When You Should Care About VRM Temperatures (Ft. Der8auer) 

Article: www.gamersnexus.net/guides/3113-asus-rampage-vi-extreme-vrm-thermals A reminder that, again, the big question is when it's reasonable to expect the VRM heatsink to be able to handle the heat load without help. This is not necessarily only a fault of VRM cooler design, but also a byproduct of the fact that i9 CPUs suck so much power through the rails. The two together mean that, if overclocking i9 CPUs, you should be actively ensuring there's cooling over the VRM. Here's Der8auer's channel: ru-vid.com

Thanks for having me in the video :)

Kyle, the real authority on bottlenecks.

Out of the box vrm thermals

I really hope we can get some practical motherboard designs in the future, rather than just "gamer" ones.

I measured all the VRMs on my gear out of the box and they have never been above ambient temperature. Nothing to see here. 😎 (Sarcasm doesn't translate well)

I have never found your independent investigations redundant and have often found them critical to my understanding. Thank you for your work.

Back in the core 2 days EVERY single high end board I used came with vrm and or north bridge fans.

Really a mobo at that price should have better cooling, it's concerning that vendors create boards while being so focused on how it looks, and not validating the actual performance

I think it's way past time that active VRM cooling became an industry standard, certainly on platforms that are being targetted at overclockers. Meanwhile, what's the best solution? My takeaway from this and der8auer's video is that, first of all, *never use a front rad,* and second of all either have a top case blower oriented as intake, or use a down blower on the CPU, or have LC or a side blower on the CPU but a separate fan "mounted on a spike" directly over VRM. And again, all of this depends on having cold, unimpeded intake from the front. It also seems that all current VRM heatsinks are more like thermal _insulators,_ and we should be removing them as a matter of standard procedure. The problem with top intake is that surely it's just going to get sucked straight out of the exhaust, and won't get much chance to hit the VRM. And even with the exhaust off completely, because the top intake is perpendicular to the board, and the VRMs are relatively low profile, any air is going to just skim right over the top of them. A down blower on the CPU will certainly pass air over the VRM, but it'll be hot air by the time it gets there, so again that seems counter-productive. Also, is overclocking even possible with down blowers? Sadly the only solution that seems to make any sense is the "fan on a spike" idea, and even then I'd want to feed it cold air from a side panel intake, which pretty much excludes all tempered glass cases, and actually excludes basically every modern case on the market, with the exception of a single Corsair model I can think of. While we're on the subject of inadequate cooling, maybe Steve could include a brief analysis of HDD cooling in modern mid-tower cases, or rather the total lack thereof. The general assumption seems to be that nobody uses HDDs any more, so the fact that they've been shoved into the PSU shroud with zero airflow is moot. That's great if you're a millionaire and can afford 2TB SSDs, but most of us can't, and meanwhile a slightly more affordable 250GB SSD is only just big enough to hold maybe 4 triple-A game titles (at an average of 50GB each) plus the OS. I just find it funny that in 2017 the only way to get a PC that actually works properly is with a Dremel.

bring back finned VRM heatsinks

If motherboard makers where to actually use heatsink fins instead of the mostly decorative blocks of aluminum like they been using for a few years now there wouldn't be any issue. I guess they were use to CPUs not consuming so much power that having a non-optimal VRM cooler wasn't the worst thing in the world so they could be more artistic with the VRM coolers that is not so much the case now

11:30 always easier to grasp while seeing them all at once. Thanks for the chart

I always know that VRM temperatures matter. But it's super rare to see a review that mentions VRM temperatures, most of them just show you pointless benchmarks in which the reviewed board's performance is only a few percent higher or lower than its competitors. Like right now, I want to build a Coffee Lake system and the Asus Strix Z370-E is in my budget range and has features that interests me, but I can't find a review that says anything about its VRM temperature. WTF ??

What about passive aggressive cooling?

so i guess..... EKWB monoblock for the win ! :D

"Silicone adhesive"

"When You Should Care About VRM Temperatures" "This video is brought to you by EVGA" Uhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh

Don't forget to cool your vCore wires.

It bothers me that nowadays they just plop those fat, monolithic blocks of aluminum with heatpipes underneath on the VRMs. This is stupidly small surface area. I miss good old razor sharp fin arrays, they didn't look as flashy, but actually performed like a heatsink should.

@20:00 I feel damn glad that I bought an Asus X99-e WS workstation board for my custom loop cooled and overclocked 5960x after hearing De8aur reference the Asus X99 workstation boards as his standard for excellent cooling of the VRM’s. I was aware of VRM’s and spared no expense on the motherboard knowing workstation boards are more stringently tested. I didn’t know the X99-e WS boards were THAT good.

