Gadgetory

hey everyone welcome back to another episode of ask GN this time I'm joined by popular demand demand by our senior Andy correspondent snowflake and also sometimes CEO of gamers Nexus we have a lot of good questions for this one as noted recently we're shooting 2 episodes per week now plus the bonus patreon episode you know patreon.com slash gamers and access to join us for that and otherwise we'll be uploading both the main ones to the main channel she's left me already before that this video is brought to you by the EVGA X 299 dark motherboard the one that we recently used to take a top 5 world record for fire strike overclocking results the EVGA x 299 dark board is one of the best we've used presently in this current generation and it has a coupon code of gamers nexus for a savings of $75 off for us customers expires on May 31st if you've been interested in the X 299 dark with its actual vrm heatsink and actual vrm cooling capabilities that was a good time to grab it use code gamers next as that check out click the link in the description below it's the first question for this week for it well for this episode of this week out of the three we're shooting today is from Looney juice who says besides a Nvidia's quarterly report how is the DIY enthusiast market doing all things considered we got past the doom and gloom articles and then into what I guess was a bit of an enthusiast boom we're still seeing new skiers of various types of products but with so much hammering the market is there even a way to get a bead on how CPC components and systems have been affected you talked about multiple SKUs entering the market isn't there like a new 1053 gigabyte model like what is going on I understand that Pascal is still doing well and understand that there's not a new generation so clearly there's some demand I guess but really a 1053 gigabyte we really need that we've got how many 10 60s and how a 210 30 is like what is this product segmentation it's super out of that I find it super annoying that those types of SKUs are entering the market but and to bring the analyst back for this question so overall things are for the enthusiast market things are doing pretty well it's actually one of the only growing segments of the PC industry right now pre-built not doing so well spree builds are actually not gonna be able to answer this from an AMD perspective now I don't know why it's so difficult for her uh the she's gonna fire me later the pre bolts are the definitely the most suffering part of the market there's been shrinkage year-over-year and anything from like Dell HP those types of people and that's because think of like your average non-gaming non enthusiasts non workstation consumer what do they need a desktop for you're probably buying a laptop or something like that or even a glorified phone in some cases I guess I'd be a phone glorified as a computer in some cases so yeah I mean if you look at numbers from most the analysts including John petit research although I know they're not the most highly regarded but overall the PC segment is not doing great but the enthusiast segments doing really well and enthusiast hardware it's driving sales that's why it's kind of disheartening to see the memory price issues and the GPU price issues because it drives down adoption of enthusiast pcs and makes everything look bad which is why companies like Nvidia and AMD are still so invested in PC gaming because they know that's what's driving growth in any kind of desktop sector so definitely over all the enthusiasts is doing just fine like I answered a few ass gems ago I'm a little concerned actually no it was that the Paks panel I was on a panel at PAX East with bit wit Kyle so in that panel I answered a question about is are we in dark times of PC space and I think yeah I mean from a memory price perspective from GPU prices it's definitely a bit of a problem right now for adoption because I don't know if memory costs half the price of your computer to build then why are you gonna build one so that's certainly concerning but I think we'll recover there with time the enthusiam ensuing pretty well overall though is just that one aspect is is a bit troubling I think we'll get through just fine though oh and also quick sidenote again you're right about the scuse hammer in the market it's a bit ridiculous in some aspects and clearly taking advantage of the current situation which I don't blame any company for that's kind of what they do but now there's kind of a limit to what's reasonable next one ten fair and Zoo says wouldn't it be smart to use high-performance top down cooler as an octave and be quiet that would also increase the airflow on the board and would result in longer component life I'm not mistaken it also cool down the VR arms and help with OC on cheaper boards really depends so in an IT Xbox or something yeah downdraft is good because it will get air over the VRMs and even if that air is warm it's still cooler than the VRMs it's still airflow it still moves air across the service and helps dissipate our air is awful at thermal concerns a thermal conductivity terrible interface to use but obviously far better than standing air which is what you get with a CLC that's for example mounted to the front so this is actually some data we never published because we only got the testing kind of half done it was right before Christmas I think