Gadgetory

everyone welcome back to another episode of ask GN we're filming another one right after this for the patreon extended version go to patreon.com/scishow to get access to that but this is the main show so if you want to leave a question for next week post it in the comment section below or on the patreon asked yen channel and we'll try and get to it for next week's episode this week we've got a couple of good ones one on how much does vrm power delivery actually matter and i took that question to our expert build Zoid to get his quote on it and i think that'll probably the most interesting to a lot of you but we also have some fun ones about things like well GN design a case so we'll get into that after this before that this video is brought to you by Corsairs new dark tour RGB SE mouse the dark poor RGB se is a wireless gaming mouse rated for up to 24 hours of continuous wireless gaming with the LEDs enabled and can be coupled with a Qi charging mouse pad for easy battery charging it has both wireless and Bluetooth antenna so the mouse can be used easily on two systems and switched between them learn more at the link in the description below quick mod mat update for everyone first a couple of you have asked when the next round will ship the good news if you've back worded it already or your back ordering it soon you will definitely get the next round of mats the we're not completely sold out of the next round yet so you have time store doc here miss Nexus not in that if you're interested in it we should be getting them shipped to us within I think our target right now is 10 to 12 ish days maybe 10 to 15 depending on when this video goes live and then from there obviously will ship it out to you also add the usual shipping time for that if your domestic or international that's your answer it takes a while to get this stuff shipped over after it's finished production so first question first real question gamer games asks asked yen a third time asking sorry the post keep posting below we get a lot of comments does vrm power delivery matter in overclocking EG 13 phase power delivery overclocks higher than eight phase power delivery thanks I'm going to I have a big answer from build Zoid I'm gonna sort of preface this by saying it depends the question you're at skin depends on all kinds of stuff like what kind of components are you overclocking how far are you overclocking them and we'll get into his answer for the rest of this so build Zoid from actually hardcore overclocking he's got his own YouTube channel you should subscribe to if you haven't he says it really depends if you ignore the thermal aspect and just look at the electrical benefits of multi-phase you basically get this lower Ripple and faster transient response basically more phases equals better voltage regulation however it's easier to stay in spec on a 10 amp to 20 amp jump than on a 10 out to 100 amp jump also ripple at 10 amps is lower than at 100 amps for a given set of components continues to say at low currents the electrical benefits of a multi-phase design aren't relevant because a single phase wouldn't be having issues once you get into very high current draws the multi phase designs will start showing their advantages as keeping good ripple and I'm gonna read this part in builds eyes for haste once I got to it and transient on a single 100 amp phase is really really hard to do now if we weren't limited by physics you could build like a stupid high switching frequency single phase that could rival a massive multi phase he continues bit by saying I think the simplest answer is that having more phases only really helps for higher power draws though I would recommend against overclocking on what on less than four phases even on something like at 7600 K I think a 73 50k would probably be kind of okay or at least the i3 shouldn't clock much worse on a three-phase compared to a four phase something like if somebody some 100k could end up requiring much higher V core setting to hit the same clock compared to a mobo with more phases or more true phases so that's probably best answer you're gonna get it does depend a lot this is something I talked to the builds I had about as well but it depends on the CPU you're overclocking as he indicated here with i3 as i-5 Zeiss Evans of course Rison same thing applies it for example if you're overclocking an eight-core Eisen CPU versus like for core huge difference in the power requirements in the thermal requirements and the of heat that the vrm is generating also again this does depend on what you kind of consider to be your stop point for an overclock if you're trying to do stuff like build Zoid then you definitely need a lot more than the average user who's trying to hit like four point eight or nine on the average Intel i7 chip or 3.9 4.0 on the advertising chip but his answer is pretty good for you so I'm gonna leave it there next question oh and actually hardcore overclocking if you want to check them out next question is a balanced diet who says if given the chance to influence the design of a piece of hardware such as a case would you do it why or why not for example of NZXT he came up to you and said Steve we're looking to make a high airflow version of the s340 elite complete with a mesh front panel and we'd like you to help us design it what would you do we've kind of talked about this before designing cases is really hard I have a lot of respect for the people that do it even when like even when their cases aren't the most positively received by us I still fully understand how difficult it is to make that because I mean we've talked to all of these PM's and product managers over the years I've heard the horror stories we work with factories ourselves now on things like the mod mats and I get it so it is hard that said yes I I do think it'd be fun to work on I don't have a design in my head right now one day we kind of have some some completely not concrete won't happen anytime soon ideas of making a one-off case just just to do it just for fun go to like a local machine shop or call some friends with the equipment to do it and some of them some designs and see what we can put together that's not happening anytime soon I'm not promising content on it today or even this year but it's something I'd like