Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

EVGA RTX 2080 Ti Kingpin PCB Review & Samsung vs. Micron

2019-03-31
hey guys build Zoid here from actually hardcore overclocking and today we're gonna be taking a look at the RT X xx atti kingpin kingpin edition from EVGA this is the as far as I'm concerned the perfect RT X 20 80 TI PCB it's so perfect in fact that I really wish in video would allow EVGA to make Titans but that's never gonna happen before that this video is brought to you by mass drop and the PC 37 X gaming headset with professional-grade Sennheiser noise canceling microphone the PC 37 X headsets are what we use in the office for phone interviews where audio and mic quality are critical making it very convenient high performance solution for gaming or professional work the headphones come with a detachable ten-foot cable for safe storage during travel a standard 3.5 millimeter plug and soft foam for firm but comfortable fit over the ears the PC 37 X headphone stand apart with high mic quality mixed with high quality audio output learn more at the link in the description below let's get right into it because there's just so much stuff to cover on this thing that that's really interesting that you know we need to get through this so then starting off with the VR I'm probably in my opinion the the least interesting part of this card because we've seen so many high-end RT X xx atti it's always very similar it's not necessarily the same thing but it is very similar you have a V curve erm that it's split into two parts normally with six power stages on that side 10 power stages on this side and the thing is you know some some of the cards out there they'll opt for like eight power stages on this side and well the founders Edition goes to seven and there's a couple cards that like there's one card that goes to eight and then everybody else goes to ten on this side and then six on the other well except gigabyte where gigabyte goes to eight on this side and then eight on the other side as well the top cards end up with 16 power stages in there in there v curve erm above the right side well yeah right side V curve erm we find the memory of erm so V mom up here that's a three phase which is standard as well we've got a 1.8 volt rail and a pecs rail in this area I'm not sure which ones which because they both do well they're both similar regulators they don't push whole lot of power and they both kind of don't matter so like well no they matter in the sense that they would like if these weren't present the card won't run but they don't really do anything for overclocking the pecs rail is well pecs rail sometimes helps when you're on liquid nitrogen but the 1.8 volts rail never does anything as that's a supporting voltage for the GDD r6 like the only reason why you'd ever want to raise the 1.8 volts rail is if for some reason you're running like 1.8 volts memory voltage but at that point I would probably I'd expect the memory controller to die so again not really a useful like not a realistic use case either that 1.8 volts might also blow up the memory chips though memory chips didn't to be really really robust and is usually the memory controllers that go that go first anyway so yeah it also powers things like the BIOS chips which run off of 1.8 volts as well so we have those miner rails somewhere in this area on the card there's going to be 5 volts regulator because of course these are smart power stages and they do not run on 12 or 3.3 volts so there needs to be a 5 volts regulator somewhere on the card to power those alright pretty much that's pretty much the top top level overview over the various voltage regulators on the card now then let's take a you know go into more detail on the V Corps and the memory of URM because those are the one they'd like the only ones you really need to care about that much even those aren't that special so we have an MP 2888 voltage controller for the V curve erm and we also have another MP 2088 for the for the memory of erm this is made by monolithic power systems so you know they make power stages they make controllers and they actually make some pretty advanced controllers so this thing is fully digital goes up to ten phases five megahertz switching frequency which is way more than you're ever gonna use because I'm not aware of any power stages or MOSFETs that well Massa you can probably find his grete MOSFETs that would do that they probably wouldn't handle a practical amount of power that's kind of the issue if you want to mosfet that can switch a lot of current it tends to be kind of slow so it can go up to five megahertz but EVGA is using it in the 100 to 600 kilohertz range which basically the switching frequency it affects your output regulation for the vrm essentially higher switching frequency reduces your output ripple but the other thing is higher switching frequencies can also negatively impact your vrm efficiency depending on what kind of power stages you're using and how the rest of the vrm is designed so yeah that's why they decide to top out at 600 kilohertz and not the 1 megahertz that the actual smart power stages support which our MP 86 9 5 6 is from also monolithic power systems these are 60 amp smart power stages I'm not sure on like I'm not sure that these have any special features I would assume they do like for example inter-cell and say international rectifier well they have body braking mode or diode braking mode which basically the same thing it's essentially a way of reducing the over voltage overshoot on a trend on a load release which basically means the transition of a VR I'm going from high