so you know it gave people time off and stuff like that it's never quite finished it but one thing we learned is that vrm thermals if you measure them properly with say a tower cool or even like this piece of garbage to my left except without the liquid cooler attached to it tower cooler like that even that will get some airflow over the vrm is even if the fans not directly pointed at them it's just because there's air moving by nature of getting sucked into the fan and that performs well enough that compared to like a front-mounted closed-loop liquid cooler where you've got it mounted like this this is the front of the case this is the CPU this is the worst possible configuration you could have if there's no rear exhaust fan in the case because in that situation you have all this intake in the very front the CPU is great it's gonna be really cool probably the coolest it possibly can be within reason and the VR abs however unless there's a fan above or behind them will get pretty hot and that's partly because you're pushing warm air into the case from the CPU so case ambience higher from the perspective the VRMs there's no airflow over those vrm components and also the cpu is cooler so you can push your overclock potentially higher which means that you're now pushing the vrm is harder because if your CP is cooler you don't really necessarily know that the vrm is getting hot and obviously we can all assume that but unless you have measurements somewhere there's no good way to know what temperature they actually are short of putting a thermocouple on it or using a board that has a sensor and to be honest a lot of people don't even know VR on temperatures the thing you need to worry about I certainly didn't for a number of years when I was a new PC builder so I don't blame anyone who doesn't and the point is if you're just looking at CPU temperature because that's the most logical thing to be concerned with an overclocking then yeah having a really cool CPU means that you could end up pushing the VR arms way harder because you don't know that they're a problem so downdraft coolers certainly eliminate a lot of the VR I'm cooling concerns but you're gonna be more limited on the CPU cooling side so if it's not an ITX case really the sort of best thing to do is make sure your TP is cool to the extent that you need it to be and then put a fan somewhere in the back like either in the rear and hopefully it pulls air across the prm's or in the top isn't really better problem with top is VT top front it'll pull air out of the case before it hits a tower cooler on the cpu in which case you actually end up with a warmer CPU because you're exhausting the cool area from intake before it ever gets there so you don't position a top fan like exhaust too close to the front of the case top intakes probably the best you could possibly do it's just that then you obviously have some dust concerns so depending how you feel about that dumps collecting on the top of the case more likely to get inside the case that that might be kind of a no-go for some people so yeah downdraft is definitely helpful in that capacity that you're talking about it's just that it is naturally limited and how well it can cool the CPU itself because I mean it's just it's less surface area there's there's fewer fins they're not as tall the fan is probably smaller and it might be weaker as well because a lot of the time draft coolers really for like either a box cooler or ITX cases and in that capacity they don't do well on a full tower where you might be pushing the overclock harder so the answer is it depends and it depends on a lot of things but you know would it be smart to use one I mean in some applications yes like in an ITX box especially where you don't have access to a lot of air and you especially don't have access to a lot of air near the vrm components because you're really just trying to get some air into the box in general because it's so small all you need is something and then the CPU cooler if it's down drives will take care of pretty much everything around it including you in the backside of the GPU you get rid of some of that radiative heat although the CPU is going to be sucking it in so the CPU will be warmer but the CPUs are also more tolerant of high temperatures than than a GPU but also VRMs are pretty tolerant of high temperatures as well more so so you had one last question here that you had was about the life of the components you're not mistaken you said would result in longer life I'm not mistaken but we're talking about let's talk about the capacitors those are typically rated for say 5000 hours of use at 105 degrees Celsius some of them are as bad as 2,000 are 2.5 two and a half thousand hours of use at 105 C or some or two and a half thousand at eighty degrees Celsius or 85 so it's either a 80 or 85 but that's obviously once you get into the 80s for rating the capacitors and fewer than 5,000 hours absolutely longer capacitor life is something you should be thinking of and a downdraft cooler would probably aid in that but if you're looking at something like 105 C at 5,000 hours as long as the capacitors are reasonably cooled and it's pretty hard to get a cab to 105 C as long as the unless you have like we did it with a front-mounted CLC no exhaust and no other intake we were able to achieve that but just you know don't be an idiot like we were intentionally