Gadgetory

this video wasn't originally in the plans we briefly talked about clock throttling due to vrm thermals in our a bit wit collaboration video but didn't take it any further after some conversations with issues following that video we decided to stick thermocouples all over the rampage six extreme x-29 motherboard to objectively monitor vrm temperatures under various conditions this includes direct cooling passive cooling and a range of clocks helping determine at what point a higher powered cooling solution is needed in order to prevent VRM forcing a clock throttle before getting to that this coverage is brought to you by EVGA and Nvidia with the destiny to 1080 TI bundle a 1080i SC to comes with a synchronous fan control for its dual fans and nine thermal sensors and again includes destiny to learn more at the link in the description below so again this is following our video on Kyle's CPU delayed where we tested the temperatures before and after D lighting and then talked about some vrm temperature issues where it would hit a throttle point and we'd have to basically put a fan on the BRM directly to fix that problem and for evidence of being post kyle delayed video I still have a silicone adhesive permanent stain on my shirt because it exploded much like Kyle's bottle did when he tried to pop the cork so Kyle is cursed and he has spread the curse to me but yes so we're testing vrm thermals basically what I've done here is we mounted thermocouples everywhere so there K types they are mounted to one of the capacitors it's the center capacitor in the capacitor bank that's basically north of the CPU socket and then just above the capacitor bank following the compass it still be north is the line of chokes and so we put a thermocouple on one of those inductors and then above that there's the line of MOSFETs and those are the international rectifier MOSFETs so we put a thermocouple on what we would assume to be one of the hottest MOSFETs basically dead center so it's flanked by chokes to the south and it's got the erm heatsink on top of it these thermocouples are 1/100 of an inch thick they're pretty damn small they do not influence the thermal transfer capabilities from the heatsink the thermal pad to the component in any meaningful way and they're basically the thickness of captain thermal tape 4 to give you an example so we use those they are not electrically conductive it they sit between the heatsink and the component so we're measuring the temperatures on top of the component not inside of it and the effects of case temperature is going to be different obviously than the internal temperature but that doesn't really matter because all we're looking at is a delta from test a to test B at what point does it throttle on our thermal thermocouples and at what point can we stop that throttling by stepping back the clock the voltage or implementing a fan so that's what we're testing this is presently being done on open air and an open air configuration means that we basically have established a baseline today implementing a case testing with a case means a whole lot of different outcomes are possible and we'll go through those up in a bit but basically for case testing we have dozens of case tests on our charts you've probably seen them at this point and for a good half the cases on the market your vrm thermals are going to be worse than what we're showing here today and then the other half they will be better it just it depends on the airflow configuration of the case we're testing baseline and then I'll give you some guidelines toward the end as to how different cases will impact the thermals and in what way other than that what we're getting into here the big question is whether or not it's reasonable to expect the Asus Rampage six extreme vrm heatsink to be capable of effectively passively cooling the vrm that's what we're trying to look into and I'm joined by Dare Bower toward the end of this video for his own thoughts on that exact question so stay tuned for his thoughts as he did the original X 299 disaster video we'll start with only passive temperatures with no extra fan to read about which cooler we used and other system configuration information check the article link to the description below for the testing methodology section we are also using the 79 80 x ii that was deleted recently some of the temperatures here are not terrible like auto settings for example the clock sits at around three point one gigahertz average with complete auto out of box motherboard configuration with the vrm component temperatures at about 67 to 70 degrees Celsius for the center choke and MOSFET with the center capacitor at 62 degrees Celsius a very lightweight trivial overclocked 3.9 gigahertz and 0.99 to voltage ID brings us up by more than 10 degrees Celsius so now we're at 80 to 83 degrees C for the fats and the chokes as measured by the thermocouple this means that the internal temperature of course is higher than this but we're not throttling yet our power consumption also increases to 293 watts at this point up from 225 when auto go into 4.3 gigahertz and 1.15 VI D is the next step and puts us into throttle territory immediately the MOSFETs are now hitting 100 cm our thermocouple measurement which means that we're at the point where the clock is throttling until they cool down and then they effectively spin back up and we throttle again we can actually plot this over time in a few ways but we'll save that for the next number step at 4.5 gigahertz and 1.2 voltage ID we're again hitting throttle territory where the CPU frequency Peaks and drops chaotically with mosfet temperature the same is true for 4.5 gigahertz at 1.2 for voltage IV unsurprisingly and is the most extreme scenario where we just wanted to torture