Why I get NO POST with XMP??? 👉 ULTIMATE RAM + XMP Guide

so why is the XMP on your PC not working especially with ddr5 and when trying to run four sticks I mean you might have experienced crashes bugs all sorts of weird things when trying to run XMP or Xbox on your PC well there's a few reasons why this might not be working and there's actually simple fixes and tools what you can do to actually fix that and in order to help you understand XMP and solve the XMP issues I've teamed up with JJ from Asus and I've put together the Ultimate Guide to understanding XMP and some of the tools how to fix these things so if you're a beginner or maybe a complete Enthusiast you've already known all sorts of things in the PC industry for years and years I'm sure you can learn something new about this conversation about XMP with JJ because ddr5 is slightly different than previous generations and if I work for Linus I'd say something that likes of you know what else is different the sponsored segment what he does not but I don't so I'm not gonna say that looking for a cheap way to license your windows check out who Keys through the links in the video description make sure to use the code tn20 to get a 30 off paste the license to the activation settings and you're all done this license is for Windows 10 but you can upgrade it to Windows 11 for free they also offer Microsoft Office 19 license use the same code tn20 to get a 30 off check out who keys.com in the video description below could you give a little bit of a background where you work how did you end up there another background what kind of position you're in at Asus yeah so I've been with Asus for a pretty good chunk of time just about almost 15 years so I've been able to see the company grow and evolve in a lot of ways I've actually handled quite a number of different roles uh probably the most uh kind of I'd say involved rules that I've had have been coordinating with our product marketing team and our headquarters team in terms of kind of product feature design and development and Community engagement and kind of bringing that all together holistically um my current kind of roller designation is the senior technical product Marketing Manager for Asus North America and like I said that covers a lot of different aspects in terms of ultimately what I do but um you know the main kind of I'd say focus is one to engage with our product marketing team as we go to market with new products such as let's say a new motherboard launch chipset launch help to be able to provide kind of insight in terms of how the community amongst how users amounts how techno media will evaluate look and kind of ultimately adopt our products and help to have them a better understanding of it and you know that like I said that because of ultimately what we're trying to do there's a lot of things that kind of ultimately fall into that for my side in terms of just being a PC DIY Enthusiast and somebody has to inform myself and understand things but at the end of the day hopefully what I'm trying to do is help to provide more insight and edification regarding our products and our Technologies as a whole so when it comes to RAM and XMP compatibility what is your experience been in Asus about that or how do you link into that yeah so that's a really interesting topic because it involves a couple of different ultimate narratives when you talk about kind of X and P so one of kind of the really good things so first we're trying to understand is what XMP is sometimes there's a little bit of kind of I'd say conjecture or even misinformation in terms of sometimes how people try to understand what it is and in short really what XMP is is an overclocking profile and it's important to always add an overclocking profile at the very beginning of that conversation the reason being is that overclocking is for one is never guaranteed and it's not part of the base specification design of a CPU product right and so essentially what that means is that when you talk about either Intel or AMD when they are designing the actual CPU and this has been now the case for multiple years so multiple generations for both Intel and AMD you have something that's called an IMC or the integrated memory controller so that's the part that's inside of the CPU we're not thinking about you know the cores or the cache or kind of maybe the traditional parts of the CPU that a lot of times get talked about when you're talking about performance but they heavily implement influence performance so that integrated memory controller has a defined set of operating specifications and essentially this is done to help to ensure that if you were to have a hundred CPUs produced or of course an Intel and AMD situation they're producing thousands of them right you want to be able to ensure that they meet a guaranteed level of operation so there's actually something I want to show here that hopefully will make this a little bit clearer when you talk about actually memory in the IMC in general but you can definitely always reference this information directly on the product page from either Intel or AMD it's actually called the Intel portable AMD does also have one so if you were to go to their website you'd be able to find this information this one right here is for Intel 12th gen but it would be a little bit different but the same format would apply essentially for Intel 13th gen and what you're seeing right here is the Intel default Port table so essentially the operating set of specifications that the memory is designed for so we see right here we have a couple of different terms this is a one slot per Channel One dim per Channel and then memory Rank and we'll touch on that a little bit because it is going to be part of the conversation but the main thing that a lot of people you're aware of is this kind of let's say frequency number or this bandwidth kind of number right here right here so this is 4800 Mt that's what the default was for 12th gen and then on of course Raptor Lake 13th gen it went up right but the interesting part here is you'll notice that depending on the configuration that you have for your system that actually number will go down so it doesn't mean that there's a guaranteed kind of consistent number regardless of the configuration and this isn't something that's new for ddr5 it actually also existed for ddr4 and for DDR3 it always has a difference based on a couple of different factors and we'll touch on those factors in a little bit but the main thing with XMP is XMP is whatever is beyond that set specification so if you went and bought a kit of 52 or 56 or 64 these are all XMP kits right and they all carry some other value points that ultimately allow them to operate at that I like that I'd like to take a little bit of a step back to to actually talk about ramp because a lot of people may not exactly know what is RAM and what is RAM stick what is XMP and what does it stand for and how is this utilized in a PC so could you maybe give an overall overview of what is RAM how how what is that part of the you know P6 we know CPU is Computing unit right we have a GPU Graphics unit what is RAM yeah so I mean insured it's pretty much it's your Random Access Memory um you want to just think about it that it's all about kind of the proximity of how the system works with the components it has available to it right and so right here you know my system next to me or of course in the laptop that I'm running this from right they every system we have I was always going to have kind of this hierarchy of how it has essentially a storage space to be able to run applications for rent on the CPU the closest form of storage is going to be cash right and then the next closest form is going to be the system memory or the the random access memory and that's usually what's referred to in like your configuration of like 16 gigabytes or 32 gigabytes 64 gigabytes and then you've got your SSD right those are your three generally points of storage in your system the ram though um it is generally going to be the most dense in terms of letting us run applications because cash usually is only literally in the megabytes you don't get gigabytes of cash yet on CPUs so um you know applications have to be really really sensitive about how they load and they're working with cash and once they're no longer able to do work in cash they're working inside of our applications and I always try to visualize this from the perspective you know I think a really easy way to look at this is probably something that's like literally opening up your task manager right you open up your task manager you're able to visualize your memory right you can literally see right now in my system I'm using 9.1 gigabytes worth of memory I can see that there's a footprint that's on there I can see my speed I can see that corresponding information right and that that's what its job is at the end of the day whatever you're running whether they're you know it's Adobe Premiere or whether it's your browser or whether it's you know a streaming app doesn't matter it's all going to be running inside of system memory so now we know when the XMP is on and you know we've got the ram we know kind of what the ram is could you explain us um what is the different parts that can affect the XMP profile of the ram so you know we've got a ramp kit that is saying 5600 Mega transfers per second and also what is the difference between Mega transfers per second and megahertz per second yeah so this ultimately is just um this one's a kind of a more kind of idiosyncratic kind of Point within the industry it pretty much just comes down to how you decide you want to define the operating metrics for the memory so the correct term right is generally going to be Mega transfers per second right as opposed to megahertz and this just has to do with actually the bandwidth versus let's say the clock Cycles right and so the easiest way again if you kind of bring up an application is you know if you bring up a utility a commonly people are looking at something and maybe like cpu-z or I guess that I should there in the task manager you can actually also bring up the information if you're looking at it from a frequency perspective you have to add a kind of a multiplier so you have to multiply something so an example of this would be like 1600 and then you would multiply that times two because you have a double data rate or DDR right is memory you multiply that by two and then you would end with your effective output and for a long time that's kind of what people were kind of defaulting to but ideally really when we talk about memory the end goal of what the bandwidth is the Mega transfers is what we want to try to actually convey and so so with the ddr5 now that we have a new specification that's what's effectively trying to be communicated at this point prior question to that right though in terms in terms of what really kind of influences memory that one's a really interesting one because a lot of the times people will again generalize a kind of certain assumptions on what they think is going to assure them a certain level of performance and a great example this is let's say if I could hear so I've got to just get a memory it doesn't matter you know whether we're talking you know uh g-skill Kingston Patriot crucial any one of the great memory manufacturers that are out there this is going to be applying unilaterally but they'll have you know a little label on there it'll be some type of XMP profile and they'll kind of think in their minds uh you know I right they go okay I think you said 5600 um it's 5600 Mt my motherboard I look on the motherboard the spec sheet says 5600 Mt and then I'll even maybe check the qvl and it says 5600 Mt so it should all work right actually no and that's because we actually have to look at a couple of different things that ultimately influence what allows the memory to effectively run so the first one is going to be the memory itself right so when we talk about the memory there are going to be a few things one is going to be the ICS so those are the chips that are literally on the memory these can come from different companies so there's generally going to be three companies at this point you have Micron Samsung and SK High necks they're going to make your chips then another part of this is gonna you said I see what does icic stand for integrated circuit but for most people it's easiest to kind of just refer to it as you know those memory memory chips right you're then also going to have other parameters that influence that including even the PCB layer count just like in motherboards there's layers that exist and the more layers you have generally allows for better signal Integrity performance because kind of like when you're routing these paths if you make the layers less right so if you go from like a six layer to an eight layer to a 10 layer um it can be more complicated to optimally route a short path which would give you better signal quality and just kind of think about it like if you're driving right if you had to kind of go through the back roads as opposed to kind of like an immediate uh you know path that you could just drive your car right around something and you'd be you know where you need to get to right um some memory might be eight layers some memory might be kind of 10 layer but the the end point is ultimately is with that memory it's pre-checked right so you're like okay hey it should work the tricky part then is if you move to the next part would be the motherboard the motherboard