doing for purposes of test and you'll be fine so yeah definitely every 10 degree Celsius increase on capacitor is kind of like an almost exponential and how much it decreases your life the capacitor itself MOSFETs take a whole lot of heat talking 120 550 C chokes inductors same thing can also take hundred twenty-five hundred fifty C depending on what you're using so caps the only thing to really be worried about and those hopefully aren't too bad demand which motherboard you choose but we've been able to get them hot anyway yeah down draft coolers are good in situations where it's a concern but if you have airflow in general in that part of the case you'll be fine next question especially if you need a cooler CPU which is typically as doable downdrafts photonics photonics says asked us on discord slapping it in here for this for that dank exposure would increasing the die size of a CPU increased cooling capability by spreading out the heat more for example in James $0.02 recent video he stacks for 480 radiators and achieves about ambient temperatures with his GPU while the CPU and on deleted was still in the 70s I remember hearing in the past with a nasty n that GB is run cooler dude a bigger die size and so less concentration of heat or something like that so though or something like that part is right yeah in the past we've definitely talked about how I don't I think I was probably talking about a comparison between two similar GPUs where the die size was significantly different and in that situation yeah a larger die might help but here's the thing there are a lot of variables here and transistor density impacts this a lot so if all things are equal if we're talking about like basically the same component but a larger die and density I mean like all the the transistor count all that stuff is equal so basically have the same component blown up then the larger one will dissipate heat better it's got a larger surface area to spread it and and the cooler won't have to deal with as high of a concentration of heat in a small area so it's almost like it kind of depends on the amount of power you're generating per square millimeter almost is a good way to think about it we're the more power if you have like a 250 watt component and it's a really small dive which is a large die and all other factors are equal then a larger die we'll deal with less limitations and how the heats being dissipated that said there are a lot of those scenarios realistically in the real world a smaller die has different components and fewer transistors may be less active areas of the die for cash or is if you look at like a block diagram of a CPU or GPU you'll see the different areas on the die or it might be CC axes might be cores like a cache and depend what components have been removed on the smaller dies to get that smaller die size it's not really a linear comparison to just straights a bigger die cooler but bigger die cooler in the instance where bigger dives poorer die are equal for the most part as far as JS JS content you're talking about so if he had I didn't watch it but I I'm sure it's good though I'd like his videos when he does stuff like that if he had a CPU in the seventies and a GPU at a meet you can't really compare them they're different architectures they are the different components they have different active diarrhea's they have different process maybe by different companies consume different amounts of power so can't really linearly compare them especially when considering one's got interfaces like Tim or solder to go through as well in addition to an IHS so way different thermal characteristics between the two I wouldn't compare those numbers because it's not really comparable but I had a note here from VSG of thermal bench as well vs she's a brilliant guy with regard to anything to do with thermals and testing and if you don't know who he is he writes for techpowerup as well sometimes so I talked to him about this question as well and he pointed out that as we enter sub 10 nanometer EUV or extreme ultraviolet lithography for making components for making silicon precision of the masking used for the gates will also impact heat dissipation but otherwise all the other stuff I said all the variables of what is on that die makes the biggest impact to answer your question very simply yes a bigger dies cooler if the two are very similar and that's been the case sometimes when we toss stuff also a multi dye approach is really interesting to you like what Andy's done with Rison and what looks like it'll happen with navi and future MCM a multi-chip module and video components that's that's got its own thermal characteristics because now you have multiple devices with an interconnect and so the heat may be kind of like hot spotted in different areas as each component processes whatever it's doing it's very interesting though hopefully that helps a little bit I don't know if I I think I directly answered that kind of next question sleepy time sleepy time it says wonder how many weeks that I wonder how many weeks that garbage AIO will sit on the bench and pieces this person is referring to the m22 we took apart clearly it's not on the table anymore however allow me to introduce you to this this was a bit of a nightmare to assemble on the stream for a number of reasons so I want to bring it up we have a full content piece coming that like