to do it's kind of like a distant content idea that I really like and and I think we could offer a bit of a unique angle on it if a company came to me I mean sure if it's kind of like dear Bower is a good example where he worked with Leanne Lee to make the dynamic case I'd be open to it for sure that said it's definitely an intimidating task like I've seen a lot of cars from our viewers really like Steve should design a case GM should design a case I we could probably get the airflow side pretty well but though everything I also like the looks that's difficult design ID is really hard maybe we'll look into it some day for fun though I'm certainly not going to be mass producing cases anytime soon but it'd be fun as just like a mock it up in CAD bring it to a machine shop or some friends who can put it together and and make it and test it just to see what would happen but uh yeah you sure sure if someone if a company came up to me and said we'd like to design a case with you I'd be like yeah let's do it that sounds fun but yeah again the compromises you have to make in that kind of design are challenging and I respect what they have to go through to do it next especially when you consider the cost of things next question Matt the tech says hey Steve I just joined patreon to ask this question but we'll be staying because I want to support my favorite tech journalist thank you I appreciate it when precision boost override is turned on on the 2700 X what is the TDI TJ Maxx of the 2700 X I can't find anything official about this I was hoping you can investigate for us I know with precision boost override turn to off that is supposed to be 85 degrees Celsius whether that's TDI or tea control I'm not sure but there is nothing out there that I can find about the throttle point when you have precision boosts turned off or on I guess depending so the tricky thing here is that Rison doesn't throttle the same way that Intel CPUs do and because Intel's had market share for so long I think we've all kind of gotten used to that for like like the understanding was when a CPU hits a 100 degrees or whatever TJ Max is for the CPU then it throttles down whereas with GP is the understanding is every couple of degrees you lose a couple megahertz because they use different boost technology GPB is three and what I'm even AMD does now in Vega cards same idea so with AMD CPUs rising TVs they're getting more and more like GPUs in the way they boost rise in to just like risin want except even more so is really sensitive to temperature so you could push 90 degrees on it TDI sure but the stability of the overclock is going to be much worse at that point AMD officially just so everyone knows their official number I I think their official number is sixty point eight degrees as the optimal temperature that doesn't mean you should buy like a 600 millimeter radiator to try and get it to be sixty point eight degrees if you're like pushing the high overclock it does not it absolutely doesn't mean you have to run it at that temperature but I'm just saying that was there when when creating their TDP numbers the metrics they use for optimal die temperature is sixty point eight so that gives you a starting point to answer the question directly TJ Maxx I think is supposed to be about 95 but we've run CPUs that don't hard shut down at that point and they don't throttle either so that's the tricky point where you see rise in two CPUs and it's not like they hit 95 or they hit 90 or 85 or whatever it's they don't hit some number and just magically throttle down 300 megahertz lower like you expect with Intel typically what happens is you just lose boosting Headroom of its stock including access far too and if it's overclocked you lose stability so it'll keep trying to hold that frequency but at some point you're gonna have a freeze a blue screen whatever and you'll have to restart now in our testing and our experience staying under 90 TDI helps a lot with to build a whole lot with stability staying closer to 80 to 85 has proven the most stable with the lower voltages that we try to use you can push higher voltages to maintain a higher overclocks but obviously the more voltage you push into it the hotter it gets do you have kind of a feedback loop where you're trying to increase stability by increasing which increases heat which decreases stability so that's kind of problem you run into it with it and I don't mean problem like a design flaw I mean problem like it's a challenge when overclocking it works pretty well as a design it's just you know expect different things when you use different products to ever talk so to answer your question it's about 95 is is kinda the hard limit in theory but it'll depend on the motherboard different motherboards will interact with t.j.maxx in different ways on Intel for example just because I'm still the most familiar with that cuz riding to still new on Intel you can hit t.j.maxx on a board like a hundred 105 with a EDT CPUs and my down clock or it might shut down it depends on how its configured or if you changed it for what it does when you hit TJ Max and what kind of protection you wanted with rise I'm like I said it's not so much of a hard number that you hit and things down clock like if you have hardware monitor open or hardware info open and you're expecting to see a big down clock when you hit a certain temperature threshold it's more likely you're just gonna see a freezer a crash then then kind of temperature or thermal throttling of the frequency except in the case of stock performance in which case you'll see some boost limitations but we're talking more overclocking here because that's where you're more likely to hit the higher temperatures anyway next question Daniel Munez says on episode 77 around the two-minute mark you mentioned that on Intel Hades Canyon nook the Vega MGP present isn't really Vega but it's kind of Polaris what do you mean by this Vegas memory controller is made oh yeah yeah so these are my notes I want to read to you so here's the thing this comes down to how do you define one architecture versus the other AMD calls it Vega so really at the end of the day it is called a Vega product it is called a Vega GPU that's the official branding but as for what is Vega or what is Polaris you have to kind of draw a