influences things it also similarly you can have a different layer count it can have what's called a topology so the way that the actual traces are laid out so the traces just are essentially lines that are on the motherboard that are routing information from the memory to the CPU um and that that can heavily influence things and then there's also the firmware or what some people refer to as the UEFI firmware or some people still call the BIOS to that is the software experience and that also determines memory um the memory performance compatibility the last part that people forget about though is the one that we talked about in the very beginning which will be the IMC the integrated memory controller so you can essentially pre-vet the memory itself right because you know let's say the memory manufacturer has done that for you they pre-checked it they are able to confirm that it runs at that speed uh the motherboard manufacturer like us were able to pre-check and design our board to be able to be confident that it can run at that speed but what you can't pre-check is going to be that CPU that CPU is always going to have some variance in terms of the values that it can get to now that doesn't mean that when you pick the memory that there isn't memory that might be a little bit of a better Target to get to but it does mean that there's more variance so essentially there's kind of three main things and then there's a lot of little subsets within those that influence the ability for that memory to run so when we are looking at let's say the same kit of RAM and then let's say this is 6000 MTS right and then we're using this on our system and we're not getting stable performance let's say in one CPU but we get exactly the same CPU same model and then it works on the second CPU can you explain us why this might be and what is a silicon Lottery and does it also affect the eyes integrated memory control of the CPU yeah so that's actually exactly what happens so it you know great way to kind of evaluate this is if you know you kind of took 10 CPUs so if you as a user if you could actually buy 10 of them you would probably run into that if you checked each one even though all of them if we can go back to that poor table we know they all can run let's say at 4 800 or in the case of the you know 13th gen that might be 5600 Mt right or on AMD you know again the rate is memory specification if you took 10 CPUs and you put them all together they could all guarantee to run at that minimum base spec beat but what you would start to notice is a delineation between one and another if you start going at faster speeds so to your point like six thousand uh we saw take for instance like in 12th gen that probably approximately about 7 75 percent maybe almost 80 percent of CPUs could hit a 6000 Mt memory divider in terms of stability right and so this doesn't just mean posting it means post Boot and use so there's always three levels to how memory can work and you and you can't kind of um forget about one because just because your system could post at a speed doesn't mean that it's going to boot at that speed and when it boots at that speed doesn't even mean that it'll be stable at that speed so you actually always have to account for all three but to my point in terms of the percentage about 75 to almost 80 percent were able to go from that 4 800. to about 6 000 right and then um when we went to uh Raptor Laker 13th gen that 6 000 actually went all the way up to about seven thousand plus so it was a big jump up in terms of the amount of kind of consistent CPUs that could overclock to that speed right but there will be CPUs that will be under that margin and there will be CPUs that are over that margin and the you know a lot of people talk about it it's the Silicon Lottery uh we terminate silicon variance maybe sounds a little bit uh more PR friendly right but that's what it is it's just variance between parts right some parts can run a little bit more and some parts can run a little bit less and we see the same thing within CPU frequency a lot of people understand that when they like buy a CPU if they were to overclock it not all the CPUs are the same right some will overclock maybe to like that six gigahertz and some might overclock only like 5.8 gigahertz right and so that's just an again it's an inherent difference you mentioned Intel here but what about AMD in their memory controller where did they slow in in the 12th to the 13th gen yeah so um of course they're a little bit newer in terms of Now ddr5 intel has now had two generations AMD only has now one generation in terms of ddr5 memory and they're also still making enhancements both companies actually do release what is called MRC or a memory reference code which helps to influence say like the interoperability and compatibility for memory that a motherboard manufacturer has to work with along with their own designs for AMD if we actually bring up their table we can take a look at this and this is an interesting point as well because when you bring up the value a lot times for AMD there's been discussion on like everybody going oh like 66 excuse me 6000 Mt is the sweet spot for the memory and that's what I should buy I don't necessarily like using that term and the reason why is I don't think you can have a sweet spot when it is not the guaranteed operating spec to me sweet spot has to mean that you have to be able to run so if we took a look here at this um uh table here from AMD we can see that their base speed is 5200 right but can actually go all the way down to 3600 depending on the rank the density and the population of the memory now again that doesn't mean that just like when we talked about on the Intel side you can't get uh CPUs like a 7700x or 7900x to be able to run at 6000 Mt but there will be variants and we see right now probably that percentage is fairly similar I'd probably say it's maybe around 70 percent 75 percent in that range that most um most ryzen series CPUs essentially will be able to run at 6000 Mt but the critical part to that is just like the intel if you noticed is the slots per channel the dims per channel are or what people generally think of when they talk about memories is like two sticks of memory versus four sticks of memory two sticks of memory is always going to be much higher in terms of its scaling potential and four dims is always going to be lower in terms of its scaling potential right so that just means that that number that I talked about uh that you know 70 or 80 marker that's in that uh that's only in that two dim configuration not in that four dim configuration so there is I think also misconception of sometimes people don't understand what means four dim sticks and four channels and dual Channel and four sticks in there so you mentioned this already that when we have four sticks on the motherboard because a lot of motherboards these days come with four sticks and you think look I've got four channels but that is not exactly true as you um already know could you explain why there are four dim sticks or dim slots on some of the motherboards and then most of the mainstream CPUs are actually dual Channel and how does this actually work and why why does it go down in terms of the memory supported frequency or XMP or clock speeds when we're entering to that four sticks so I think that the first one right there we want to tackle there is just you know the point in terms of kind of the Dual Channel versus like let's say quad Channel and that's ultimately defined by the again the IMC so your CPU so your platform all mainstream platforms that are available from Intel and from AMD are all dual Channel there's no quad channel so you only go into quad Channel type memory specification when you're going into the high-end Enthusiast desktop so for Intel that would be like the x299 chipset which will be replaced in the not too distant future and then on AMD side right that would be threadripper threadripper Pro based platforms right so like trx40 wrx80 right you go into more memory channels but for all the mainstream series users those are all going to be dual channel so that's the first part that kind of uh just keep that in mind right is that's influenced by the CPU now as far as the dim slots right why do you have two versus having four technically there's no uh like I'd say uh formalized reason as to why you have to have four right uh we as a memory mother as a motherboard manufacturer can produce boards that only have two slots like our uh z790 Apex board or take for instance or like a mini ITX board um although generally on Mini ITX you're doing it because you're more constraint just in terms of physically space four I think actually became the reason why you saw it so common and why it's even maintained its place within the industry is because of memory density and what I mean by that is literally the amount of memory that you have so as a user right when you're buying memory that's the way you think about it you think about it in terms of density or the capacity right so 16 gigabytes 32 gigabytes 64 gigabytes well not that long ago even to get to like 32 or like 64. it might have been that you had to go with like a four dim configuration right and even years back even an affordium configuration you could have not even got 64. you you would have had to go to like a high-end Enthusiast desktop and had to have like a motherboard that had like six or eight dim slots to even be able to get to 64. ddr5 has made this quite interesting because the density for ddr5 is really moving up quite quickly but that's part of its benefit as a new specification right right now we're moving into ddr5 can give you 64 gigs and two dims and a lot too distant future you'll be able to get 128 gigabytes and two dims right um so the value proposition is going down a little bit in terms of why you supposedly have to have four but there is a reality that has happened over the last few years that people love Aesthetics in their PCS when they're building them and so a lot of people like the aesthetic of a Ford in population so while it doesn't necessarily offer a technical Merit outside of density right you can still have of course more memory with four slots than you can have with two slots what we see most people actually adopting four dims is not generally because of density now you you a lot of your users a lot of your viewers they are creators so they might maybe be in configurations where they maybe need 128 but to be honest even for uh definitely in photography I'd almost say it's almost no scenario would need 128 for photography really only 3D animation professional videography uncompressed workflows we need that density and memory but that's really what it breaks down to right is that if you need the density you would go with four but if you want really I think good density but High scaling you definitely want to go with two and the last Point excuse me I think the last question was why is it harder this one gets a little bit tricky without getting to too much of the minutia but an easier way to I think about it is kind of like weights right if I was to give you um two five pound weights right or versus two ten pound weights versus two 40 pound weights which one's going to be easier for you to lift right the one that's lighter right and that's pretty much what you're doing is that you're asking the memory controller to address and to hold a lot more memory at a faster speed and that's much more strenuous so density will also affect that value so even when we talked about two dims the that still plays a part so if you go from like a right now the best performing kits if you talk about scaling are going to be a 32 gigabyte kit that's where you like you can go like 6066 7076 almost even 8 000 uh with with really good boards and really good CPUs but that's not going to be with the 64 gigabyte kit the 64 gigabyte kit guess what that's going to come down because you added in more stress to the memory controller if you want to reach let's say 64 gigabytes right because this is like the sweet spot where you could easily get it with two sticks or four sticks of RAM and a lot of people like the Aesthetics of having four populators like you have there on the desk you've got beautiful RGB of the four sticks you don't have any teeth missing and if they want to reach 64 gigabytes for example is it easier to reach that XMP of x amount let's say 6000 MTS with two sticks of Ram or four sticks of ram yeah so it's always going to be still easier with two right so the you know what we try to communicate to a lot of people in the community is ultimately going to be the rank density and population anytime any one of those go up you're always going to be going down right so if your density goes up your your speed is going to go down if your population goes up it goes down and you can kind of pull those in different scenarios right so if your population doesn't go up but your density uh goes up right it's still going to go down but it will still scale higher than if it was both Rank and excuse me if it was both density and population right which would be you know you're going to like a good example that would be like 128 gigabytes right which right now you still couldn't only do with a 4-dim configuration right but yes to your point it's always going to be easier in a two-dim configuration even if it's at a higher density than it would be an afford him now there is a little bit of motherboard uh what's called topology that can influence this all General kind of most