current output to low current output or when you do that then you basically end up with a bunch of excess energy stored up in your inductors and that tends to cause a bunch of voltage overshoot now body braking mode can prevent that by essentially burning off the extra energy on the body diode of the of the power of the low side MOSFET in inside the power stage now I'm not sure that these have this because I can't get a full datasheet for these because it's not public I was lucky enough to actually get efficiency curves for these so that they supported up to a 1 megahertz switching for 1 megahertz switching frequency so yeah that's basically why EVGA is restricting the vrm up to 600 kilohertz because well you don't want to go above 1 megahertz because at that point these will stop working properly and at 1 megahertz they're also going to be very inefficient so that's kind of the the limiting factor on a lot of power stages is just how far you can push the switching frequency before they become a completely start malfunctioning or get really really halt I noticed that these were advertises slight more accurate than than the average smart power stage which I didn't really bother to verify too much they also include a built in temperature monitoring over current protection over temperature protection so basically everything to prevent the vrm from blowing up if something does go horribly wrong and yeah that's what makes them smart power stages this is like a new standard of voltage right like power stage from Intel pretty much so you know we're gonna start seeing these a lot and the thing that's kind of interesting with with this vrm from EVGA is that this is the first twenty atti that I can think of that actually uses 60 amps more power stages and not 70 amps 70 M smart power stages note that that makes a huge difference to the vrm efficiency as those are our efficiency figures and quite frankly this right here is better than all of the other twenty ATT is that I can remember covering partly because these actually and I'm gonna partly blame that on the fact that these come with a datasheet that's actually SPECT at 1.2 volts not just 1.8 so for a lot of the other power stages I would take like 1.8 volts efficiency figures and then try scale them down and with the scaling down I would be very conservative because as far as I'm concerned under estimating how much heat something puts out is a lot worse than overestimating it so essentially with a lot of the other cards there there's a good chance that the efficiency figures for them are too high which makes these look better than they should be but nonetheless these are definitely very very good and more than you'd ever really need to worry about it's also worth noting that there's h2o like the h2o 400 amps here is it's the bit generous tin from EVGA basically told me that in their testing of the kingpin edition of this card right here 2100 megahertz core clock and 8450 memory clock the card would pull about 420 Watts under 3dmark so you know I guess if you over vaulted the card a lot um and maybe hit 2200 megahertz or something you might be able to hit that 400 amps but actually necessarily you could get close to that but you won't necessarily hit it either way even at 400 amps output this VR I'm barely produces any heat plus that heat is split and spread across like two separate sections of erm so yeah this is gonna run really really cool and it's gonna run even cooler if you've seen the video of this card in the EVGA lab when it was running this thing hasn't really elaborate vrm cooling system like oh those things gonna have no like I'd be very surprised if this had issues dealing with 600 amps output because you know you'd have what you'd have a slightly more than half of that so let's just say like 45 watts on this side and there's a massive heat sink on on the vrm for this portion of the card and then there's another really large heat sink on the other end of the card as well so if you just put air flow over that you should have absolutely no problem dealing with with 68 watts of heat so 600 amps output and the vrm here is ridiculous oh well yeah really Achilles overkill I mean even at 700 amps you know given enough airflow this should be no problem to handle for shorter benchmarks there's a lot of other cards that you help to use the U P 9512 which only goes up to eight phases then P 2888 goes up to ten phases the Galax hull of Fame card uses an infant roller that also goes up to ten phases and then the twenty atti Strix as well as the 20 atti matrix which those two cards share the same PCB those also use an MP 28 88 in a 10 phase configuration this one's standard ridiculous overkill it uses yet more mps more of the MP 86 9 five sixes so you know 360 on power stages for powering your djidi djidi dr6 more than enough absolutely no concerns there in terms of power delivery either and the MP 2888 for voltage control up there now then things that I think are far more interesting than the V RMS on this card I guess let's start off with the memory chips so normally when you buy a NVIDIA GPU you get whatever memory chips happen to be on the factory floor that day I won't besides what memory chips are on the factory floor that day is whatever memory chips and vidya decided to ship out because normally when board partners buy when board partners by the the GPUs they'll get the GPU and a set of memory chips with that GPU which basically means you're gonna end up with whatever Nvidia decides to ship as Nvidia really doesn't care about what kind of like memory overclocking Headroom you know EVGA is kind of going out of their