Gadgetory

everyone welcome back to another hardware news recap lots of news this week sunny Cove of course dominated the news cycle but also some interesting discussion from MIT on a new 2.5 nanometer transistor 3d transistor approach using fin FET design but the interesting thing is that it's a new manufacturing technique involving basically an ion cannon so we're talking about that for this week very interesting stuff and the future of manufacturing process for transistors and silicon although a very distant one at that rumors on Andy Vega 2 are in the news as well as rumors on the RT X 2060 becoming less of a rumor every day and then also some more brain drain going on with AMD as more talent goes to Intel before that this video is brought to you by MSI's meg xiii ninety ace motherboard which primarily offers high frequency memory support for forty five hundred megahertz kits and tight timings tuning but also comes equipped with a twelve phase V core vrm at ready for the ninety nine hundred k a pre-installed IO shield and an infinity mirror for RGB LEDs atop the left heatsink the RGB LED is synchronized with Emma's eyes mystic light to support RGB and rainbow LED strips using the infinity mirror to cascade lighting learn more at the link in the description below quick note before diving in if you didn't see the disappointment build that we posted you should absolutely go check it out it's it's the most cinematic thing we've ever made but it also has some real content in there after the cinematic opening maybe the editors can put in a clip from that build now or something so you can check it out and with that we also have the disappointment build t-shirt which has effectively concert tour date listings of all of the most dissapointing elements of 2018 s Hardware cycle including things like spectra and meltdown crypto mining depleting all the GPUs available Ram prices the r-tx launch just by it and the verge pc build which we've listed as worst PC build so you can check that out on stored I Kara's Nexus dawn that if you want to pick one up it is a great way to support us and to commemorate the disappointing moments of 2018 by wearing a constant reminder of them so moving on engineers construct a 2.5 nanometer transistor this is some really really interesting stuff to talk about researchers from MIT collaborating with the University of Colorado which is responsible for working on a new fabrication technique a modified thermal atomic level a qing working together with MIT they have managed to build a 3d transistor coming in at 2.5 nanometers using a fabrication process that works at an atomic level so this is of particular interest because we see Intel working towards 10 nanometer and the and it's fabrication partners working towards 7 quick note on that 7 versus 10 isn't as easy as 7 is 3 lower than 10 and therefore is 3 better there are there are other issues to consider like density you can have a 10 nanometer process that is better than a 7 and reprocess or a 40 nanometer process that's better than a 10 nanometer process depending on the architecture the density and things like that so it's more than just the number in nanometers it's more than a numerical difference to consider that said though it is important to continue to shrink things so that the amount of transistors in a given area can increase as well and density goes up and this is a topic we discussed probably in about 2010 with a former Intel employee who worked on the tick-tock cadence when it was first created and he had noted to me in that interview that there was at the time a clear line down to 8 nanometers he thought and after that we would have to use ion cannons by his description of them to me and that turns out to be looking pretty accurate actually so what they've got here in order to do this new approach the researchers use 3d transistors they use FinFETs which many of you are likely familiar with and they stand vertically like fins hence the name fin FET it's kind of got a fin on the end helps with leakage and things like that and we've discussed all that in the past as well there are the fins are thousands of times thinner than a human strand of hair and to create these smaller fin vets the team is using a modified thermal al e or atomic level atching technique and the result is FinFETs as narrow as 2.5 nanometers this theoretically makes them more efficient than commercial counterparts from what the MIT report said MIT is website does a great job at describing the process writing the following microfabrication involves deposition growing film on a substrate and etching or engraving patterns on the surface to form transistors the substrate surface gets exposed to light through photo masks with the shape and structure of the transistor all material exposed to light can be etched away with chemicals while material hidden behind the photo mask remains the state-of-the-art techniques for micro fabrication are known as atomic layer deposition Ald and atomic layer edging al in Ald two chemicals are deposited onto the substrate surface and reacts with one another in a vacuum reactor to form a film of desired thickness one atomic layer at a time traditional al II techniques use plasma with highly energetic ions that strip away individual atoms on the material surface but these can cause surface damage these methods also expose material to air where oxidization causes additional defects that hinder performance and the researchers here believe that the method could be migrated or integrated into existing fabrication processes and the real-world impact would be in the form of dents or chips higher performance and potentially increased yields so it's a very interesting article we would encourage you to check out the rest of it if you want to learn more about the future of this technology and we'll have it linked in the show notes in the description below if you want to check that out from mit.edu and it is just just kind of set the expectations here this isn't something you should expect in a cpu tomorrow or next year or in two or three years it'll be many many years away but the research has to be done in order to ever integrate into some kind of commercial application so really cool stuff to see Sonique ov we'll just recap this if you've missed our videos on sunny Cove the most pertinent one for you to check out would be our discussion with David Cantor a technical analyst in the industry runs real-world tech