motherboards right now that are produced on the market um you're going to have What's called the daisy chain topology as opposed to a t topology design and this is again gets back to that Trace layout Design One favors a four-dim configuration and one favors a two-dim configuration but if you kind of think about what the the benefit of ddr5 is right is really this impressive level of bandwidth that can be supplied for the memory it doesn't necessarily make sense to prioritize for dims right because we know that ddr5 can offer so much more bandwidth and the best way to get to that bandwidth is to do it through two dims so as a motherboard manufacturer and actually really it's not just Asus all motherboard manufacturers we are prioritizing a two essentially kind of a two dim design even on four dim motherboards because that's what allows us to get the better scaling right and you will still see a difference between like a two dim board and a four dim board so if you took like a current uh like this is an extreme board or hero board but it doesn't really matter price price point does affect it a little bit in terms of some elements in terms of the design but uh you can do take for instance like on z790 most z790 uh Asus motherboards will be able to comfortably with a good IMC like I said 7000 is the minimum and many will do about 74 to about 7 600 with two dimms uh when you go into then like something like an apex board which might only have two slots because there's what's called less signal interference it's an optimized topology where it doesn't have to do with those four slots it only has two slots you can take that same CPU and probably almost scale that to almost about you know 78 to almost 8 000 so even still get more margin uh within that so if you really really care about that really high level of scaling you still have to always account for that silicon variance and that CPU right and there's also some silicon variants and Boards not you could take 10 motherboards and there's going to be a little bit of variance but the variance generally between boards able to hit a certain value tends to be not as great as the variance between the memory controllers right between those same set of CPUs and you mentioned dim capacity or density and then you know obviously the amount of sticks you have but he also mentioned the rank um you know memory rank which is not often mentioned on specs of the ram kit so when you are buying the ram kit it's not somewhere in the front there you might be finding it somewhere in the you know fine print but is this something that people need to understand especially creators shopping for a ram do they need to know what's the rank of the dim sticks they're buying yeah 100 because again if we go back to that poor table um so we go again back to either the Intel or the AMD it doesn't matter whether we take a look at amds or Intel's again there but that configuration right there when you see that two by one rank that's when you have your greatest speed the moment you've noticed even if you uh if you you change your rank there right so four by one still would give you then 36 and 4x2 rank you can see goes down to 36 right and then uh if I bring over that Intel portable you can see how in the prior generation right it was also similar that even if you were keeping one rank but you were changing kind of slots per channel right you were dropping uh you're you're dropping down right so it is important definitely to keep in rank rank is probably the most immediate way for performance to drop down even more than a density but they're not always together but generally they're together so generally the more dense the memory you increase the likelihood of going to a higher rank in terms of your memory um it's kind of hard because some people would go like how can I look at it well a lot of times you can't see the memory module because it's underneath the heatsink and some people would think oh if there's like chips that are on both sides it's going to be dual Rank and that's not correct you can actually have chips that are on both sides and it still can be signal rank so really the only good way to do it is you have to actually get the information from the manufacturer you got to look in the specs and find out is it a single rank kit or it is a dual rank or it could even be potentially a quad rank kit but that would be um like I said pretty rare unless it was like a really really high density this was 12th gen that you just showed us but when we come to the 13th gen and I believe that there is whether you're a single or dual rank it's still 5600 Mt right actually I don't believe that for dual rank um not for 5600 for 5600 it's just it should be only for single rank configurations but this is kind of the strength that um Intel has uh in terms of their memory controller and the memory code and then also working with a good um like I said board and firmware is that you can always massage essentially how much you're able to kind of tune out of the performance right um and and how much margin exists there like I said it's already quite impressive that in both Generations whether it was from that 4 800 or on now on Raptor Lake right 13th gen 56 that literally you're you're having you know the ability to be talking about that you can run speeds you know of uh 66 72 7600 speeds that's that's we've never in the history of memory seen that type of scaling within uh memory Generations it has never happened before um so it's really really impressive but at the same time that's The Balancing Act Right the speeds have literally gone so fast just for rough difference keep in mind right right now we hold the world record for DDR memory overclocking it's over 10 000 Mt we this has been achieved in an ultra short period of time you literally didn't even double you went over double what the initial introduction was for memory speed on the on the standard right in a really compressed amount of time it's exciting because as enthusiasts this means tons and tons of bandwidth is being given to you right um but then the challenge comes into is that you're talking about these extremities and and what ultimately uh is going to affect your stability when you're trying to get to some of these extremities could you simply explain what a single Rank and what is dual rank Ram if you if you see one sometimes even it's the same Ram provider or make that provides you single and dual rank so what what does it mean what is single and dual uh rankings yeah so it has to do with essentially the bits uh that are on uh the memory itself right and so again there's no easy way for the user to know this information but some essentially it has to like I said do with the amount of addressing that's happening right and so when you're increasing the ranks it's how the information is being accessed and so the more ranks that you have the more simultaneous access to those bits it's again maybe think about it like being in an environment where you had multiple people talking right at the same exact time um if I had maybe one person there pretty easy like me and you right we can have a clean easy conversation even if probably both of us are not talking at the same time but the moment we probably brought in somebody else into this conversation and you have them start talking at the same time guess what your brain starts to probably get a little bit like wait what did he say and what did she say I'm getting it just it's getting a little bit more complicated for me to understand this whole conversation um and so that's probably the easiest way to it to address it but again probably the best way for users just to try to find out whether it affects them is going to be checking you know uh with the memory manufacturer and see if it lists that information for the most part though um rank isn't going to be as critical as density because still a lot of what you're seeing out there is going to be single rank memory so it's it's not like you know you have to be super super concerned about this it's more generally the challenge of like I said also the the population and the density being a factor so again even if you got four dims and they were all single rank right but you were increasing that density or increasing that population you're still going to be bringing down your scaling so remember doesn't matter whether you talk about rank whether you talk about density or whether you talk about population anytime you bring up one of those three your scaling is going to be coming down is there a difference between AMD and Intel CPU memory controllers you mentioned that already that one has maybe like AMD is just the new dog in the game and it's the first generation of ddr5 memory controllers Intel with their 13th gen is newer you know of second generation is there anything else that's different between those memory controllers there is uh and it gets kind of tricky because it depends on how you want to do Performance Tuning um so you know with AMD you have something that's called The Infinity Fabric and uh kind of how that works in relation to feeding memory from the CPU um you know in the past you I generally want to have kind of this what's called one-to-one kind of value so you want to try to kind of match the memory to be able to running at a certain gear or certain value in relation to kind of what you had in turn into your Infinity fabric Intel actually also has something that's similar to this it's called the gear ratio and generally you have kind of a gear one or Gear 2 but ddr2 pretty much is always running in a Gear 2 speed I think for most users though these these things don't really play A Part um I think the way that most users are ultimately purchasing their memory they're ultimately just trying to figure out what are the effective kind of ranges that I can consider when I'm looking at one platform at the other right and so I think when you're looking at AMD again remember their Baseline it's 5600 anything over that is an overclock and it's not guaranteed right like I said out of that 10 CP samples you could have uh some CPUs that wouldn't even be able to hit maybe 6000 Mt so you would have to go down to a slower speed and I can show you actually how you do that it's called dram divider testing right when you want to maybe actually test your scaling speeds over that are going to be possible on AMD but they won't be as consistently possible as they are going to be on Intel as I noted right now on 13th in the really the Baseline for the vast majority of CPUs is going to be 7000 Mt plus so I say that quite conservatively that about you know 80 percent of CPUs are going to be able to do 7000 Mt and it's not uncommon for them to be in the 72 74 to even up to about 7 600 range and that would even be on a uh four dim motherboard not even a two dim based motherboard you don't generally see these type of memory speeds on the AMD platform AMD platform you're generally going to be targeting a lower value in terms of that speed but it kind of works out right because the sweet spot right now in terms of memory density and and speed in the market is I would say a 6 000 to about 6400 and then depending on your timings that's kind of The Sweet Spot the the then the higher the speed or the higher density you start to get into more expensive kits and definitely if you're if you're somebody in super performance minded then you know it's it might be worth you know allocating more your budget for that but you kind of can see that for for both platforms the memory that is most commonly being considered would generally be able to work on both of them why are some Ram kits and why some Ram kits have better XMP stability than others so you mentioned that this is wow this is a really really good question um so go on yeah so uh different manufacturers might kind of like phrase this a little bit differently but you kind of have to think about it that again you're talking about an overclock profile right and that profile uh has to be defined and you kind of it can be quite tricky for them to Define that profile and have it work for a lot the largest amount of CPUs and even the largest amount of motherboards because some other boards might have maybe have like a six layer PCB some other boards might have like a 10 layer PCB some other boards might have maybe more uh optimized firmware code right for your memory management there can be a lot of kind of discrepancies and this this doesn't necessarily mean that the XMP profile is the Andor or be all best option in some situations it should be kind of more of like this kind of Base starting value essentially and you may have to potentially maybe add a little bit of voltage or you might have to adjust the timing to actually have it become stable on your system and this kind of goes against the grain a little bit of what people think about X and P where by default they think should always be that I can just drop it in and have it work so it can be quite tricky for the memory manufacturers perspective because you have to account uh for essentially that memory has to work on potentially hundreds of different motherboard configurations and so there's some that might be a little bit better there's something they'd be quite a bit better and there's some that might be a little bit worse so you kind of almost have to tune kind of for the lowest common denominator and so some memory manufacturers might decide to maybe be a little bit more aggressive on their profile than others right to maybe give you a little bit better performance