way to either like hold back on using their supply of GDD our six that they get from Nvidia and just like filling out the other cards with other memory chips or they're buying GDD are six directly from well buying GDD are six from some kind of vendor that is always gonna be Samsung and the reason why this is important is the Samsung GDD are six is consistent like is consistently good which basically means like there there are micron cards out there which do very good memory over clocks but there's a lot of micron cards out there where the memory overclocking absolutely sucks so with the Samsung memory you're pretty much guaranteed at least you know like good memory overclock I don't want to be overly generous but the the Samsung GDD are six should most of the time hit like 1200 or higher offsets easily higher I would be very surprised if it didn't do higher but I don't want to be you know I don't want to hype this card up more than I already will with with memory overclocking figures that aren't realistic but there's a lot of things EVGA is doing here that should make the the memory absolutely clock like crazy so one one thing is the guaranteed Samsung memory chips which all generally clock better the other thing EVGA is doing that I'm a huge fan of is a significantly beefed up output filtering set up for the memory VRM and one of the things I do to Nvidia cards these days a lot is just whack a bunch of capacitors on to the memory rails because I think it was a gtx 1070 where i first noticed that if you just added a bunch of capacitors the memory overclocking suddenly got way better okay it was a micron gtx 1070 it sucked and then I added a bunch of capacitors and it stopped sucking so much and I had a twenty atti founders Edition to play with and that's exactly what I did I whacked a bunch of extra capacitors on the memory rail and the card basically picked up some plus seventy-five megahertz offset that you know this card right here should have that taken care of from the factory which is great this has Samsung memory and this has significantly better output filtering for the memory of erm compared to all the other cards it also helps that this card will have memory voltage control which is another easy way to get a bunch of extra memory clock as well because the GDD r6 the samsung G DDR six definitely scales with memory voltage the the memory memory overclocking on this thing should be absolutely killer now that we're on the back of the card it's also worth noting all of this so you know not only have they significantly improved the output filtering for the memory vrm they've done the same for the vcore now on the vcore side i actually don't know how much impact this is gonna have I have a sneaking suspicion it's gonna have very little but I've never tried like I've never tried adding this many multi layer ceramic capacitors to anything because I'm not a pick-and-place machine and this would suck to do so yeah I like on paper this should help in practice I'm not sure how much of a difference this would make but it is definitely the most like like you can just look at all the other cards none of them have this many multi-layer ceramic capacitors all in the output of their V curve erm and the reason why this is significant that it's multi-layer ceramic capacitors is because multi-layer ceramic capacitors have a far superior frequency response compared to you know what you normally find which is like aluminum polymers or tantalum polymers or any other kind of sort of like electrolytic type capacitor right because they're they're still electrolytic they're just solid electrolyte so yeah and the thing with the frequency response is basically that means the capacitor is better at filtering out high frequency switching noise from say the GPU core though admittedly like the it also is affected by distance and these are rather far away from the GPU core which is why I'm like well I'm not sure how much in fact these would have then again you also have just like transient or spot like these would still also help with smoothing over transients so when the GPU core pulls a lot of current and then suddenly very little current and then suddenly a lot of current again the multi-layer ceramics will help out with that you know since they are fast the issue most of the time the issue with multi-layer ceramics being used for like handling transient response is that they don't have a lot of capacitance to them but EVGA is kind of solving that by just having so damn many of them so yeah it's just like that's the thing is like these are like if you add this many multi-layer ceramic capacitors I'd be very surprised if it didn't do anything but I've been surprised by like I've done other things and ended up surprised with when it comes to adding capacitors to V curve erm so yeah I want to say this makes a difference but I'm really not sure it should but you know I would like on liquid nitrogen I'd expect this to actually help a lot on ambient I'm not sure that you'd be stressing the you know like the card would be pulling enough current to really make a difference so the bulk capacitors now this is an engineering sample card and you can tell because like this is missing and there's stuff on the back of the card that's also missing but and that basically means that these capacitors you see right here they're not final these are not the ones you'll actually see on the retail cards because these are tantalum polymers from Panasonic and the retail cards are going to be on Panasonic SX series capacitors and those are SP caps whereas these are pauls caps and the difference is SP caps are aluminum