few lines to decide the Vega M GPUs on the Intel nook are a semi custom solution AMD worked with Intel to make this this Vega branded GPU it uses Polaris technology he uses Vega technology because at the end of the day it still has a high bandwidth cache controller so I was HPCC to me that's a very definitive this is Vega feature HPCC is something that was really specific to Vega it was a big deal with vega's development early on and that's a defining characteristic that I would I would point to that and I would say that's Vega architecture it also has HBM too so it's got high bandwidth memory something that Polaris does not have although Fiji had high bandwidth memory 1 HP I'm with the fury X but Polaris clearly doesn't have HP on to you so that's a big differentiator as well as for other stuff there's differences of well like I said polarities GD r v GC n 4 vs 4.1 2 GC on 5 on Vega and you end up with a kind of just an existential question of how do you define it because it is kind of Vega but it's not the Vega that we expect when we talk about Vega 56 or 64 so yes you bring up great points here in in your question but it's it just kind of depends on if you're using strictly andis marketing or you want to call it like Vega asterisk kind of not but kind of is PCWorld has a great article on this they kind of started talking about it there's some other stuff out there as well so yeah Vega M doesn't currently have DX 12.1 support but it could be added probably the Vega M stores HPCC it's got HB m to it does not appear to have drawstring binning rasterization or rasterizer z-- or rapid packed math I guess so technically yes those are Vega features but we don't really see them used too often anyway and I would call HPCC a a higher echelon of Vega feature than rpm but that's just kind of define it myself yeah I think a couple other things were pointed out device idea indicates Polaris but that's just a device ID doesn't necessarily mean anything there's a high rob count versus a low CU countin that's pretty unique that comes into the semi custom design so when I said that what I was referring to is it's not the Vega it's not Vega 56 or Vega 64 it's not the same thing that you think of architecture as those because it's got some polaris tendencies but it's not 100% Polaris either it's a unique thin in-between and for the record we really like the Hades can you know that's probably my favorite thing I reviewed this year just cuz it's so interesting because Intel and Andy work together on something that it's really good like yes it's expensive but it's still really good hardware and I can appreciate that in isolation of the price and that's because they work together instead of trying to kill each other for five seconds and you know with something really cool so you end up with a semi custom half Vega half Polaris type of thing that Intel probably had some specification on depending on whatever they were trying to do with the box next question rusty Shackelford says as stupid as this is going to sound what's the difference between HB m and G DDR memory chips is because how they're manufactured how they're programmed or something simple like the resources they're made from I think this question is referring to price because in the video this was posted on we were talking about the cost to make HB m 2 versus gddr5 were you talking like 60 plus dollars for a four high stack HP m2 versus nine ten dollars for a an eight gigabit per second gddr5 piece of memory so I think that's what it's in reference to the difference is first of all it's kind of zoom it out actually it's really convenient and unintentional but I have a Titan V in front of me here so let's just zoom it out start with basics a card like this has HP m2 on it it's got its surrounding the the die on top of the substrate but technically there's something in between the sub-state substrates excuse me and the memory of the die and that's something in between is an interposer so an inter poseur is created at the fabrication plants just like HBM and other dyes interposes are limited to the same lithography limitations as any other dye that you make so they can only be as big as the largest dye that your fabrication plant can make even though it's not a it's it's not like a memory module or a GPU it's an interposer that sits under the GPU and the memory modules and it bridges everything together it handles all the communication so I've got this stack of things where kind of have like PCB with BGA ball grid array into the substrate into interpose a GPU HBM all that stuff whereas with G DDR memory your memory would be surrounding the outside of the card so you end up I mean you've all seen it you know with memory like this instead and that's the major difference that memory is BGA I think it's a 180 ball grid array and physically HBM is closer to the GPU so that's an advantage you don't deal with as much latency it's higher bandwidth as the name would suggest now depending on how many stacks you have that could be kind of negated like running four stacks we're just - for instance it's kind of like running multiple memory modules on a flash device versus fewer where you kind of have saturate the amount of channels available that could limit your bandwidth it if you only use one stack of HBM for example like the Haiti's Canyon look you don't have really fully realized benefits of the high bandwidth memory except for a few and a few of them are way lower power consumption pretty low heat flux on HP m2 as well so try to cooling hbm's much easier and power wise it uses a lot less power and I'm not giving numbers because I don't keep them on the top of my head anymore but if you go to our content that's called I think it's like how much does HP M cost or how much does Vega cost to make I went through some of the power metrics there and what the wattage actual number numerical wattage differences are I went through the price differences stuff like that I would recommend watching that content it was all in my head at the time we made it about other things that are different with HP m they are 3d stacked DRAM dies so HBM when i say for high that's referring to kind of how big the stack is vertically it's kind of like I don't know you know in a way like v-nand or 3d net where you're stacking