some memory manufacturers might have a 10 layer PCB be designed versus an eight layer PCB design and that actually means that the signal Integrity would be a little bit more tight and actually a little bit cleaner and that can actually provide a little bit higher level of stability at certain values so it can be kind of a tricky in terms of that and there's also some information that is important to understand if you check certain manufacturer's websites they will even detail to you the information that we're talking about here it's not always a parent on the page but there will be a portion on the page that you'll be able to bring up and it might show you that hey this XMP profile is dependent on the quality of the CPU and motherboard combination and what they're pretty much telling you there is saying is that even though this kit is rated for this speed it can't necessarily be guaranteed to run at this speed because of some of those factors that we talked about right but the last point that I think I would make there in terms of that XMP profiling is just because you can't hit the XMP profiling doesn't mean that the XMP profile doesn't have value um it has been pre-screened by that memory manufacturer to be able to at least know that is able to run at or around that speed with that similar kind of voltage in that similar set of timings and that still does have value because otherwise if you didn't have that information you would have to do it all manually right you'd have to be buying a kit and you'd have to go okay if I don't know how to set the divider speed if I don't know how to set my timings if I don't know how to set my voltage you'd have to know how to go in there and dial in all those values you mentioned about massaging the memory with a bit of More Voltage and the less voltage sometimes I know this is a little bit more like overclocking side for people who are more enthusiasts than are probably not for creators but I think it's still interesting to understand what does it mean you mentioned that some of the IC manufacturers there are three main ones Samsung Micron and SK High Nix and tell us like what's the difference between these because actually we don't see a lot of these Ram providers out there because there's there's not like a Samsung mainstream you know XMP Ram kit right yeah we have Corsair we have G scale we have all sorts of you know Kingston all these different manufacturers looking at your memory right so you know like yeah so here you know I've got Corsair or right here I might have my G skill or here or my exactly Patriot or you know the one right here next to me Kingston doesn't matter that's the way that we think about our memory and even when we go to the manufacturer's websites they actually generally will not list the IC provider um and there's a couple of reasons to this one is the IC providers can actually change depending on sourcing Demand right so or production right so sometimes um you can actually start with one production of an IC and then it can change into another production of an IC right so this has actually happened even with SK High next right where they had MDI and then they went over to like a die and they start to have different properties one might be able to scale it to higher frequencies at maybe tighter at lower voltages than the prior Generation Um even within the same uh company right so that was SK to SK right um and the reason why this is also again important is the other big No-No of memory which is don't try to mix and match modules um so you know we I don't think this has been brought up yet but for one there is no four dim XMP kit it doesn't exist on the market you cannot go buy a Ford dim kit so everything for ddf5 yes for ddr5 correct for ddr4 you can definitely buy them um uh but for ddr5 you can't buy a four dim kit it doesn't exist so you're always buying essentially two kits of memory right and this is where the false presumption kind of comes into play is that people go there's XMP for four dims right because I can still go into the motherboard right like so right now if I went in here and I enabled it I still see a value um but you have to remember is that that XMP value was defined and validated for that two dim kit so for that specific ICS at that specific timings at that specific voltage now does that mean that it can't potentially work no it can potentially work but there's also things that you don't even see that happen behind the scenes from the motherboard manufacturer that we might have to change other little sub values that actually aren't even part of the XMP profile to get it to work so we consider this almost like a massaged form of XMP because they're having to be kind of adjusted to be able to reach essentially that operating parameter and this all falls under something that we call auto rules for any mother brand manufacturers they're Auto roles these are Auto adjustments that we're doing to attempt to try to streamline or improve upon the experience for you but it doesn't change all those kind of other factors that we talked about another tricky part to this conversation too when you talk about those ICS and kind of where they might kind of benefit one versus the other is if we gave like a reference to like Micron Micron makes outstanding memory very very good quality in terms of its stability and reliability but generally it was lower scaling than you might have saw in uh seen for let's say Samsung or SK High necks so most of uh microns kits were usually like 4 800 to about like 5600 Mt but you saw Samsung kits definitely doing all the way up to like six thousand and then you saw SK hynix being a lot of the memory that was generally in that like 62 64 6600 range so an example of this is I have a kit of memory here it's actually an a data it's an xpg kit and it has two different ICS from two different production points now why would that be important well think about it if one set of dimms were designed and they were tuned with certain profile in a certain timing range right um they kind of want to run at a little bit a different level than another level right and so maybe you buy a dim a year apart and you think it's the same exact dim it might even be rated by the manufacturer exactly the same right but the inherent kind of range or tolerances that it might have can be different so you could have this one that could be Micron and you could have this one that could be Samsung and you've all of a sudden mixed them right now if you bring those values down more conservatively it's probably easier for them to kind of work together I kind of think about this like like maybe like maybe like a handoff you know like in track and field like in the Olympics you know where you gotta pass the Baton right you know it becomes a really good Balancing Act of people knowing how to pass baton but you know a new team that maybe hasn't worked together that well what happens sometimes is they fumble when they're passing that baton right and just the experience it doesn't work out that well and that's kind of actually similar to when you mix those kits if you're mixing maybe one IC with another IC you're probably not going to get that great of experience ideally if you are going to mix dims you would want to match the ICS and uh you can show this but we thankfully at least offer in our UEFI our UEFI firmware we do give you the ability that you can see the actual SPD information and the and the actual vendor so you can actually know like who who makes my memory is it you know is it well you have a memory manufacturer but who makes the ICS right uh whether it's Michael SK hynix or Samsung if we're talking about um Samsung SK hidex Micron and so on you mentioned that there is a bit of uh difference in there what are some of the difference that you have seen as a motherboard manufacturer and that why what are some of the better ones and what are some of the lower ones if someone's looking for a look I can see SK High next here I can see Samsung here I can see a micron here which one is better um it kind of comes down to depending on I guess you know what your what your goal is Right Micron I think has really been their memory to a very outstanding yield and consistency and quality at their kind of Target brand this allows them to be very I think price aggressive which is like if you go to like crucial or their website you can buy that memory but they're not necessarily maybe tuning it for a high level of clocking right whereas let's say maybe the SK side they're going like we're tuning it to allow for even greater scaling so if the memory uh manufacturer wants to work with us and Bin these so sort these in the facility add more voltage we kind of give you this room that if you're willing to add more voltage produce more heat consume more power right we can give you even greater scaling right so SK hynix is probably right now the the leader in terms of giving you that fastest speed possible in terms of kind of kits that are available but you can get still like I said very very performant kits of Micron based memory right and very very performant kits of Samsung based memory but most of the highest speed kits that you're currently going to see on the market right now and keep in mind if you guys watch this you know a year down from though this changes because different companies are producing different ICS right so like I said right now SK hanix is just transitioning from a higher a high performing IC that was called MDI now you've going to your a die right this could change again Samsung could maybe come out with a new revision and that new revision all of a sudden starts to have really really great scaling right um and uh you know but for you as a user you generally don't have to worry about this too much because that's the Java kind of the memory manufacturers them trying to find the best IC to the Target that they're trying to provide to you as a user and that's usually going to be that that that empty number along with the timings and then along with the voltage right that's going to be you know what influences those kind of three values yeah and I think for viewers it's important to know that the most likely the average user who just applies XMP don't really need to worry about the IC providers and even if they change it's still it's completely fine this happens in the all over tech industry everyone does it it's not just memory where you see it in ssds we see it in sometimes gpus we see it and apple does it on their iPhones it happens everywhere even if you're it's just yeah it's a part of Supply Chain management right it can be uh you know for a lot of different reasons maybe you go to a new production process and you're going to change over there's going to be a lot of different reasons so this isn't like I said a negative factor uh it's just part of the reality of just you know electron mix production right let's say that you um you got yourself the ram kit now and you've found out that for some reason your RAM is not stable right why are some of the reasons that your RAM is not stable and then we're going to talk about some of the tips you have for people to improve or get the stability perhaps in some of the Asus motherboards what are some of the tools you have in BIOS but first of all what are some of the reasons you might not have the stable XMP like in your mind what's the checklist we've already like kind of covered different aspects of it already but just if someone's thinking over so yeah the first the first one is it could just be that you have reached kind of the um I'd say that that's scaling for the CPU right and so remember if we talked about like out of 10 CPUs you know 10 of those CPUs would be able to hit like 6000 Mt or now like 7000 Mt it could just be one you're at that little bit of a more of a margin so you're just kind of pushing the limit right um and that also would translate into the density right maybe it's a little bit too high of a density and also at that speed and that's important that when you make that kind of memory purchase you kind of have to be prepared that you can either have a higher probability of success or a lower probability of success right it would be great to see that when you go to the memory manufacturer's page or you go to you know your e-tailer to kind of know is this what's the kind of the realm of likelihood of success right but that's kind of what you have to think about um as far as really being kind of the determining Factor right um and this also assumes that you're running even two dims right because the moment if you did four dimms we know that that is already going to be much more difficult than it was running um excuse me uh four versus two right now are there some things that could potentially also be holding you back sure maybe one of them could be that the voltage is maybe a little bit too low um so that could mean maybe hey I have to increase the voltage a little bit uh for the and memory controller itself right but most of the times you also have to balance out the fact that if you're going to be increasing voltages you're also going to be adding more power consumption you're going to be producing more heat right and even potentially adding more degradation uh to your CPU right which is not something that you necessarily want to do so even though you might be able to bump it a little bit more the question comes in is should you be bumping a little bit more right so that is also something to account for um the other one could be is that the timings right keep in mind