polymers and Panasonic makes some very very low ESR aluminium polymers and the reason why we want low ESR is essentially it makes the capacitor is more efficient and it makes them just better like better at doing their job I mean panasonic also makes capacitors that go all the way down to three Millions but those are ridiculously expensive and I don't think I've ever seen them used on any GPU ever because of how expensive they are like I guess at this point everybody's gonna be wondering what's their lifespan rating well it's only 2,000 hours 105 degrees Celsius so they're not high endurance considering that this vrm is probably going to be running basically at ambient I'm not worried okay I think it's and this is actually something worth noting is a lot of the very low ESR capacitors just straight up don't come in high endurance specs like the three million yes are capacitors you can't buy them in five thousand hours they just don't make them they also don't make them in high voltage ratings either so these are basically capacitors that is just like they're optimized for raw performance above sort of any kind of consideration for long term you know high temperature and urines because they're still gonna last forever if they're at like 50 degrees Celsius they're just not gonna last forever if the vrm was super super halt which it won't be because the vrm is very efficient right and really well cooled so yeah like I I don't see a problem with with the capacitor selection the EVGA is golden for I mean I can I kind of wish it went for the three million ESR ones but at the same time I completely understand that they didn't because those are as I said like ridiculously expensive and I think that with everything that this card is doing it's just kind of like okay you got a stop plus there's like a pile of multi layer ceramics on the back you know shaving off a couple of milliohms on the on the ball capacitors probably wouldn't make much of a difference considering what the back of the card looks like let's move on to other cool features we get triple eight pin power connectors which normally I don't think is that cool the thing with this card is the triple 8 pin power connectors allow this card to ship with a max TDP of 520 watts so when you max out the power slider in an EVGA precision for example this is what the power limit is going to be for the card and the reasoning for that is is a hundred fifty Watts 150 watts 150 watts and 70 watts and now if you remember a lot of my past videos I say that these power connectors can handle a hell of a lot more power than 150 watts and that's true the issue is and video what allow their board partners to ship cards with power limits higher than 150 watts per power connector or at least I think that's the situation because the way and videos power limits are actually implemented is there's not really like a full card power limit it's a per power connector power limit which is why this card can have a 520 watt BIOS you know power limit because 150 is in spec in spec in spec and still in spec and also the per power connector power limit is the reason why we have all of these MOSFETs and all of those and lat and also this choke and that choke and these guys and those guys and well I didn't notice any in this area so I'm not sure if this is power balanced or not but basically because the power limit is per power connector you end up in this really awkward situation where theoretically it's possible depending on how you have your V like vrm connected to the power connectors it is theoretically possible to end up in a situation where one power connector is pushing 200 watts and another one is pushing a hundred and or without all of the power balancing stuff that would cause throttling on the core clock so that's bad because you know effectively it reduces our power torque like you know you have your theoretical 520 well power limit but if you're pulling trying to pull 200 Watts from one of the connectors and a hundred from another one you're effectively limited to I'm gonna screw this up 470 Watts by instead of the full 520 so you need all of the power balancing to make sure that at any given time there's a hundred and fifty Watts across like if you're maxing out the total power limit that it is actually shared properly among all the power connectors and the way these MOSFETs do that is basically they reassign phases from one connector to the next connector on the fly so you know if like we have one phase here that's pushing way more current than the other one well there's gonna be different currents flowing through each phase you know reassign the phases so that they're all pushing added up there they're pushing the same amount of power for each power connector so that's what all the power balancing is for now if you're on liquid nitrogen there's going to be an extreme overclocking BIOS that's gonna appear on some forum I'm not sure where actually no I have a I have a good eye I have a rough idea of where it will show up and that BIOS will remove such silly things to constrain us as power limits because on liquid nitrogen Nvidia cards get very very power hungry for example 1080 tea eyes will regulate will average 800 watts when running on liquid nitrogen easily they will average clay easily 800 watts well Tim from EVGA always actually releases a a really cool guide for for overclocking these that usually mentions the average ln2 overclocked power draw and at least for like nine ATT eyes it was like 11 huh it was I think 1100 watts average or something like that for 1080 T eyes it was like 850 and his guide so for these we'll definitely