and is one of the one of the leading experts for any kind of CPU analysis especially third-party so you should check that out we'll recap some of it here Intel also silently launched the be 365 chipset that was not discussed at the Sun eco event and then they did discuss the Jena Levin and XE graphics not T GPUs but IGP stuff that's still pretty interesting along with a hybrid x86 approach to design Intel's architecture day was a pretty major event they don't really do architecture days that actually in in years from what we can remember here but so it's good it's good to see Intel getting a grip back on what's going on and acknowledging that they need to engage to try and assure everyone that this ten nanometer stall won't happen again and at this event it was densely packed with things like sunny Cove and the new iGPS and eventually a DGP you approach probably but D GPUs were not explicitly discussed at the event aside from a passing note that well they're not here but so if you were waiting for that and for Arctic sound sadly that news did not come to be it's a couple years away though Intel themselves noted a 20/20 window for an unveil of a d GPU when they did their set your graphics free campaign that they did a while ago so for the major announcements here there's the new suncove microarchitecture for the core architecture this is contained within the greater ice Lake product that you've likely heard about in the past but is is technically different by definition of sunny Cove being explicitly the core there's also the Xeon and the core series chips that will bear sunny Cove and later I believe they were willow Cove and golden Cove and further Intel made announcements for the new XE graphics brand and also hybrid x86 CPUs so not related to Intel's architecture day was the be 365 silent chipset released if you didn't already see that one so in top hold the wraps off of sunny Cove here this is something of a clean slate architecture for the company it's been stuck on skylake for years now and stuck on 14 nanometer for years - in tiles credit 14 manager has had significant uplift to a point that it's really pretty impressed moving from 4.2 gigahertz to 5.0 plus depend on if you count overclocking or not but if you don't still 4.2 to 5.0 40 nanometer is pretty darn good but it's not exciting and it has stalled Intel's progress to a point where Andy has been able to catch up in a significant fashion something that up until reisen's launched was really not something that we would expect because the FX series was just not competitive in the ways that Rison is so Intel now has to get moving on this and the new architecture is tied to a new smaller process node Intel has since deviated from tick-tock for years now but they're still pushing and Intel has a massive roadblock with a shrink to 10 nanometers they're stuck with 14 and the sunny Cove is the first of three new architectures in Intel's roadmap that are designed to be portable between nodes and will be Intel's first raw IPC increase since Skylink sunny Cove will debut on the 10 nanometer node and is slated for the first half of 2019 although it isn't clear which chips the architecture will accompany for our part we did speak with representatives at the Intel event it sounds like Mobile's getting an immediate focus from what Raja Kaduri was telling us and desktop will be a bit later but this is all stuff that will be will be coming out over the next year - so interestingly sunny Cove chips will be paired with Gen 11 graphics as well and will be branded as well maybe not publicly but will be ice Lake functionally something that Jeff Kampmann from tech report noted when he took a photo of the heatsink that was labeled icy Lu even though Intel was not officially calling it ice like so accompanying this launch of the jenna levin graphics engine is an increased count of execution units significantly from 24 to 64 with the compute performance seen an increase as well pushing Intel into a higher class of performance - about 1 teraflop now when we saw the demo we have some b-roll of it we'll play it on the screen Intel was showing a Tekken demo where on the left side you have the or at least camera left you have the new gen 11 graphics and then you have the old IGP graphics versions now these demos are interesting but the graphics settings matter a lot to understand if they actually mean anything when we looked at the graphic settings they were 1080p at 85% resolution skelion which is odd to say the least so it's basically 9 18 P which isn't a thing but that would be your functional resolution and the graphics quality settings were high for anti-aliasing and medium for everything else is something we checked in the graphic settings for each device they were set the same so it was a fair comparison it's just an odd choice of resolution but it's an undeniable improvement it's just whether or not that will compete with things like a PS or low on D GPUs that will matter and how much it moves the price target for the low-end CPUs that have those IG P's integrated Intel is also finally moving to implement adaptive sync something to talk about years and years ago it won't require any external hardware and will be compatible with free sync monitors beyond Janna Levin is Intel's new X II graphics brand which will take the company into discrete gp2 territory at some point details are scarce in town mostly teased list more than anything else however it seems that with X II Intel is aiming for scalability hinting that they intend to design GPUs ranging from integrated solutions all the way to mid-range and enthusiast discrete GPU is even targeting data center these designs will all likely be underpinned by the same architecture and Intel also reaffirmed its plan to bring discrete GPUs to market by 2020 lastly Intel debuted a new hybrid x86 processor as they called it and this is built with a new 3d chip stacking process Favreau's which is as they told us I believe Greek for awesome although they haven't really looked too much into that but fibrosis is what they're calling it and this will allow various dyes and silicon to be stacked something about which we spoke with David Cantor previously so this is used in conjunction with emmab or embedded multi die interconnect bridges it'll allow chips to be built with a mix-and-match approach combining CPUs iGPS and different IO elements with a high-speed interconnect Intel demonstrated a working sample