that depending on how that memory profile was designed could be maybe some of those timings are maybe a little bit too tight and so maybe they might have to just be adjusted a little bit um right that means so let's say maybe I bought a kit of like uh you know again it's like a G's code kid it could be like a c28 kit well maybe that I'm gonna have to tighten that excuse me instead of having to be so tight maybe I have to loosen it up just a little bit maybe go to like C30 maybe C32 versus c28 right um so that could be something that could be also influencing it but in most situations um it's generally going to be like I said a limiting factor there between either the the margin that you have between the CPU so the IMC or the margin that you might have also on the motherboard side because that can't exist a little bit um in terms of that and then the last one would be the density right which is you know whether it's the densities achieved by four dims or whether it's just because you went with 64 versus a 32 gigabyte kit I'm thinking from a crater's point of view perhaps maybe someone who's not just already built their system but they already have their parts list right and they want to make sure that look I want to make sure that my memory is actually stable and my system runs stable so I'd say probably probably look at the CPU you look at your memory controller of the CPU then you look at the qvl list could you explain what is a qvl list of yeah so yeah so the qvl list is kind of again it's a really tricky one because I see a lot of people they always make this recommendation they're like oh just buy memory that on the qvl and this will assure that it'll run and actually it doesn't and I know people might find that surprising because they're like but wait you hear the motherboard manufacturer what do you mean the cube the whole point of the qvl no the qvl you want to think about it is almost like a proof of concept but it's not just it's it is a real proof of concept because we actually have to run it at that speed the difference though is again if we take what are the three things that at least on the highest level influence memory speed it's the CPU the kit of memory right and the motherboard right well on the qvl what could we 100 guarantee we can guarantee that kit and we can generally pretty be pretty consistent by consisting the motherboard but we can't confirm or control the quality of that CPU which is what you have as the user and you're putting in right and so this is where a lot of users get tripped up because they're like qvl says it my motherboard says this on the spec page that it can support this right and I know in general just Intel supports this CPU frequency so why can't I get there well guess what that IMC which has variance that's the part that you can't so that's the part that you kind of always just have to account for and that's why um you know we we at least do you know like I try to put out there in the community we try to put out these overclocking insights or kind of guides or message points to say Hey you know if you're picking your memory uh like now for like 13th gen I'll tell most people hey if you go up to 7 000 with a 32 gigabyte kit of memory you've got about an 80 likelihood of success the moment you jumped into like 74 or 76 or 78 guess what that percentage starts coming down right so you have to be prepared for a higher likelihood that even though on paper right and even to a degree validation from the qvl right the qualified vendors list they should work there's still some variants within that and that's going to bring down that probability of success a lot of the motherboards support different generations of CPUs which also have different memory controllers of generations as well right so yeah for like z690 could be an example of that where you know z690 you were able to put like 12th gen or 13th gen and also vice versa you could run a 12th gen and a z790 based motherboard right and this will also happen for am5 at some point when even newer am5 CPUs come out and you can run that like on x670 or like b650 or something like that um and there is a difference because some people go like z690 is there any difference between z790 and there actually is um we were able to like from generation to generation improve what's called signal reflection which is again it's just saying that the overall noise when anything is electrically talking to each other there's noise that gets generated and we're able to tighten that up make it a little bit finer and so an example of this is uh z790 on average like from from board to board even with like the same CPU you'll generally get a little bit more margin within z790 than you would with NZ 690 but z690 differences with a 13 gem part in there I can still run 7000 Mt right but I have to bring along that um 13th gen because that IMC is what's really driving a large part of that the board imparts a part to it right but if I would have put 12th gen in there maybe the most that I might have seen out of it might have been you know 60 62 or 6400 right but the moment that I put in that 13th Cent I'd be able to get in there but then maybe I take that same 13th gen and I put on z790 and now maybe that Peak number might all of a sudden maybe be closer to like 72 74 7600 so you can see that on average I would say it's somewhere between about like 200 to about like 400 plus megahertz bump um between z690 dz790 that's very very interesting because I was going to ask you let's say if we have a mother how much does the motherboard actually affect the memory speeds as well let's say we have a 5600 MTS kit which is quite you know conservative if we took like uh 13 700k and i7 um a z790 motherboard z690 motherboard b660 b760 for example maybe had some h670 for example or something like that because some of the motherboards are lower and motherboards Intel has locked or motherboards have have been locked X and B so XMP doesn't work from them but let's say all the ones that XMP is allowed on all of them um is there how much of a difference there is because some of the people think look I'm going to save a bit of cash go with the b660 motherboard and you know 13 700k for example does that affect as much of the XMP yeah so definitely for most most situations generally you wouldn't usually run ddr5 in gear one you you know DDR uh ddr4 on Z on Z series chipset you could definitely consider running in gear one but um when we talk about kind of the board influencing it um whether you're talking about kind of more entry level boards so like say for instance us we've got like you know our Prime Dash a right versus something like here I've got the extreme right it's a big price brand Delta right you know like here you might be talking about like you know 250 299 motherboard versus like a thousand dollar plus motherboard right how much of the difference are you gonna see because sometimes people just say overclocking is better not that much really uh the you know I was actually like in the prior generation I showed hey from the prime motherboard all the way to this High board all of them could do you know 6000 Mt or all of them were able to do 7000 Mt so really there's not that much where there is a little bit of influence that does occur is that higher end boards that will have a higher what's called layer count right um maybe going uh to like I said like eight layer or like a 10 layer motherboard because essentially when you have more layers you can more optimally route the signal paths to have less noise and overall just a better type of transmission experience that can give you a little bit of a margin or a little bit more of an edge in terms of stability especially at higher values but to your point like in those lower values if we're talking about like 5600 you know um 6000 Mt 62 64. it's not really that much it can improve it right so you will have higher success rate on that but it's not a large degree the CPU was still much more of an influencing Factor there so the quality of that CPU is always going to be a much higher determiner of the overall quality of the experience you're going to have so if you took that really good CPU and you put it into a 10 layer motherboard versus you know an eight layer motherboard versus like a six layer motherboard it just keeps getting better because that CPU is still that big one right um those just help to kind of give you a little bit more margin but that's the main part and if somebody does want to save money and they want to go with like a B series chipset right so like b760 just you know came out the beginning of the year that's fine you can still get very very good ddr5 overclocking experience on there as always though do keep in mind it gets forgotten about because people kind of I think understand it for CPU but they forget about it for Dram and this is you know Intel and AMD right technically uh dram overclocking is also voiding your warranty uh so you do have to be mindful of um you know that Balancing Act Right of performance to everything and this gets kind of confusing because you buy the XMP kit and it's like it's all validated and that has a warranty and your motherboard has a warranty but remember your CPU also has a warranty and there are guidelines that are noted uh for the CPU resident to those guidelines tell me more about that because last time we had this conversation there was a bit of a difference between AMD CPU warranty and Intel CPU warranty uh I'm not sure if that has changed because um in terms of like the overclocking and that as well I think for creators I've always told creators look overclocking is not worth it for you if you use this as a tool the stability and CPU degradation and all of that is just not worth it so I don't think this is for this but I think I have a few listeners as well who are a lot of you know very very high in the PC Enthusiast so when we're talking talking about the warranty of CPU AMD and Intel what have you found some of the things so that one said you know I got to give a little bit of a you know let's say a PR type of message here where I tell you is that you'd have to defer to check the guideline notes that are denoted by both Intel and AMD on how they want their CPU you can at least though confirm some of these points of values an example as if we take for instance here and we take a look at you know amd's Expo base page right so the Expo is kind of the preferred kit you can actually see right here that it notes you um it is underneath footnotes so here you'll see kind of information regarding AMD Expo but you can go into the footnotes and says overclocking and or under voltage which is also something people do right in CPUs AMD processors and memory including without limitation altering clock frequencies multipliers and memory timing voltage to operate outside of amd's public specifications will avoid any applicable AMD product warranty this something that you have to be mindful if you have more specific questions of course we are not Intel or AMD we would recommend on the motherboard side it doesn't affect our warranty because the motherboard has been designed with that in kind of mine so if you did want to engage it doesn't affect us but if you kind of want to be cognizant of all of the spectrums of what you have per warranty per device it is something that you do have to account for right and this is important because there are some memory kits a lot of the memory kits you purchase are going to be X and P kits so inherently by engaging that you are going to be potentially engaging in that overclocking even though you might not think it's overclocking it is overclocking so if you want to run like Pure Stock standard then you actually have to buy a kit that is not an XMP kit it's a kit that's designed to actually run at what is called the jadex spec right and there are partners that do do produce these kits like crucial I don't know how um you know where you may be in the part of the world but hey for instance if you go to them they actually have just pureja get kits and the great thing is that they'll run at that jdeck value there's no you know worries or concerns in that and this really only happens in the pcdiy realm because if you were to buy like a system you buy like a laptop they're all pre-validated to run within essentially these Factory standards right it's really only you get into this whole conversation in the DIY Arena where so many things can be selected by the user right but uh if you run an XMP Ram without the XMP because obviously XMP does not come enabled by default on the motherboard then that is not overclocking right correct yes so you could run an XMP kit on a motherboard not engage it the only disadvantage of that though is that it depends on how the profile is set for that memory to run essentially at its default setting so an example of this is a lot of XMP kits that exist out there they might actually run um at a much lower speed than than let's say a native jdeck kit of memory wood so an example is again if like I had like this 6000 kit I put it in on my motherboard it might post and default to let's say like 4 800 right it wouldn't it might not default to like 5600 so it would be under the spec or speed that I would actually want where if I got let's say a actual validated Jada kit that wasn't XMP it would actually default to like 56. does that make sense so you could be in this kind of little bit weird scenario where um you know how do you make sure that you're getting exactly the right speed right that's designed for your