find out what exactly they pull either way with 3/8 pin power connectors we should be more than more than sorted because each of these connectors can do at least three hundred and sixty watts if you have you know a halfway decent power supply now then other features we have a USB B micro that's the wording and that gives you monitoring capabilities for the card over USB which is pretty awesome then here we have an e V Bob header so if you have a TV bought you can use it to control voltages on the card which is uh I mean at this point like I wonder how many people there are who are still active in extreme overclocking and also have Eevee BOTS because those things are pretty rare because EVG hasn't been making them for ages above that we have a four pin fan header for some reason I'm here oh then we have the I assume that's for the OLED display that the card comes with so that'll allow you to monitor things like GPU clocks fps power consumption voltages like GPU memory voltage PEX voltage 1.8 volts and 12 volts it'll also allow you to monitor temperatures across the 12 different sensors on the PCB as well as the thermal paste monitor which is which is a sensor that goes was some like in - I think - 60s so you know goes pretty far below zero it doesn't go all the way that's light as it doesn't go as low as a K type so personally I would still it like if I had one of these cards and I was taking it on liquid nitrogen I would still glue a k-type thermocouple to the back of the GPU core because that's actually a really good way to figure out if your thermal paste failed because even if you're at like full pot and you know temperature sent like the the the normal temperature sensor that this comes with won't work at that point and regular 1080 T is and twenty eighty T is they won't register anything at that point the temperature readings going to be completely wack but what I do is I glue a k-type thermocouple to the back of the core because if you monitor your Delta through like from the beginning of your session and throughout your session if your Delta suddenly just jumps way up for no apparent reason you know your thermal paste just failed and I guess with this you if the card doesn't immediately crash I would assume with this thermal monitor the the thermal paste monitor on this it would actually start showing you know potentially like minus fifty degrees while your ln2 pot is at minus two hundred and if that's happening you know there's a problem anyway so it might not be that much of an issue that it doesn't go super low but personally i like i like to be able to monitor the delta you know and took really negative temperatures but it is also kind of awkward because you need twice as many thermometers for all of your GPUs and everything above the OLED connector right here oh and also the OLED display will allow you to display custom graphics if you feel like making those above that we find the Eevee ball I mean not the Eevee ball header but a voltage probe it header there that's what evj calls it essentially this is a hookup for digital multimeters super convenient it fits snugly onto digital multimeter probes well there's like a connector that goes in there and that then terminates in connectors for digital multimeter probes that fit snug so you can hook up your multimeter and you can just set it to voltage and you can monitor whatever voltages you feel like from your multimeter so in combination with that OLED display you could you'd actually have a very easy time monitoring like two voltages at the same time without you know needing two multimeters so it's pretty neat also this is a lot like there's a lot of extreme overclocking cards out there which basically just give you like voltage read points on the PCB itself like they don't give you any kind of hookup it's just like the solder bumps or something and as like it's it's nice that the they thought about adding that but at that point I may as well just solder wires to the back of it like two capacitors on the back of the GPU core at least from my perspective it's just like well if you're not gonna give me real hookups I can just and I'm gonna have to solder to the car to anyway I may as well just solder directly to the back of the core right like at that point it doesn't make a difference the this this is actually really convenient so it basically eliminates one of those things that you would have to you know so like on a lot of other cards you'd have to do Sol during to get these kinds of hookups here it just comes with it which is awesome that's kind of the main reason for the kingpin cards is just like you know these are cards where you can buy the card to insulate the card maybe add the extra k-type thermocouple to the back of the core if you're like me and then throw it on liquid nitrogen you don't have to do anything else is it just you don't have to worry about soldiering anything modding anything removing resistors bridging anything you know bridging resistors whatever don't have to do anything like that buy the card install the extreme overclocking vials insulate run next we got a triple bios which well since you are gonna be expected to like you know if you're gonna be installing the extreme overclocking while so you will have to flash the BIOS on the card and if in the BIOS process about BIOS flashing process anything goes wrong well having three biases makes it that much harder to break your card right like instead of having to brick just to BIOS is before the card is terrible before the card is unusable and here you have to you have three attempts to not screw up and then we get some LEDs for indicating