compresses basically three hours of Maxine's youtubes and testing some fans into something that makes more sense so if you did catch the stream though the I found out after the stream it was still kind of leaking overnight so tightened everything all the fittings are all tight the tubes are clamped on with two zip ties they didn't include anything else with that but it was still leaking and it was not leaking out of the tubes the fittings are connecting anything that is kind of external to the cooler so we let it sit overnight I drew kind of a water level here came back the next day and it dropped like another I don't know half inch on this marker and looking at it closely there are actually four screws on top of this thing up there so there's four screws here and two of them we're not really tight at all and to it the other two were maybe about one full turn away from being what I would consider acceptable tightness or torque so if you caught the stream this is just a teaser to what we're doing putting together after it's all tested and everything but basically a lot of the leakage problems from the fittings kind of being a pain in the ass and some of my own mistakes I made I certainly was not flawless in that stream but it was fine I liked it it was a lot of fun actually but some of the problems outside of external to user error were these four screws is not being tight enough and it just it sort of leaked leeches out of the cap on the top and proceeds to drip down the fins which is a real nightmare to clean up I basically just blasted it with an air compressor because it's the otherwise you can't get in there so to answer your question that that cool that you're referring to is gone I mean it's in the box and probably turned into like art or something but now we have this one and you can look forward to this I still need to do some testing on it I realized a lot of things during like Aptus I mean being on stream is kind of a lot of pressure because there's a lot of stuff that's going on we had an issue where the stream went out we didn't know why I had to restart OBS and that fixed it but a lot of pressure so I forget stuff which is why we don't stream too often because it is actually a lot harder to complete the same objective but it's way more fun I love interacting with viewers we're planning to do more streams in the future so be sure to catch them but it's honestly it's mostly just like three hours of me interacting with chat for the most part so not too exciting to watch out to the fact that said my point is there it's a much different environment than a pre film thing so I forget stuff and I forgot I really wanted to connect to quick disconnects which I'm gonna do after this video actually so all we that would have solved a lot of my problems with like stupid stuff like the order I did the cap and the the pump cap but reservoir Peck cab so yeah we're gonna stick these in there and then expand into radiators for some additional tabs because people really wanted to see that anyway that's cooler on the bench now and that's some behind the scenes as to what we found after the stream was over I had time to look at it more closely let it run overnight some leakage from manufacturing stuff some leakage from fittings kind of sucking the ones that they shipped with it some leakage from me screwing stuff up so it was it was fun though we'll do more of these in the future I promise next one final Spartan says do you think CPUs will ever go back to direct eye cooling I feel like significant performance could be gained if direct eye we're less risky probably there's two things to consider here direct eye is more dangerous in that the now you have in tow names you have to warranty that component probably if the user installs a CPU cooler that is maybe out of spec like a partner just a partner like one of the major cooler manufacturers didn't do the torque calculation correctly or something or the user say tightens one side down harder than the other and it cracks the die that's why it stops because with direct die you do have risk of cracking the die even when there's like a retention kit and everything in there you can still do it so I don't think we're going back to direct die for the manufacturers now also consider outside of potential concern where you have a lot more RMA components because people are cracking the die on the silicon or scratching off a small SMD or something like that outside of those things keep in mind that the better they make their products out of the box the less they can segment their products the less they can sell you on the next five percent increase in performance so certainly there's some some kind of you know maybe planned obsolescence there but in this particular instance I can forgive it for both companies because I mean I wouldn't trust 90% of users with a pre dealer at CPU anyway so and the other 10% being all the people watching this channel because if you think about people who are maybe just getting into it kind of scary they just give them a bare CPU and then they buy whatever cooler maybe is not perfectly compatible you crack the die and have very expensive parts RMA shubo says why is directx12 underutilized it's been so long since this new API is launched but games aren't using and ones which are not using it they're not taking any advantage of it why is that I answered this not too long ago in a