memory um is there a memory controller difference between I5 i7 I9 ryzen 5 or iPhone 7 ryzen 9 because if you look at the specs they all have the same same memory memory spec I know if you go let's say I 5 at 12 13 400 or when you go before um under the K basically if it's not okay then it sometimes has a different value of the memory control as well that's what I found on the looking at the specs but is there a difference between them memory controller like I5 13600k versus 13900k that one's a little bit trickier generally just because of the way that the kind of the production process works um you generally do find that usually the best memory controllers will be on the higher end CPU sides of the fence but for again the majority of the target what the consumer is targeting whether they're buying again like a 7600x versus like a 7900x or you know a 13 400 versus like a 13 700 or 13 900 you generally would be seeing very similar scaling you shouldn't generally see that large of a delineation um but like I said there is always going to be CPU to CPU variants right so you always have to kind of think about it that um more so for maybe kind of jumping from one uh product segment to another so I5 to i7 to I9 take for instance right it's always more important to kind of think about it that if you had 10 of that same exact CPU so I had 10 13 600ks or if I ate 10 3400s right 13 400s that there is going to still exist variance between any single one of those so I think that's the most important thing to keep in mind is that over overall scaling is is quite good in that generation that's the more important part to kind of account for right is that like 13th gen has a better or let's say higher speed memory controller than 12th gen series CPUs right and that generally holds true regardless of what 13th gen series CPU look at but you still have to always account for that regardless of that 13th gen if I had multiples of them some would be a little bit greater in terms of their ability scale and some would be a little bit less and some would be kind of right there in that Africa I'd like to go into a bios and have a look at a little bit some of the tools that Asus provides to actually give you um you know stable XMP and I'd like to kind of enter the this from very very beginner point of view like how do you turn on XMP where do you see your IC provider what are some of the things that you can see in BIOS that tells you something about the XMP and RAM that you're running yeah let me go ahead and get my little set up here I'm gonna twist over my keyboard so um here I'm just I'm going to essentially reboot into the UEFI you can do it actually a couple of different ways um you know a lot of people just say spam the delete key and that works actually I like doing it in Windows actually going through the advanced uh option the advanced option and windows it actually allows you to go ahead and get into the UEFI without having to spam anything so it's great because I can engage it and then just literally if I was to walk away I'd come back and it would be in the UEFI so I always prefer to do it that way but this is kind of the way that everybody will tell you to do it it works but you know there are better ways at least if you want to try the advanced command prompt option or like I said the advanced startup options it works well but with that we've gone ahead and entered into the UEFI now depending on your board every manufacturer is going to be different this is Asus right so we of course really have I think a great UEFI some people may see depending on their motherboard what is referred to as kind of the easy mode interface so if I was to start off here that's what it would look like is your easy mode interface you can always clearly see your information over here so it'll show you the actual amount of memory that you have installed so the density so we can see that we've got 64 gigabytes some people get confused by this because they're like I see this value and then I see this value and actually what this tells you is that the 4000 is just actually what it's running at so it's running at 4000 Mt but it tells you actually right here that the base profile that could be supported actually is 48 but then there's also the XMP profile we'll see directly below that the actual dim slots so that tells us which slots are populated A1 A2 B1 B2 for reference if you're first putting a system or you just want to confirm that you've done something right your primary banks are pretty much always going to be A2 and B2 yeah you do not want to put things next to each other it actually would be in opposing Banks so again A2 and B2 and then right below that under XMP we can see if we wanted to engage it we would literally just go and we would click over and we would have it enabled and you can see right there it would enable our profile now that's great under the easy mode there are a couple of little bit more options that you have once you go into the advanced mode so if I go into the advanced mode interface let me go ahead and actually just redefault this again here so I'm just going to do F5 F5 is just going to kind of reset the board back over to defaults and let me go ahead and back in here in the advanced mode sorry I just needed to see where I was there so for us under the advanced the default icon option you would be looking at would be under extreme tweaker and then AI overclock tuner and you'll see that you have a couple of different options now manual is how you would actually overclock your memory if you didn't have an XMP profile that means that you would have to kind of know all the things on how to kind of tune your memory so I'd have to go okay I have to pick my memory divider and then I would actually have to Define timings and I would have to define the voltage so I'd have to go over here into terms of timings and I have to set all these timings and same thing I'd have to go down here to my voltage and I'd have to change that but the great thing is with an XMP kit that's what the memory manufacturer's done for you is that they've essentially pre-screened the memory and they've kind of tried to Target a value that they feel pretty confident is going to work and you can select it now on Asus you will see that there's different XMP profiles you can see right here we have an XMP profile one and an XMP profile two some people go like what's the right one to use technically there's no right one to use it can we just give the option because it can improve the likelihood of potentially getting you to have a higher degree of stability or even better performance an example of this is xmp2 is actually what is considered our Pure default so this is exactly reading the information just as it's provided by the dram manufacturer so if you want to try something xmp2 is not a it's not a bad place to start because it's like the most pure basic form of XMP is that always like this that X and P2 is like the base good like place to start or like the most stock because sometimes it's XMP one because it's a bit confusing because you think XMP one would be kind of the first priority yeah but actually xmp2 is the is the most like what the memo uh the ram kit comes with say look this is the XMP profile right it's the most part of Pure Stock profile so yes xp2 is always going to be kind of that most Pure Stock profile option xmp1 is what we refer to kind of as like an Asus specific profile and you might go well why does Asus have a specific profile well if you remember kind of all the part of the conversation that we had a key thing to understand is that we understand our motherboard design better than um than everybody else right because we designed the motherboard we designed we've coded the firmware we created the motherboard topology and so sometimes we actually might have more tuned sub values that might be in there to allow us to maximize the compatibility or even the performance so you can attempt to load in xmp1 so those are your kind of your two options that you have available now technically there is a third option that you also have available um then I would actually have to drop in another kid and I'll do that in a moment so you can see but we actually have another one that's called XMP tweet and XMP tweet is a specialized set that depending on a kit of memory that you have installed that we've even done further work to we realize that there's even more kind of tweaking or optimization that can happen to that specific kit of memory to allow it to even have better performance and so we will give you another value so instead of just seeing X and P1 or X and P2 we can actually do XMP tweet actually I think I have a screenshot so I don't need to swap other kit I think I can show you via a screenshot but those are essentially your XMP options after you've done that you're pretty much set and then you would just you know you would go to uh exit or hit F10 and you would save and exit so if I understand that correctly Asus if you have a certain type of ram kit Asus motherboard will understand that there's more potential in the RAM and they'll give you even better other faster timings or faster frequents or something like that with the XMP tweaked correct yes and so the the goal there essentially is that we realize that some actually kits of memory do give you the ability to essentially have better performance so let me see I think I have the screenshot here so give me one moment and um see if I can bring this up right here so this is what it would look like if you uh saw it on essentially a board where you installed the memory and generally you would only see XMP tweetier under a two dim configuration you wouldn't see this under our four-dim configuration because generally again four dim kits don't exist right so two dims are generally going to be the optimal configuration but you can see right here it says XMP one xmp2 and now you see an X SMP tweet so that XMP tweet would be essentially that other option that you have available so this would even be I would say like maybe the best option if you wanted to engage one for let's say again best performance you've decided that you want to overclock the memory and you see that as an option you could go with that but my general kind of rule of thought though is always when engaging memory and doing kind of overclocking as I go first always what's going to be the highest level of stability and assured success so I do usually default to two and then to help people test your memory and that's the other kind of cool part that we do have within the UEFI that I can show you here so yeah I wanted to ask you about the memory testing if people are getting like the memory stick right they load their XMP on and they want to know look is this even stable like is my CPU able to support this if my you know how what what's some of the tools Asus gives so a cool option that we have is right here you'll actually see that there's a mem test option built straight into the UEFI so this is not going to be on every single one of our boards but it is on a large number of our motherboards and we do list it on the product page so you can't always confirm that if you're interested in knowing this now if you don't and you are you know maybe running on also another board another another motherboard manufacturer you can get a bootable memory test um so you can get a utility to do this but we just streamline the experience for you by building it straight into the UEFI so right there you could literally just click yes and it will reboot into the UEFI environment and it will allow you to go ahead and test the memory so this is great because again if you remember kind of in the very beginning of our conversation I talked about that there's three ways to assess memory right you have the post that is your power on self-test right and so the power on self-test is the first part so an example to the power on self-test if I just get out here that's when we first kind of start our system that's your power on self-test so your memory has to be able to pass literally just going through that initial power on self-test the next part is the boot the boot is when it boots into Windows now it could crash during the boot stage and the last stability test is actually stability underneath your OS environment actually being used so you have to actually ideally pass all three forms to have memory be stable so this is the third part right yeah so the the third part would be actually going into the windows environment and then being able to run your memory right so again three things have to occur you have to be able to um you have to be able to post you have to be able to boot and then you have to be able to pass within Windows now once you're inside windows there are different options uh one option is like if you're using an Intel platform you can use XTU extra that's Windows a utility provided by Intel which is free that allow you to run a memory test I like an application that you can't get for free or you can pay a small amount to essentially support the Creator it's called occt it's a fantastic memory test that you can run to check system stability and there are a couple of other applications that exist out there as well to systems system memory but you want to run that once you essentially get into the windows environment so I'm sorry so I'm back in here and if you then wanted to kind of check your memory right at that point the easiest way to check it is actually in the task manager even if you don't install utility you can just literally