HW news - Nvidia & AMD fight for 5NM, intel alder lake, RTX ampere, Ryzen Pro 4000
video description
Date: 2020-05-10
Related videos
Comments and reviews: 10
Zazethe
A silicon atom has a Van der Waals radius of 210 pm. To make a transistor you need more than 1 atom, at least 3 i suppose if you are optimistic, for gate, drain and source. As atoms cannot enter each others radius, 3 times 210 is a best case 620pm or about 0.62nm. However, because you need to dope the silicon into Positive and Negative charges, you need to combine it with other atoms. So you can pretty sure smaller than 1-2nm transistors will never be created. And even that would require immense the precision of basically placing single atoms precisely on a substrate. So after that it will be up to designs and architectures, Quantum computing perhaps, or Terrahertz capable switching materials(but not smaller), superconductors perhaps for less heat.
reply
A silicon atom has a Van der Waals radius of 210 pm. To make a transistor you need more than 1 atom, at least 3 i suppose if you are optimistic, for gate, drain and source. As atoms cannot enter each others radius, 3 times 210 is a best case 620pm or about 0.62nm. However, because you need to dope the silicon into Positive and Negative charges, you need to combine it with other atoms. So you can pretty sure smaller than 1-2nm transistors will never be created. And even that would require immense the precision of basically placing single atoms precisely on a substrate. So after that it will be up to designs and architectures, Quantum computing perhaps, or Terrahertz capable switching materials(but not smaller), superconductors perhaps for less heat.
reply
Patrick
I don't imagine that your response on the picometer comments will satisfy people, since they're responding to a comment you made about running out of numbers after nanometers and not to the concepts in the paper. You then responding as if they were responding to the concepts of the paper really makes no sense. I wasn't one of the people that commented, but did notice your unfortunate phrasing and thought it _would_ engender a lot of comments. I think this recap of the same information as last time with just leaving out that one comment and acting as if people didn't understand and saying people didn't read the paper (I'm sure some did), is just throwing oil on the flames.
reply
I don't imagine that your response on the picometer comments will satisfy people, since they're responding to a comment you made about running out of numbers after nanometers and not to the concepts in the paper. You then responding as if they were responding to the concepts of the paper really makes no sense. I wasn't one of the people that commented, but did notice your unfortunate phrasing and thought it _would_ engender a lot of comments. I think this recap of the same information as last time with just leaving out that one comment and acting as if people didn't understand and saying people didn't read the paper (I'm sure some did), is just throwing oil on the flames.
reply
Carvin0
There is a natural lower limit to semiconductor technology dimensions which is 1 (one) Angstrom. 1 Angstrom is one tenth of a nanometer; a unit commonly used in crystallography. So a 0.5 nanometer technology is a 5 Angstrom technology. The Angstrom is the natural unit to measure atom sizes and spacing. For example, the nearest neighbor silicon to silicon atom distance is 2.35 Angstroms. Nothing in the semiconductor technology will be smaller than a few Anstroms (that is, a few atom spacings). Picometers, femtometers, etc. are in the realm of nuclear and particle physics, not semiconductor physics. 500 picometers sounds silly, but 5 Angstroms sounds cool, doncha think?
reply
There is a natural lower limit to semiconductor technology dimensions which is 1 (one) Angstrom. 1 Angstrom is one tenth of a nanometer; a unit commonly used in crystallography. So a 0.5 nanometer technology is a 5 Angstrom technology. The Angstrom is the natural unit to measure atom sizes and spacing. For example, the nearest neighbor silicon to silicon atom distance is 2.35 Angstroms. Nothing in the semiconductor technology will be smaller than a few Anstroms (that is, a few atom spacings). Picometers, femtometers, etc. are in the realm of nuclear and particle physics, not semiconductor physics. 500 picometers sounds silly, but 5 Angstroms sounds cool, doncha think?
reply
Fladder
I am truly starting to worry about 5nm technology and 5ghz tech and smaller. I forsee quantum vulnarable security isseu's with smaller 5nm chips, I think they might be dangerous to external pinpointed microwave multi-pointsource intermodulation radio emittors you can probably read and write / manipulate these chip externally. By building this small you will inherently create microwave receivers/radio's with quantum propertiess. With a specific microwave burst they can deactive and activate all chips at once. We are at grave danger.
reply
I am truly starting to worry about 5nm technology and 5ghz tech and smaller. I forsee quantum vulnarable security isseu's with smaller 5nm chips, I think they might be dangerous to external pinpointed microwave multi-pointsource intermodulation radio emittors you can probably read and write / manipulate these chip externally. By building this small you will inherently create microwave receivers/radio's with quantum propertiess. With a specific microwave burst they can deactive and activate all chips at once. We are at grave danger.
