Gadgetory

seven nanometers can you visualize how small that is let's try a strand of human hair is about 100 micrometers thick and that's about 14,000 times wider than seven nanometers in fact with this scale we could measure the thickness of a single DNA strand clocking in at around two and a half nanometers some of the smallest effect of transistors we've ever created we're just a single phosphorous atom across that leaves out the silicon aspect of it but still and that's about 0.2 nanometers in diameter so much for the end of Moore's law so we are literally fabricating processors in the same scale as us as human beings and here's why it is so important not just to clarify seven nanometers does not describe the size of the transistor in a CPU we discussed this in more detail right here the length actually describes the precision of the fabrication how detailed you can get think about it like I don't know the size of a pixel smaller pixels take up less real estate on a screen meaning you can pack more of them into the same amount of space the result is a higher degree of detail more information per frame basically and that's why fabs like TSMC and GlobalFoundries and micron use this notation perhaps above all else the node shrinkage signifies a bump and efficiency and that's what we'll discuss first so in short seven nanometer processors are extremely efficient surprise surprise Apple was the first to push a seven nanometer processor to market dubbed the a12 bionic you might have heard of this before the iPhone 10s boasts nearly seven billion transistors in an 83 square millimeter surface area that is smaller than like pinky nail just ya visualize that by contrast the Xbox one exports a similar number seven billion or so and a package significantly larger roughly 360 square millimeters this is because the Xbox utilizes a 16 nanometer fab so the node shrinkage alone allows us to cram similar amounts of compute power into much smaller packages that is obviously one pearl another trait of a smaller transistor is its lower capacitance defined simply as the amount of electrical energy a component can store now in the case of CPUs transistors act as switches and require threshold voltages to both open and close smaller transistors inherently have lower threshold voltages since the gaps between sources and sinks are smaller right they're smaller transistors and the energy to function effectively it's does lower Advice ignition amount a byproduct of this is frequency meaning the gates themselves can open and close faster if less energy is required to build up in the process its overall this increase in precision allows these processors to save power and operate faster which is a win win for the consumer there are drawbacks including leakage which has been a concern for the past five or six years but we're constantly coming up with innovative ways to solve that problem another benefit of smaller more precise CPU architecture is space obviously remember that 812 Bionic chip we just talked about it's die covers less a square centimeter of area meaning Apple had plenty of additional space to work with it means a bigger battery means more RAM it means a larger GPU more cache it really just depends on what the manufacturer deems necessary nvidia for example took the opportunity with its 12 enemy to process to cram additional compute tensor and RT cores into that massive t you want a to die in the cases of both the 2080 TI and titan RTX we just call that like the Titan R of course there are thermal limitations associated with any electrical work done but assuming you've got a beefy enough cooling system the benefits of a smaller architecture in this case are limitless and allows companies like Nvidia to experiment with innovative ideas and new ways of solving complex problems but with the seven nanometers fabrication comes it's financial and technological challenges another reason why their entrance into the market is such a big deal production cost scale rather exponentially with no shrinkage that's just relying on historical context there and foundries like TSMC currently supplying Apple Qualcomm and Nvidia among others are feeling this strain higher degrees of precision mean more mistakes to put it lightly this is one of the reasons why I expect the lower tiers n2 chips in the case of AMD to be extremely affordable if the yields per core are lower and we expect to see C X's per die and maybe up to 4 CC X's in total then Zen 2 for poor CPU should be an ample supply by default this is why prices for higher core count chips don't scale on a price per core basis in a linear fashion if the yields for an 8 core variant are 40% let's say and the yields for a 4 core variant are 80% then the 8 core chip should cost well over twice the price on a consumer side since these failure rates cut in to AMD zone margins Intel mind you is dealing with a very similar failure rate with their in-house fabs which is a reason why we keep seeing 14 animator refreshes from them not the only reason but it's a big one so I hope this short little video at least shed a little bit of light into some of the reasons why the 7 nanometre leaf is first of all so difficult and so important in the industry I expect the year 2020 will be a sweet time for PC gamers and enthusiasts and it'll be a great time to finally just appreciate competition in the space I think we need more entrants but I'll take whatever I can get at what are you guys thinking this was a short little video I could have gone into more detail but just bringing up the whole 7 any of your conversation is good because uh we're looking forward to a lot in the next two or three years like the video you know what to do dislike it for the clean tops of feeling or if you hate everything about life you click that red subscribe button if you haven't already now I'll catch you in the next one this is science studio thanks for learning with us