for an unnamed customer whereby Intel combined one Sonico of core and four atom cores in a 10 nanometer design coupled with a separate 22 nanometer i/o tip connected through CVS or through silicon via Intel announced that they're planning to roll out an entire product with Favreau's design two chips coming out or entire product line as an aside for that be 365 chipset it's a 22 nanometer variant of the 14 nanometer B 360 you can think of this as similar to the previous news of the h3 10 C chipset which was a specific deviation from the 14 nanometer process so that Intel could free up stab space and have partners produce the new extra low low end chipset but be 365's bit news it uses the KB Lake PCH it's no doubt another move to free up fab space as Intel waits to try and get to its new architectures on the horizon because 40 nanometer is still in short supply and it's beginning to impact sales or at least the pricing in a significant way to the extent that Intel has removed about two million units from the DIY channel and moved them over to OEMs and sis instead so Intel's continuing to struggle to meet 40 nanometer demand but at least there is a target on the horizon for relief new reports of AMD filing a trademark for a new Vega logo suggest that the company isn't done with Vega just yet looking at the logo one could interpret it two ways either Vega 2 or as the Roman numeral 7 VII denoting a seven nanometer shrink depending on how much you want to read into it further fueling the Vega refresh rumour mill is the discovery of a new device ID and the Linux drivers Veronica saw one new Vega 20 ID and six new they get 10 IDs while logos and device IDs can only lead us to speculation it's pretty clear that Andy is planning something with Vega and that it's not completely gone with 64 and 56 so this will presumably exist alongside or just before na'vi and hopefully we'll learn more at CES 2019 we already have mediums booked for just about every manufacturer so make sure you check back in early January for our coverage of that maybe they'll have something there for us to look at also in the rumor mill the RT x 2060 now less and less of a rumor as more reports come out it packs 1920 cuda cores the leaks suggest that this is launching early next year and also asserts that the 2060 will boast in 1920 CUDA cores and the tu 106 GPU silicon the card is also rumored six gigabytes of gddr5 a comparatively the current crop of gtx 960 is run 1280 CUDA cores six gigabytes of gddr5 or gddr5 x after a recent refresh still no word on tensor cores or real-time ray tracing for the 2060 cards but they do have our TX branding and along with many others in the tech space we've speculated that the 2060s might not really have enough power their raw performance to drive real time ray tracing in any meaningful fashion unless you're gonna play it at 720p or something so we'll see we'll see what happens but we're just around the corner from when announcements typically happen so keep an eye on the space as we move towards CES and first quarter 2019 the next news item is an industry news item but it's an important one this is following several hirings that Intel's done from AMD so it's kind of a brain drain going on right now we're a lot of AMD talent in RTG specifically or the Radeon technologies group is moving over to Intel the most notable of these in recent history was Raja Kaduri who was functionally the head of RTG at AMD when he was there and Raja has been at Intel for about a year now along with Raja went Chris Hook one of the marketing leads at AMD and also Ryan shroud from PC perspective since this changed though another Hardware reviewer Damian Triola I believe it's pronounced from hardware dot F R has headed to Intel and Intel poaching more talented AMD RTG members because Damian here although used to specialize in GPU reviews at the hardware our french review website has was since employed by AMD and is now being brought over to Intel as of November 2018 it seems like Damian's jump ship to Intel and is now in a position of tactical marketing again within intel's at gaming and graphics department that's recently grown and intel also recently snatched up again brian shroud formerly a pc perspective another former reviewer moving on to the last news item and video shares slide again to half value now of the recent high and a major investor Softbank wants out so we recently reported on Nvidia's earnings and how they suffered what Jensen Hahn CEO of Nvidia called a self-induced quote crypto hangover as it would seem the Hanover isn't subsiding anytime soon because Nvidia put a lot of eggs in the cryptocurrency basket which failed to really pan out when the bubble burst at least for mining anyway so anybody has been left as a result with excessive amounts of GPS in the channel particularly gtx 1060 s and this is something we discussed previously where 1080 s 10 DTI is leading up to our TX launch we're finally starting to drop in price because it was time for those manufacturers he's got stuck with the bill of all the GPUs to dump their inventory and he got affected as well so they didn't escape the as anybody called it crypto Hannover completely unscathed either but as such and vinius had the temper it's quarter for earnings expectations with this new dose of reality and they've lowered the expectation by more than 1 billion dollars subsequently the stock market shares have started tanking in in price taking a nosedive recently and that nosedive seems to have not found a bottom quite yet NVIDIA shares recently hit roughly half of their peak value from two hundred and eighty nine dollars and 36 cents to $149 at least as of the time of this video worse yet one of Nvidia's biggest shareholders the Softbank group wants out according to a report from bloomberg so Softbank is considering selling sometime next year after Nvidia is less than stellar stock market performance the company originally invested three billion dollars in 2017 and has since profited about three billion dollars they might just be leaving any way to get that to collect on that check of a hundred percent profit not bad so that's it for Hardware news as always you can catch up with with additional coverage by subscribing to the channel go to store documents access net to pick up one of our new disappointment build shirts as is hanging right there and go to patreon.com/scishow cameras axis helps out directly we'll see you all next time