open up your task manager and go there and you would be able to check your speed now I reloaded back the defaults um so uh it's now at a lower value but if it was 6000 Mt you would see 6000 if you know it's 48 whatever you would see that corresponding value but you can also again you can use utilities from different vendors um our Armory crate utility will actually show you uh information XTU can show you information cpu-z can show you the information just one thing I will show here for a cpu-z though is keep in mind that for that memory um it you still have to do a little bit of math if you're going to use CPU Z so a lot of people use CPU Z and they'll make make a recommendation but I think task manager is more immediate because it actually gives you the correct value if you go here on cpu-z and we go to memory you're going to see 2000 which will show which will be confusing to a lot of people because it'll be like wait two thousand that's less than four remember you have to multiply because it's DDR so you would have to multiply that times two and that would let you know that you're writing uh four thousand right yeah did you mean ddf5 Excuse me yes yeah yeah uh well I said DDR just in general right because if you had ddr4 memory you would still have to do the multiplier and if you had ddr5 you'd also still have to do it so yeah but I do think this is the CPU Z is still important because I've had instances where uh task matters just can't read it or reads a random number of memory sometimes even laptops because laptops memory sometimes is integrated and stuff so task manager doesn't quite know what to read but cpu-z you can easily read the right frequency there so that's a good tool could you show us more about the memory tested there so let's say we've we've got into the BIOS we've hit the XMP on we've got X and P two on it's the most default one or the best one right yep and then we've pressed save we've got into the windows and now we're going back to the bias to test the Ram or is this the right order yeah so you you uh actually ideally if you ran through that mem test the mem test will usually have a default that it will run usually I think it's set to uh four passes so it actually will come up to you with the entire green screen that would tell you it actually passed um depending on the amount of memory that you have that could either take shorter or longer um I would say on average probably going to be maybe about like 30 minutes or something in that range so once that's passed then you would want to go into windows and at that point you should feel pretty confident um but mem test I'd say it's not a hundred percent a validator in terms of having the best level of stability but it's a good indicator of generally system stability so if it's passed there I still recommend using maybe a more formalized memory test but that you could run again within Windows right so that's that's an option that you have but the kind of the hierarchy should be set the memory first go into mem test and then after you m-tests boot into windows and then run your memory test there but there are still a couple of other little things that we have here in this UEFI so a new option that we also introduced and it actually applies to z690 so retroactively we imported this into is we have a lot of automated overclocking Technologies uh really kind of a leader in this space and so we've had our aioc technology but now we actually also have a cool little technology to get you a little bit more information about the memory controller so if I go here under AI features again your board has to support this so it should see be one of our aioc enabled motherboards you go here you'll see that there's actually a little uh value right here that says get m m c s p so that's memory controller SP and so the SP essentially is an internal logic number that ASU says defined for multiple criteria that help to kind of give you a little bit of a determining Factor about the relative quality to your to your CPU so you can see right here we have an SP information for my CPU and now I also have SP information for my memory controller so we're the only manufacturer that does this there's a lot of advanced logic that goes in to be able to Define this information but essentially the higher the value the better the quality that is there are multiple factors to it and it's not an end-all be-all Arbiter to it but an example is let's say maybe you had an um uh an mcsp like let's say on 13th gen that might be like 73 you know let's say in the 70s right you could probably have a CPU that would comfortably be able to do probably like you know 7600 Mt right so it is a way that we're helping to maybe provide some information that uh you don't know too much but you could put this in and then maybe you could join our group right um and you know ask hey what what value do you have what value do you have and then at least it helps you to kind of know what might what range you kind of might be in as opposed to just arbitrarily trying something because you can of course try it you could just set your XMP and see if it crashes and then you would know how good your memory controller is but if you want to at least get a little bit more information this is also something uh that you need Asus uniquely has so we have this integrated mem test we also have this s just about the SB value there the higher the better correct generally yes I don't want to use this as an enderbr because there's a lot of people that get stuck on SP values um when they think like it's the only Arbiter you can have some SPS it could be lower and you could still have a better memory controller than maybe somebody else so it's not let's say an all determining attribute right it's just one part and what does the SP stand for uh we don't have a designation for it it's just an internal internal uh it's an internal is it just Asus internal because I've heard that some of the CPUs have that as well if you know the German overclock right there Bower he yeah he was doing some videos about like CPUs that have very highly Bend you know the SP value was XX amount or whatever yeah but part of that is because so many we're the number one overclocking brand in the world right um led by so many of the overclock enthusiasts will use our boards to pre-screen and vet their CPUs because as an overclocker right you may test hundreds of CPUs right so this is one of the best options for overclockers uh because they can literally just put in the CPU check it and then put another one put in another one until they keep until they find one that it's you know much better because all of a sudden they might have all these epus and they go wow this one this one was 121 and and all these other ones were like you know 105 111 90 84. right you know whatever you know the value range would be so um we get a lot of people that will share these values in the community but it's not too uh like I said over complicated just showing you an option here that if you want to try to understand a little bit more right or maybe you know you buy two CPUs and you kind of want to see whether maybe one is a little bit better than another um you can do that because the only other way that you can then test for your memory is to do actually um what I call dram divider testing so you may have to do this regardless it's let's say you get into a situation where your CPU um maybe you load up the XMP it crashes and things don't work the way that they're intended well what do you do right um do you just go like oh I returned the kit not necessarily um it just might that you might have to massage it a little bit so an example of this is let's say here the profile 6000 right C32 so it's a pretty good kit of memory right here one if you again wanted to find out more information you could also go to this little tool right here the SPD information and this will actually tell you a lot of great values where you can see right here it actually tells us this kit it's SK hynix right it tells us even really great information like the serial number the part number the actual production year all that information and actually shows us also right here each one of the profile points right so the jdeck value the XMP profile and the voltage table it's all listed in there and if you guys ever don't know one really great feature that I also love in our Asus boards when people have to share information and they take pictures using their cell phones hey it works but you can always just stick in a flash drive into a motherboard hit uh function and F12 and you can save a screenshot straight to your flash drive so it's much easier better way to share sharp and clearer screenshots um than you know taking a picture and getting the glare and everything like that right um but if we head back over here the easiest step you would do is let's say hey six thousand is not stable so what I do well just go here to the dram frequency divider and drop the divider so if it wasn't stable at six thousand go and drop it down so it might be like hey fifty eight hundred and now I'm Gonna Save exit and reboot now this of course is not the native value but you're still benefiting because you have that higher voltage that's been defined for the memory and those timings so you're still going to get the C 32 38 3880 so you're still going to be carrying over this optimized profile so you're still getting a benefit from an overclocked kit of memory it's just that you're running it at a little bit more of a conservative value because maybe your memory controller can't run at that speed right or maybe it's the configuration you're running can't run at that speed and that's an easy way to test finding what is going to be stable on your system and I call this simply just dram divider scaling right that's all you're trying to do is you're just trying to find what is going to be stable so if you tried one configuration and didn't pass then just go in and drop the divider and again if it wasn't stable there then again you go back and eventually at one point you're going to to find a divider that would allow it to be entirely stable right so that's the first easiest test there is probably one other last point which could be you could modify some voltages and I could show that if you want but that's I'd say the easiest way to kind of attempt to test something outside of the XMP profiles themselves right where you may have to test both X and P2 or xmp1 and then if xmp1 or XMP do it doesn't work then you want to try changing your dram divider most creators probably will leave it at that point and won't start messing up around with all the voltages very simple right as far as just you know you're changing is right like hey you can just drop it a little bit of a bump all right and then see if everything works through and just get running from there right you did mention that when you get into the windows so XMP you click the profile on and then you do the mem test in BIOS and then go to Windows what is some of the best memory tests on Windows that you can do because I've in my experience I have seen that I can get the post I can put the XMP on I can get to Windows but then sometimes programs start crashing I get bocs sometimes not even like in the first hours after like eight hours or a few days later I get a boc and it says it's a memory management controller blah blah something like that and so what are some of the best tests you can do on Windows like last thing you I think you mentioned something just to go over that again yep yep so actually let me let me go ahead and show you right here um actually this this kit is actually an engineering sample kit and there actually is a faulty bit configuration so I'm actually hopefully I can show you actually what it would look like if if it got an error so I mean one second here to go into this configuration here and if I remember correct from this kit configuration that I have if I set this I think it should run on this configuration if not I can I know I can test I can show it another way for sure but let me just try this one I don't remember on this engineering kit if this will work if not I can swap out to another one and I can show you for sure how it would end up doing it but um what we're going to do is we're going to go into operating system and I'll show you just kind of two very basic ways that you could actually test the memory uh one like I said through XTU and the other one through occt but the main thing is that essentially what both of these utilities are attempting to do is fully load the memory right um and what you'll generally find is that if you don't test your memory in this way um it could be essentially that you're able to post you're able to boot you're able to start running applications but once an actual application sits into data bits or part of the memory that essentially has instability that's all of a sudden where you get the crash you get a blue screen you get some type of a kernel issue right that pops up and lets you know that there's a problem that's present in the system right and so here essentially what we're doing is instead of going through and running applications we're essentially just immediately placing a load on a large part of the memory generally at least about 80 to 90 percent of all that memory and we're also adding an element of heat heat can influence stability for memory it's it's not that critical of an issue a ddr5 most ddr5 kits on the market all have a little temperature I see that you can actually she find out the temperature that they're running at and if they can get too hot this can increase the likelihood of instability which is also a little bit of a factor I think especially for creators because if maybe you're running like an extended render and it's actively sitting in memory for a long period of time you know when you first start the render it could be like at 45 degrees but then like uh four hours in it could be like at 65 or it could be like 70 degrees and the temperature can play a part that sometimes if temperature gets to a certain point it can actually bring down instability so that is a little bit of a factor but you don't have to worry you don't really don't have to worry about it too much that it's not that much of an issue but if you are kind of trying to be conscious of your system and airflow and things like that having good air flow to your system will help to mitigate this concern but let's see if we are on Windows here and so um from there you can I'll show I guess first through here through xdu you just open open it go ahead and open up xdu so actually see that's actually a good indicator right there XTU actually already gave me the error so generally it's already telling me this probably wouldn't happen if you're running a normal memory config some applications are written to be a little bit more sensitive and some are going to be a little bit less sensitive let's see if they have CCT will open up here and here I'm expecting actually this configuration to actually not be stable so you can see right here um where it's got the memory right I've gone ahead and defaulted this from 80. 