layers to create the memory and that's why you end up with like on Titan video with four stacks as opposed to eight individual memory modules as part of it density is different to of course another dish I think I noted a couple other ones interposer yield so this is interesting to some of the cost comes from the yields also the demand for HBO is pretty low right now relatively but yields alone you have you have your HBM die yield you have your inner poser yield your GPU yield and then you have your package yield which is should be high at that point but the inner poser has its own yield has its own cost it's like twenty twenty-five dollars cost and then you've got your HB m that's in the 60s for a four high stack plus the GPU and I don't know what those cost unfortunately so that's a difference but yeah much closer to the GPU much lower power consumption vertically stacked which means that it's smaller so there's a much smaller footprint on the board you can make I mean I wish they did but I think that Vega nano was displayed by Andy a while ago that's an example of something that would benefit from all of those things so I'm but check out our previous content for a like a 20 minute detail on it next one wraps up 100 says GeForce experience our Radeon relive does it relive yes yes I think it's pronounced relive yeah so um this is just showing how we cover so much stuff that obviously I can't memorize all of it forever it kind of starts leaving eventually but relive I liked a lot I we used to use shadowplay almost exclusively for capturing b-roll of games gameplay footage stuff like that and shadowplay I have a lot of problems with now yes I know I know we just ran ads for it and it's still decent software but the thing is when Nvidia introduced the registration button that was kind of a big problem for me because it used to be with GeForce experience I think 2.0 or whatever it was you just launch it and you'd run shadow play you didn't have to log into anything there was no involved you just launched the thing he hit alt f9 it record your game and now he needs to log into an account which I had really dislike strongly clearly they want your email address clearly they're doing something with data logging whether or not it's malicious I'm not gonna comment I I don't I don't have any reason to believe that it is but also they're collecting data clearly so that's not the problem I have with it though although I can fully appreciate why you would have that problem my problem is as inconvenience here's and shadow plate hasn't been updated at the same pace as Reliv so Reliv was like it was really surprising to me when it came out we did a couple of videos on it and compared it to shadow play overhead for shadow play in games and when I say overhead I mean let's say you play a game at 100 FPS and you start recording it clearly you expect some loss of performance in framerate and typically if we're talking 100 you would see a dip down to 97 with shadow player talking less than 3% on average relive was really damn close to that I mean in most cases it was about the same and these days now that it's been updated and improved further I would expect it's about the same if not better than shadow play for overhead so I do really like it if you have an Nvidia card you should probably just just stick with shadow play but I don't even know if Reliv works with with Nvidia cards anymore I know that their previous one GVR technically worked with all devices but that was a raptor thing that and he worked with them to bundle it in with one of AMD's software developers I don't know it was a big mess so gbr was was pretty messy but really if I like a lot the softer is really good and he's done a good job to build their software package in a way that's like much easier to navigate really small stuff like navigating folders with shadow play I don't know if they've updated this recently but for a while when you tried to navigate between folders you couldn't type in a folder path you or even click on like the windows directory through a normal Windows Explorer dialog you had to navigate through this stupid custom built-in file tree and dig your way up out of the grave your and to find the file path you wanted whereas a three live you just basically type in a path and he was there so I really liked relives navigation I thought it was well-designed when it came out I'm sure it still is I just I haven't looked at the updated versions but I did genuinely like relive a lot and if I could take their navigation I think that's really the main thing I would want to see shadowplay improve on shadow plays just fine functionally as a software they're both great at what they do it's just one as much better UI and UX than the other and that's real it's a credit to Andy there bill bruce says Steve have you ever watched your videos on point 5 speed I watch them out like 2x when I'm previewing that before we publish AHA next question count him out of here says how do you feel about people calling you Tech Jesus I'm not here's what I want to know who started it does anyone know who started that was it just commenters or was there a youtuber who started that I feel like it might have been pong Kyle but I didn't let me know if you specifically know where that came from clearly I'm no carpenter though if you've seen some of our attempts ed wood stains next front wraps up oh yeah I guess you had you in there twice I have several 100 under Steve would ever be employed by one of the manufacturers in the future no not going anywhere I've zero interest I have worked for companies in the past I worked for Dell it was a good experience but I'm not really big into the bean employed by people thing I like to kind of do stuff my way so no I mean yeah I am NOT interested in that I plan to basically stay here forever so if you don't like that I'm sorry you're stuck with me that's it for this one thank you for watching patreon version will be at patreon.com slash gamers Nexus and we have lots of cool stuff coming up this week so we're done most of it tonight including more stuff with this thing an IHS so check back for that subscribe from where you go to store that gamers Nexus dotnet to pick up our GN crystal tear down cubes and the GM mod bat thanks for watching I'll see you all next time