it here's a look at average of frequency for each configuration as plotted against fat temperature at 4.5 gigahertz and 1.2 for voltage ID we're throttling 230 104 megahertz and so we're about 1.5 gigahertz below the target frequency the next one shows us 4.5 gigahertz at 1.2 voltage ID where we were hitting 33 11 megahertz so better but still significantly lower than it should be at a low a 4.3 gigahertz and 1.15 voltage ID we managed to climb to 36 77 megahertz still about 1 gigahertz below the targets and that's still from throttling 3.9 gigahertz at 0.9 voltage ID does manage to actually hold its thirty nine hundred megahertz configuration and that's because the vrm temperatures are so much lower that they're not throttling internally technically speaking of all these tests the lowest configured overclock configuration is performing the best and that's because it actually holds its clocks without any throttling despite the lower input setting to demonstrate that this has nothing to do with clock or voltage stability on the CPU side we can refer to the next chart this chart shows that vrm component temperatures when we add fans as shown here none of them with the fans anyway are throttling with an intake Sun and Maglev fan positioned 1.5 inches away from the vrm heatsink and an end ext one forty millimeter fan positioned as exhaust directly above it we keep vrm component thermals in these seventy to eighty degrees Celsius range at 4.5 gigahertz and 1.2 for voltage ID it's about 60 to 70 when at four point five and one point to zero voltage ID we actually don't know the true difference versus the fanless configurations because the throttling means that the temperatures would have been higher if left unconstrained this is even true with SPID disabled and before going further just to kind of illustrate what we've got here with the configuration that has fans added to it it's an open air test bench as stated and the Sun and fan is mounted immediately next to the VRM it could not really be any closer without chewing up the EPS 12 volt connector and the current clamp attached to it so that fan is spending at over 2,000 rpm it's 120 millimeter fan it is effectively a corsair maglev fan it's the supplier version of it that's sitting right next to the BRM heatsink spinning as loud and fast as it can and blasting the heatsink with air the NZXT fan is above it and it's just simply acting as an exhaust to pull away all that heat including radiative heat off the back of the GPU and get it up and into the room unfortunately for everyone in it but it cools down the vrm so doing that with an obnoxiously loud configuration we end up again depending on the clock at maybe 60 to 70 degrees celsius for 4.5 gigahertz one point two voltage ID which is completely reasonable that's a very good temperature for the vrm to give you an idea the maximum specified internal temperature for these MOSFETs is about 150 degrees Celsius so for external temperatures looking at something like up to 125 C or really an operating range and obviously lower is gonna be better for MOSFETs but yeah at 60 to 70 we're perfectly fine that is our thermocouple basically case temperature so it's not the internal temperature internal is gonna be higher than that but again it's all relative Delta's between each other so obviously that's a big improvement but it does require a pretty serious cooling setup that's allowed so keep that in mind you can definitely do decent cooling set up with a case we'll talk about that more in the end though here's a look at CPU temperatures for each configuration you'll notice something interesting here in the most extreme scenario of heavy throttling the CPU temperature is lower than when we add fans at 4.5 gigahertz and 1.24 voltage ID which is intentionally on the high side to make a point we see that the CPU is 84 degrees Celsius for the cooled FETs or 75 degrees Celsius for the passively cooled FETs this demonstrates that the throttling is not a problem that's sourced from the CPU we still had thermal and power Headroom for the clock the CPU was doing just fine the cv is only hotter in the fan test because it was not being limited by a choke point not throttling on its frequency and so was able to operate at the full specified frequency and voltage and therefore operate hotter on the CPU but the V RMS are cooler than previously keep in mind again that CV temperature here doesn't mean much for the throttling configurations because they weren't actually performing at those clocks and voltages our best overclock tested for stability and frequency was 4.5 gigahertz at 1.18 voltage ID which is actually pretty damn good compared to our chip Kyle's performed well here and here's the temperature over time for that problem MOSFET went under this peak overclocked configuration you can see the temperatures are rapidly increasing and decreasing rather than reaching the steady-state or sending ad infinitum that's because the board is throttling the cpu if we plot the cpu frequency over the temperature you'll see that throttling aligns with each and for this test we're at about 470 watts on the CPU which means we're producing around 35 watts of heat on the BRM the cooler can't handle it without backup in this instance because it's just not enough here's the full table and raw spreadsheet form just in case it's easier for anyone to have all the data at once average power consumption plotted a range of 225 watts to 530 watts depending on the clock and the voltage ambient was monitored every second of each test and kept generally close to about 28 degrees Celsius and the ambient column is averaged over the whole test that we also internally plotted over time to ensure consistency capacitor mosfet and choke temperatures are also represented