290 I prefer 90 just to put a little bit more on there and all you would do is just go ahead and click this again guys you can go ahead and download this utility from that vendor it's a it's a freeware application I strongly recommend though do consider actually getting the application there's a small little like you can support the patreon it's like about like five dollars or something like that great great utility to be able to test actually your CPU your memory your graphics card even your power supply you can do a lot of really good testing within this application um you'll see it gives you that link right there you would be able to click um start and when you go ahead and click Start you'll then be able to go ahead and stress test the memory right and so if you have faulty bits either one one to two things that happen either your system might immediately crash right there or it might take a little bit of time and eventually when it finds the memory bits uh that fail you'll actually get little red markers that'll start coming up that'll actually let you know that the memory has actually failed right so we'll see if it comes up there I'm going to go ahead and let it let it run for a little bit and oh okay um one of my uh last questions I was trying to find this out but I didn't find this out it was Pro art is like asus's top end crater like kind of lineup do you know if proat motherboard also have the mem test built in the bias uh yes we do have actually mem tests on not on All Pro Models because we do have Pro models in different segments right so um for one mem test is only on Intel platforms due to differences in Intel's memory code and amd's memory code um there's currently some limitations on the AMD side that doesn't allow us to have the mem test so if it's on an AMD Pro art motherboard or even not an Rog motherboard doesn't matter there's no mem test so it's only on Intel and then on Intel it would be on Z series but not like on B series I also do have a suggestion for Asus what should the name the SB uh what would it stand for it should be called silicon Pottery value it's silicone Pottery okay I'll definitely I'll make sure to pass that over to the team we'll see if maybe in the next revision maybe a little update we'll add in a little uh information and when you click on the little information tab it'll tell you when you look at the member of Kit one of the first things you'll see is like the megahertz you know like six thousand seven thousand MTS but then you see that sometimes the timings number actually gets bigger as well now what's your experience in timing versus megahertz you know actual speed of ram versus timings how does this affect so ultimately um you know the main thing that you're trying to do when you have like a kid of memories that you're trying to get overall hopefully the the lowest latency possible and so the timings are a byproduct of multiple parts of how the memory operates so one is going to be the bandwidth right the memory operates right and then it's timings and then you kind of bring those together and you have your end output latency um and so there's a lot of default logic though that people say that automatically just always get lower timings and that's always better but that assumes that your primary timings which are the values that kind of get communicated a lot are always going to be the most important so if I head back over here because I think I'm running that test right now so if I just look at this screenshot for the UEFI and here we take a look at this image we can see right here that again 6 000 right and then we have timing so 32 38 38 80 right the lower this value the better now there is generally some truth to that the general generally usually the lower the value but it's not always a full indicator of the performance of the kit you could actually have some situations where some timings could be higher in terms of let's say the primary timings but Sub sub timings are actually different and that could actually have better performance and this can be kind of confusing because sometimes people will take two kits they'll put them together and they'll think on paper this one should be faster or this one should perform better but on this one it actually performs better and that's because the timings which there are a whole bunch of them some timings are more important to some application workloads than others right and it really kind of depends on how that timing has been optimized and that gets a little bit to kind of what we talked about with like the XMP where we did like the XMP tweet where we try to make sometimes a tweak to some of these values where we know that that actually really gives us a performance uplift and it might not actually change any of those primary timings it changes tertiary timings um that actually tend to have more performance impact were this ultimately though probably affects the user more is tighter timings tend to require memory that has More Voltage so if you're maybe manually making an adjustment or if you're even buying kits of memory a lot of people don't account for this but the tighter that frequency is and also that higher that that memory divider is Right generally it's going to be more voltage so if I can go back here to my screenshot uh your base value for ddr5 right is 1.1 volt you can see already right here we've got 1.35 but it's not out of the realm of possibility you could be looking at like 1.45 right so that's actually quite a bit more voltage right it's producing like consuming more power producing more heat and same thing if if you wanted to push a kit of memory more so you want to try to get more performance out of it you might also have to increase the voltage so that's you know just something I guess keep in mind for for timings but um generally I'd say the rule of thumb right your default is C40 so anything below that great but you don't generally have to kind of go from the perspective that you have to absolutely get the kind of the lowest value it is generally going to be a good rule that the lower generally the better performance will be but there are other sub timings that can sometimes have a bigger effect on performance we'll have to do that test for craters run some of the benchmarks with the same frequency but lower timings and then see how much does this have affect to having the lower timings and then do the different frequencies and see what the performance difference is that would be interesting yeah for sure for sure and that's also the other thing too is some applications treat things differently um you know a lot of times in the very beginning there's the discussion on like how much ddr5 provided you a benefit and one really important part of ddr5 is because it's bandwidth is so much higher than ddr4 it provides more bandwidth to each one of the CPU cores and so the more multi-threaded an application generally the more that ddr5 shines if that makes sense right a less multi-threaded the where the applique location the less that you will generally find that ddr5 is going to provide an immediate performance impact right but if you have an application that's using a lot of these threads because it actually actually has to feed from the DDR channels right you actually are supplying that to the course you actually get a performance uplift so the best multi-threaded performance will always come from ddr5 and then also higher bandwidth ddr5 so you know creators that are doing maybe more workflows I'd say like on the videography animation production workflows hype side they would probably see more benefit if you're somebody like me like I'm a hobbyist in doing a photography and then I do some moderate kind of video editing on the photography side I get some stuff but you would have to really do like a lot of decompression or compression which would use more cores to see more of a bandwidth perspective so like when I go and take pictures with my dog if I filled up like a full like 64 gig uh you know SD card with like tons and tons of of raw and then I'm going to batch convert all my raw to maybe you know some jpegs or something like that I could start to use a lot of cores under that situation specifically under that batch conversion then I would see more benefit from dram bandwidth but in some other situations if it's more just cpu-centric right then you know the memory being between like six thousand to like 66 to like 7000 you might not notice any almost any performance benefit so yeah exactly the more um the application used more cores probably you'll see more benefit for more bandwidth yeah so for example people who are doing Lightroom their wedding photographers they're doing they load a lot of photos and then they batch export them to something else and Lightroom classic very well utilizes the multi-calls but then at the same time if you're working for Cinema 40 and the cinnamon to R23 you're doing the test you've got a really you know fast kit of random paid a lot of money you're not going to see any performance difference in cinebench R23 it's only the difference when you're working with a lot of data that needs to go through the ram right when we see the benefit of multi-threaded and then the big Ram capacity and the like faster speeds that's where we see see that like in video editing like you said and photo editing yeah that's where it's always important of course you know just you pick the right platform that favors the application you're working at right where you got to look at frequency you have to look at the IPC right so the instructions per clock you got to look at you know ring bus you got to look at dram you know all these little different pieces on how they influence your ultimate your workflow right thank you very much for your time I think that's it for me the last thing is did that memory test pass now or is it oh let's see let's see did it did it pass or did it oh there it goes yeah look at that so I was corrected I knew it was going to fail there so this is normal as is expected here but as you can see right there it did find errors but I picked a configuration that I knew would be able to error but not generally hopefully crash with a with a stop error so that you could see what it might look like if you maybe had a crash in Windows that was maybe marginal so the system could still work like you could still open browser maybe thinking if he's still working but you wouldn't get a full crash so yeah at this point um like I said the next step would be you could do a couple of different things you could one you could go attempt to try maybe those dram dividers you could attempt to maybe relax the um the memory timing Maybe by like a value of one so like if maybe it's like a C30 kit you could try to bump up maybe the primary timing to like C32 if you wanted to you could also attempt to maybe make an adjustment there to the the dram voltage but um you know that's kind of up to you just depends on on on on you know what your kind of goal is right what you're trying to but generally if you you know you you kind of distill this all down to a lot of what we talked about if you try to generally try to be sensible about what you picked then you know you should probably have a pretty good success at probably not running into an issue right that's uh it from me thank you very much for taking the time to explain RAM and XMP to us fantastic man thank you

Channel: Tech Notice

Views: 141,901

Keywords: tech notice, best tech, XMP guide, Ultimate RAM XMP guide, ultimate XMP tutorial, ulitmate tutorial for ram xmp, how to get xmp working, xmp fix, xmp tips, RAM xmp, ram xmp not working, ram xmp bios, ram xmp asus bios, ram xmp profile crashes pc, ram xmp profile not working, ram xmp explained, intel xmp explained, xmp profile, how to enable xmp, how to, how to set ram speed in bios, extreme memory profile

Id: wzVZgTP2204

Length: 101min 47sec (6107 seconds)

Published: Tue Apr 04 2023