nasty end I don't remember which one it is I'm sorry but we had a really good response from a developer who writes for Vulcan and directx12 engines and he basically was talking about how difficult it is just I don't know maybe search direct x12 on the channel look for nasty and recently in the last couple of months but the short of it is directx12 in vulcan are actually really hard to write for because it's with DirectX 11 you base and OpenGL basically all of the optimization all the shader optimization the way that the code is processed the way the different cores are you or the different functions of the components GPUs are utilized that's basically handed off to am the Nvidia and they build it in their driver team their driver teams experts on this topic this is all they do is optimize an engineer for game software so that it runs well on their particular video card and they're really good at it and suddenly with DirectX 12 in Vulcan you're taking all that responsibility and you're dumping it on the head of a game developer that's probably really limited on time already they have to focus on getting the game out the door optimizations not really a big concern and so a lot of the time what you'll and also it takes years to get that kind of experience also what you see is an issue where DirectX 12 is sometimes just stuck well not as much lately but used to be just stuck on a game for purposes of marketing today it has dx12 not only a Microsoft thing to push Windows 10 and the thing with that is a lot of those games were just wrapped so it was a direct x12 wrapper it was still DirectX 11 but it was basically imagine it having just like you have like the game and DirectX 11 and then you have a circle around that that says DirectX 12 and all it is handles the call but it's not actually DirectX 12 it's not leveraging any of the resources or advantages of it so you end up in a scenario where actually a wrapper for DX 12 has some overhead slightly worse than the X 11 that's why in a lot of the original comparisons I think battlefield is one of them and total war even a lot of the original comparisons the X 11 and 12 you had an issue where DF 12 was worse and if not an average framerate almost always in frame times where you'd go from say I don't know average frame time or well yeah I retrained time like 16 milliseconds to occasional spikes like 300 milliseconds which is awful that's a third of a second you're staying at the same frame so yeah that was part of the problem was I think turned a lot of people off because the stuff was just wrapped and it said the Exile but it wasn't the other one is a soup hard to do the actual in vulcan from game programming standpoint and it's just i mean it's not the kind of code they don't only have to worry about that's the driver developer's job but suddenly it's not so what i think it's going to happen is these api's need to be integrated at an engine level and it's starting to happen with unreal and add an engine level the engine programmers can do a lot of that challenge tackling of optimizing their engine in general for the api and those Endon developers have resources with nvidia and AMD to further optimize for the graphics cards and get help from people who've been doing it for potentially over a decade at each of those companies so i think that's kind of what has to happen is it needs to come down from an engine level because it's just it's it's too hard to ask a game developer to do all of this new hardware level optimization via software when they're also trying to make a game it's a big cost overhead especially when they probably know how to use DirectX alone and it might be fine for what they're doing yeah I have a way better answer though that was pretty good but I have a way better answer in a previous SVN where I read a developer's responsibility have like one more is it yes Z no nuke says nothing like making some sandwiches and sitting down only to see a new gamers Nexus video I included this in here because this made me smile as comp because I'm not a big consumer of content these days I watch it but it's not only from like a competitive standpoint or we're like we'll watch videos from I mean obviously I watch stuff just because it's fun too but watch videos from other channels just to see what is this car channel doing was this fashion channel doing whatever what can be learned from these people outside of our industry and adopt into our industry so it makes it harder to watch stuff just for fun like I used to do but I remember watching like competitive Starcraft replays with like husky Starcraft or watching GSL gstl stuff like that at 6:00 in the morning before going to bed and it was my favorite thing to get some food and then come back to town and watch youtubers or I guess it was just youtubers really talk about professional Starcraft and that was a lot of fun for me so seeing that comment they day but anyway thanks for watching subscribe for more go to patreon.com/scishow and exit stops out directly go to store it on cameras excess net if you want to pick up one of our mod mats which are basically here at this point they are in the airport a couple miles away or one of our 3d laser engraved tear down crystals which the large one is basically picking up for hand-to-hand combat and also we have an autographed version now if you want that thanks for watching I'll see you all next time