reply
Ivan
Regarding the case modding contest, it is important to note that they are looking specifically for mod of Alienware Aurora. If you read all rules carefully you will see how constrained participants are... this is nothing more but marketing propaganda of a pre-build dell computer that plagues the PC building community and should not exist. As a maker, pc enthusiast and engineer myself I was really eager to participate in the contest until I read the small print. How unfortunate TPR :/
reply
Regarding the case modding contest, it is important to note that they are looking specifically for mod of Alienware Aurora. If you read all rules carefully you will see how constrained participants are... this is nothing more but marketing propaganda of a pre-build dell computer that plagues the PC building community and should not exist. As a maker, pc enthusiast and engineer myself I was really eager to participate in the contest until I read the small print. How unfortunate TPR :/
reply
Filip
You know, i'm pretty sure anyone interested in process node size is quite aware of pikometers being smaller than nanometers. I also recall exactly 0 instances of someone having a problem with previously used 0,xx um (micrometer, micron) naming scheme. In fact, I believe some (:wink:, :wink:) IC manufacturing company got named after exactly that. Still not the point of course, but my inner social media warrior felt the compelling need to address that particular line of reasoning.
reply
You know, i'm pretty sure anyone interested in process node size is quite aware of pikometers being smaller than nanometers. I also recall exactly 0 instances of someone having a problem with previously used 0,xx um (micrometer, micron) naming scheme. In fact, I believe some (:wink:, :wink:) IC manufacturing company got named after exactly that. Still not the point of course, but my inner social media warrior felt the compelling need to address that particular line of reasoning.
reply
Blahorga
I see a simple solution to the nm problem. We simply define the process by the inverse potrzebie of whatever the process developer feels like, and turn the responsibility for calculating this over to the marketing departments. And now we end up with a number that has no bearing what so ever on anything real and is totally useless, and by making the marketing departments responsible for this number they will be streamlined for marketing purposes. Problem solved...
reply
I see a simple solution to the nm problem. We simply define the process by the inverse potrzebie of whatever the process developer feels like, and turn the responsibility for calculating this over to the marketing departments. And now we end up with a number that has no bearing what so ever on anything real and is totally useless, and by making the marketing departments responsible for this number they will be streamlined for marketing purposes. Problem solved...
reply
Jimmy
To allow their marketing to state a value and unit 'for public consumption' without that being standardised across the globe is pretty shocking, though in the end, a CPU's attached unit size doesn't really matter for the educated public - they'll get all the purchasing info needed via a comprehensive look at benchmarks and reviews, using the old and standardised, price-to-performance metric (despite what Ngreedia is trying to do).
reply
To allow their marketing to state a value and unit 'for public consumption' without that being standardised across the globe is pretty shocking, though in the end, a CPU's attached unit size doesn't really matter for the educated public - they'll get all the purchasing info needed via a comprehensive look at benchmarks and reviews, using the old and standardised, price-to-performance metric (despite what Ngreedia is trying to do).
reply
manmeet
I honestly want to be the devil's advocate here, Intel claims that it's 10 nm is equivalent to tsmc's 7 nm but if benchmarks are to be followed, the same can be said for Intel's 14 nm and tsmc's chip being used in amd processor. I just hope with all this time intel has had time to develop a well designed architecture for their 10 nm chips. I really want to see what they come up with in 2021 to fight off their competitors.
reply
I honestly want to be the devil's advocate here, Intel claims that it's 10 nm is equivalent to tsmc's 7 nm but if benchmarks are to be followed, the same can be said for Intel's 14 nm and tsmc's chip being used in amd processor. I just hope with all this time intel has had time to develop a well designed architecture for their 10 nm chips. I really want to see what they come up with in 2021 to fight off their competitors.
reply
epiccollision
Why are we explaining nanometers to consumers ? A pc/laptop consumer does not care or is ignorant....cpu consumers might care, but that numbers also doesn t tell you anything, for a while amd and intel were on the same node and they were not equal in performance. Are we going to start quizzing people? Ignorance to a point might be a good thing, do you need to understand how ammonia is made it use it effectively?..no
reply
Why are we explaining nanometers to consumers ? A pc/laptop consumer does not care or is ignorant....cpu consumers might care, but that numbers also doesn t tell you anything, for a while amd and intel were on the same node and they were not equal in performance. Are we going to start quizzing people? Ignorance to a point might be a good thing, do you need to understand how ammonia is made it use it effectively?..no
reply
Add a review, comment
Other channel videos