which voltage regulators are actually running so n VDD is our core v RM so if there's anything wrong with the core that won't light up if the memory is bro if this doesn't turn on that won't light up if the PEX rail doesn't turn on which is one of these that won't line up 1.8 volts it you get the idea 1.8 volts is monitored and 12 volts is also monitored so essentially if there's some kind of issue causing the card to not fire up and one of these LEDs is oh isn't lighting up well you know you can skip a whole lot of troubleshooting with that well you can skip a good chunk of troubleshooting with that not a whole lot because at that point like if you're V Corps member v RM and your memory of erm aren't running then it's like why aren't they running and might be something like oh the five volt rail isn't running which you might have to hunt down at that point but uh still simpler than having a card that has no display out you know no display and you're sitting there like well I guess we're gonna be taking a multimeter to it and trying to figure out why it's not displaying anything well I think at this point I've covered everything on the front so let's move back to the back and this being a you know engineering sample while we have LEDs for the the triple bias to indicate which BIOS your own but this being an engineering sample it's missing some features over here so we have a voltage offset switch which I'm not sure what exactly it does but the like what exactly the offset settings are but essentially the idea is that you can offset your volt core voltage by a certain amount this can be very useful with twenty eight like well on ten eighty TI's this would be very useful because 10 ATT is on high core voltage you don't want to start up though I assume since this card you'd be setting your voltages and software that shouldn't be an issue I guess you would also be able to use this to override any like like software voltage limitations you can just go above them by a certain amount like whatever the offset gives you so that's potentially useful I'm I'm not sure I would ever use this myself but it's it's here if you feel like you need it then we have memory load line settings and GPU load line settings now the settings if I remember correctly are basically reduced droop and then droop completely turned off and you do kind of have to play with them because in some situations you may find that having a little bit of droop or all of the droop is actually better than having no droop because droop exists for a reason and I'm not gonna go into explaining why it exists but essentially on liquid nitrogen it's actually it's really obvious if you set up your load line settings wrong on liquid nitrogen because basically you can be like you can set a certain voltage then adjust your load line and you're gonna start crashing in benchmarks instead of finishing them it's that obvious all right ambient that doesn't really happen because at ambient you're not so temperature sensitive but on liquid nitrogen it's really really obvious when your load line is set up wrong so that pretty much covers everything on the r-tx 20 atti kingpin Edition the most you know complete art like the zeiss I think at the start I said this is basically a perfect RTX 20 80 TI and and I stand by that it does everything I can't think of a single thing well no I can think of one thing I would try I would really like to see this row of multi-layer ceramics replicated down here because those are for the memory power right and I and I I just have to wonder could you make the memory clock better if there was more multi-layer ceramics down here for the memory chips that are over there instead of just the chips that are up there right this kind of seems kind of imbalanced and also you can't put any over there without shifting the VR I'm back and that would potentially haven't you know negative consequences for core clock so I don't want to mess with that bark but I do wonder you're like what would happen if there was more of them down here could the memory overclocking be even better but yeah other than that you know I can't think of anything I would change on this card it's it's that good the only card that I can like out of all the 20 ATT is that I've seen the only one that comes even close to this is the galaxy all of Fame and the thing with the Hall of Fame cards is they they don't have a lot of the things that this has and they're kind of more awkward to get your hands on the extreme overclocking utilities for whereas the whereas the kingpin cards historically have their BIOS you know just kind of placed on on on extreme overclocking forums and if you need that BIOS and you know where to look you can get that BIOS so that's that's always been kind of my like so that's that's the really awesome part with the in additions so yeah that's it for the video thank you for watching like share subscribe leave any comments questions suggestions down in the comment section below and if you'd like to support what we do here with gamers Nexus there's store dot gamers Nexus not net if you'd like to pick up some merch like mod maths shirts glasses and if you'd like to support us directly there's the gamer sexist patreon there's links to both found in the description below or in the comment section or in both actually I'm not sure Steve takes care of that stuff so anyway I have a youtube channel called actually hardcore overclocking where I do a bunch of other overclocking relates and stuff it'd be awesome if you check that out as well and that's it for the video so goodbye
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.