here so this thing this is as community members EDG calls them a thermal blanket this basically is an insulator that sits on top of one half of the vrm heatsink covers it up and it's your i/o cover so we did test this on and off just for an a/b test with it full credit to asus it's one of the least stupid RGB implementations i've seen they have an on-screen display in it and it's like animated and plays temperature readouts ironically it's a it's pretty damn cool unfortunately it does increase temperatures a bit but it's not catastrophic so in quick testing and not even don't even need a chart for this quick testing basically revealed about a 2 to 4 degrees celsius difference dependent you were measuring so measuring one of the ICS that just sits directly under the heatsink that's covered by this that temperature difference was about three and a half degrees Celsius increase by having this on it as opposed to not and the MOSFETs and chokes weren't that much more either they're like two to three degrees max and they are covered by the other half of the vrm heat sink so it makes sense so now the worst thing in the world that said it does increase temperature a bit so if you're after absolute best possible performance then get rid of it generally speaking another two to three and a half degrees Celsius isn't going to change anything for you but you know it an insulator and it doesn't really do a lot for you other than look cool so it's an option to remove but Asus did do a good job I'm not making a completely pointless IO cover it's only partly pointless so that's pretty big praise from us for an IO cover so yeah not terrible not terrible but it does increase thermals now as for the rest you definitely need airflow on this thing for overclocking high core count CPUs this is a problem where as we've shown now numerous times and case reviews and auxiliary content cases really do not provide very good airflow at all anymore and they especially don't provide good airflow over the BRM and for all of you air cooler enthusiasts who take every opportunity you get to say how much better knock to it is than everything else keep in mind not even air coolers will save you in this scenario unless their downdraft air coolers because the problem is this heatsink for this particular motherboard is situated in a way that it's shorter than most memory and so you've got memory and if you have four sticks which you should for this platform you have two on each side of the socket and the two on one side are gonna block any potential air coming in from over there they also generate heat that heat is gonna get trapped in there unless you do have an air cooler that's an advantage and then on the other side you've got the other half of the heatsink with the other two sticks of memory and it's it's covered by this thing so not an air cooler is not going to fix your problem and liquid coolers aren't going to fix your problem in fact the better cooled your CPU the higher you're going to be able to overclock sort of generally speaking and the higher you try to overclock the worse your vrm thermals so as their Barrow says in a moment here in the back half the video if you put a 420 millimeter cooler on there you have no problems with CPU thermals but now you're gonna throttle on the on the erm so the problem is if you're building an X 299 platform in a case which probably are if you buy any computer the item to consider would be potentially a top intake for the case that would immediately resolve all issues if you depend on your case configuration you might not even need to run it at that high of an RPM you just need some airflow over it alternatively CLC mounted in the top near the motherboard so there our instances where in a lot of cases you can configure it so that the CLC is mounted on the top front of the case that would be a problem here because it's gonna suck all the cool air that comes on the front immediately out the top of the case and exhausted what you would want to do ideally if there's clearance is mounted towards the back of the case so that you get some airflow over the VRMs and then that air gets pulled through the radiator that would be ideal there's not always clearance to do it but try to if possible otherwise get some kind of other intake over there to supply air flow or just stick a fan on a spike put it right over the VRMs a radiator in the front of the case is probably the worst thing you can do because you probably have a dual axial or triple axial fan for your GPU it's an open-faced fan that means all the heat goes out into the case and then if you have a front mounted radiator not only is that pushing hot air onto your GPU which is another problem all together it's pushing a hot air onto your VRMs and you also have hot air from the GPU to deal with so now you've really got a serious BRM problem and it's gonna look worse than what we showed here today so those are the things to keep in mind with cakes now for the very last part here the big question is whether or not it's reasonable to expect the rampage six extreme a $650 motherboard in the u.s. to maintain lower vrm temperatures than it is here because thus far we haven't said that this is a design flaw with the asus motherboard because it might not be maybe just these 18 core CPUs that pull 500 Watts through the EPS 12-volt cables are just too much for the VRMs to handle that might be what it comes down to now issues told me that they're vrm could handle it to some extent so that is a consideration but we still haven't decide whether it's their fault or it's just a high-power Intel CPU so I posed that question to dear Bauer the CPU is take a lot of power at 500 watts it enters into the realm of being too demanding with a passive sink and basically the next segment