda the point of my post. Sure, these people have a lot of experience, and AMD has good designs. But AMD's designs aren't revolutionary compared to Intel's offerings, despite being very different. Intel doesn't need to side-step their performance, they need a breakthrough.

In reality, they'd have to license the patents anyway if they wanted to use them!

Yup, probably.

[starshipeleven]

This would make so much more sense if Intel didn't dominate the market with like 99% of laptops and 90% of desktops.

It's more likely that Intel is just scrambling to actually produce something that people sitting on their 5+ years old Intel PCs will feel interested to upgrade to, and just failing to get decent results.

Aaand my superior abilities to see the future are confirmed yet again (surprise surprise). https://www.extremetech.com/computin...ller-to-fix-it

Apparently Intel's 10nm process is delayed to some point in 2019 (after it was initially expected by 2017, and then delayed to 2018).

While others are still going to newer process, like 7nm or whatever.

[schmidtbag]

Aaand my superior abilities to see the future are confirmed yet again (surprise surprise). https://www.extremetech.com/computin...ller-to-fix-it

Apparently Intel's 10nm process is delayed to some point in 2019 (after it was initially expected by 2017, and then delayed to 2018).

While others are still going to newer process, like 7nm or whatever.

Doesn't that problem require a different set of expertise than what people like Keller has to offer? I'm not implying that is the case, I'm actually wondering.

[starshipeleven]

Doesn't that problem require a different set of expertise than what people like Keller has to offer? I'm not implying that is the case, I'm actually wondering.

Yes, that's manufacturing process, which is not Keller's job.

But if they want to keep competing (mostly against their own older products at this point) and they can't just count on shrinking the process node to work as fast/reliably as it worked until now....

they can only try to improve the architecture. And since they can't afford to be dicking around in this situation they hired Keller for the job.

This was going to happen sooner or later as the limits of silicon are reached. Shrinking process node has started to go into "diminishing returns" land years ago already.