AMD 7nm lead over Intel

SpyderTracks

We love you Ukraine
Interesting. I was reading this only the other day too...

Yeah, it's so obvious it was zero to do with a relationship, Intel Management have been renowned for inter-office relationships over the years. His dismissal is purely down to letting AMD get this far ahead.
 

ubuysa

The BSOD Doctor
Yeah, it's so obvious it was zero to do with a relationship, Intel Management have been renowned for inter-office relationships over the years. His dismissal is purely down to letting AMD get this far ahead.

Indeed. This paragraph was key for me...

During Krzanich’s short time as CEO, Intel underwent major restructuring and bailed on the smartphone market. The company cancelled its Atom processors that were supposed to compete with Qualcomm’s Snapdragon line, and even signed a deal to produce ARM-based chips in Intel factories. Intel is now focusing on the GPU business, and the future of artificial intelligence.

The term 'ship without a rudder' springs to mind. Innovation and diversification are vital in this industry, but not at the expense of your core business....
 

Tony1044

Prolific Poster
When that story broke about him leaving Intel, the first comment I saw was "He clearly took Intel Inside too literally" which tickled me.

7nm really is something. I wonder what their failure rate is.

Putting my old electronics engineer hat on, if you can shrink the fabricration, it has three major advantages: lower voltage which means lower power, the ability for more transistors per chip but also (and this one always amazed me) it takes less time to shift the electrons from your 0 to 1 and back which means faster processing time.

And we run CPU cores at voltages that are so low that back in the day wouldn't even have registered. We wouldn't have been able to power anything from them.

It's been a very long time since I did any serious hardware work but for those of you that're interested, the next time you open up your PC, follow a couple of the tracks on the PCB (motherboard etc) and you'll often see that they take sudden, unexplained turns. This is actually also by design - it means that all the tracks on the high performance buses (such as from memory and CPU) are exactly the same length...so all the signals arrive at the same time.

And to think the first Intel processor (the 4004) which we had to study as part of our processor design courses could be printed out on a few A3 sheets and be [just about] readable. :)
 

ubuysa

The BSOD Doctor
Am I mistaken, or did I read somewhere a short while ago that the wavelength of the light used in the photolithographic 'printing' process is becoming a limiting factor? I just looked up a 7nm wavelength and it's beyond ultraviolet light and into x-rays. How do they do it?
 

Tony1044

Prolific Poster
3nm????? That's nuts!!!

It's mind-blowing, isn't it?

I remember about 20 or so years ago reading in the electronics trades that we'd hit the limit for what lithography could do in terms of transistor density. If memory serves me, IBM labs were working on a technique with tunneling electron microscopy that could physically shrink the components within the die.

And more recently, Intel created a three dimensional transistor.

We've come an awfully long way since this bad boy: https://upload.wikimedia.org/wikipedia/commons/b/bf/Replica-of-first-transistor.jpg
 
Top