is a brief conversation I had with their Bauer who is a pro overclocker / engineer and does a lot of xoc stuff so let's throw to that clip is that pretty much in line with what you were seeing yeah absolutely I mean it just shows that once you're hitting a certain power consumption of the CPU which on your CPU seems to be like 4.4 yards and well if you want to run higher at higher volt and power consumption will be too high that means that the MOSFET essentially also dissipate too much heat which then causes the vrm to throttle because the surface area is simply not enough if there's no airflow and that's exactly what we have seen on I mean basically all boards on the market I think is it just like is it just completely unreasonable for me to expect that the asus rampage 6 board should be able to basically passively cool itself to some extent or like how what kind of cooling should we be expecting out of the heatsink that's on there I guess that's always a matter of perspective my personal opinion would be that if you sell a product no matter what kind of floral it has to work in all conditions where you would typically use it and especially on like h EDT platforms you can expect that they are running for a longer period and also on higher load or I don't know rendering applications or whatever and it's very likely that you run into situations where it's gonna be really on the edge or above so my personal opinion would be that or at least I would expect that it would work even passive cooling and looking let's say 8 to 10 years back I think motherboards back at the end they had to to dissipate a lot more heat from BRM and all from chipset so back in the days the cooling solution was much more advanced and what we've seen this generation so I wish that yet a surface area was poor was basically better you know you know the Asus workstation boards right and those ports they have really really good cooling and usually at least I didn't test the X 299 versions yet of the workstation boards but on X 99 the cooling solution was extremely good because they use those fin like thinner fin designs arranged in days and that's what vendors usually try to avoid these days on normal main boards because they think it looks cheap but yeah I heard that area there yeah I can understand their point of view because if you look at the heatsink design what we can see now it looks very good like you cannot argue with that but right the performance is simply not there because you cannot make up the surface area with this kind of cooling solution you would need so much bigger cooling blocks so if they would adapt to the workstation cooling solutions I think it would be a lot better I mean it also depends on the condition how we push the CPU and the question is always is it reasonable to expect that the CPU would always run it at 500 watt or something like this but I'm sure you tested it yourself and if you check the power consumption in Cinebench r15 the power consumption is very similar to prime95 yeah and I hear a lot that people are saying that like prime95 load is like completely unrealistic but if you compare it for example with with Cinebench and you see it's extremely similar heat load it's very similar and also if you look at something like blender which is another rendering application blender uses a VX just like newer versions of prime do and we were also still burning like more than five hundred Watts on blender so it's a pretty similar power load yeah yeah so if you have an application like Cinebench which is essentially just like rendering and it's not using a VX you would expect that this has to work so that would be my my point that something like this has to work if it's like some obscure prime95 setting and that's the only setting that doesn't work I would be ok with that but there's several conditions where it doesn't work especially when you keep the CPU is quite cold let's say within 420 radiator and just CPU only water block then you can release the power of those chips and you can get like 4.7 4.8 even out of the 18 core which is quite massive but then the issue is if you use this kind of cooling solution then you have no cooling on the BRM so there would be so much more possibilities but as you said yeah like 4.3 4.4 probably what you can get if there is zero air flow and going back to what you said earlier with different situations in different cases I did also some testing in in different cases and it really depends on your overall setup so sometimes you have a case where you for example you mount the radiator of your arrow directly above the vrm and then it's the temperatures gonna be absolutely fine for sure because the the fan will blow directly on the BRM cooler and that's gonna be okay but I don't know in some cases you have no space on top you might mount the radiator in front something like this and then you have a VGA that is not blowing the heat directly out of the case and it's gonna put the heat inside the case and then you run into situations where it's gonna be really bad for the temperature yeah so I'm pretty much covers it our thanks to durbar for joining for this one as always you can go to patreon.com/scishow and access to helps out directly subscribe for more we might look more into these things it becomes a question of do the other motherboards do it better and maybe maybe the apex the apex board should because it has a small fan you can attach to it but yeah that's it for this one it runs hot be careful basically their Barrow was right in his first videos we've effectively validated that at this point so is subscribe for more you can go to store that gamers accessed on net to pick a shirt like this one without the silicon adhesive staying on it or one of our new stickers see you all next time