Intel a few months ago: AMD is gluing together their server CPUs. Glue is terrible, just horrible. It results in unpredictable performance Intel today: We are going to glue AMD GPUs onto our CPUs because glue is amazing. Hell, we are going to glue our entire processors together! We love glue! You have to wonder if there were a bunch of Intel engineers that knew this was in the pipeline that just facepalmed as hard as they could when they saw the infamous "glue" marketing slide.

Great video. The attention to detail and the professionalism in which you handled the topic is remarkable. This is something that I don't see being discussed anywhere lately. I already had this kind of problem with my AMD fx8320 and my MSI 990FXA-GD65 v2 motherboard. I can only put the cpu @4ghz without having VRM temps going 100°C+ and making my cpu throttle HARD. As soon as the VRMs are actively cooled, I can go as far as 4.4ghz on the cpu.

one of the more interesting videos. Putting aesthetics first is a problem at every level of PC construction (parts design) and has only gotten worse. Think thermals first and choose your parts carefully!

Could you make a video explaining the Intel + Radeon partnership

I have reversed the rear fan as intake and it has worked as long as thares top fans and or airflow vents

I have a BeQuiet case and found my best options for air-cooling (I leave it processing 3D 24/7) was to put two fans blowing in from the basement, two 140's blowing in from the front, and three 120's blowing in on the top. I let the pressure blow everything out the back. All incoming is fine-mesh filtered and I have a HEPA filter blowing into the corner to push clean air in and push the warm exhaust out of that area, making a high pressure area around the desk to keep dust down. That's all noise though, but when I turn the fans down and the big HEPA off, everything's quiet for recording. Temps dropped perfectly since I did that. I use a BeQuiet big cooler and it's on a 5950x. VRM's on the B550 Unify-X run cool when not overclocking, in the mid 50's with the fans rarely coming on and still processing.

Glad you guys shedded some light on this. There are many people that claim they're some short of experts but in reality they're full of crap. Thanks again for the video.

Steve/GN, this comment may not get a thousand likes but I still wanted to say that I really appreciate what you are doing here, truely informing the world and always giving a critic point of view on everything, always based on tests. I hope you get the best, and never sell out! The world needs this :)

This was kind of new topic for me. I was happily overclocking my CPU to 4,7GHz. However in stresstest VRM temps rocketed to 100C, which didn't make anything to throttle but I didn't feel too good with those temps. What was surprising was I was forced to tune down the clock to 4,0GHz to limit VRM temps to 90C in 2 hour stresstest. I'm not too familiar with VRM temps but somehow it feels that maybe around 100C is max on continuous use. So next I need to be looking some VRM cooling.

Shameless Karma grab for "OUT OF THE BOX THERMALS"

Radiator in front is the coldest solution means getting best CPU temperatures and hot air in the case is no problem at all because its a continuous flow ... for the GPU u can add a case fan to the bottom that bring cold air straight up.

This VIDEO wasn't originally in the plan , We briefly talked about clock throttling due to VRM but we still made it of 24 min.

I made the mistake of buying a Aorus B450 M and now I am looking at aftermarket VRM cooling solutions. Thanks for the vid GN.

I remember Boards like Gigabyte p45 extreme... A friend of mine had it, and it had really sick mosfet and chipset cooling. You can google it. Nowdays Boards are kinda shitty with that. Form over function i guess.

Some informative stuff from both you and Derbauer...

We used to have VRM sinks with real fins an surface area, rather than just hunks of aluminum. Last one i owned was an Asus P6X58D-E had a vrm sink with real fins. Many of these new boards have just hunks of metal and no real surface area, My MSI X370 Gaming Pro Carbon is one such board with just blocks of metal for a heatsink. A block of metal is not a long term heatsing.

Interesting. I'm a design engineer so there's always a battle between the engineer wanting to design something amazing that will last forever and the accountants/sales strategy teams who just want the item to function well enough for the next few years before a new generation comes available at which point they want your current bit of kit to become outdated and redundant so you go out and buy the newest, latest and greatest... Am looking at installing a AMD 3950x onto a ROG B450-F motherboard, replacing the 2700x CPU currently installed. I want it to work, but also want it to last. At first glance the TDP of both CPU's is identical at 105W so I assume this means the VRM's should not get any hotter with this newer CPU? (same number of VRMs delivering the same amount of power... .although the 2700x seems to work at 1.2V whereas the 3950x seems to operate at 1.4V. As P=IV and power is the same this should mean the higher voltage flowing through the MOSFETs causes a lower current to flow when used with the 3950x. The graph on the following page, (not convinced it is correct as would expect the power loss curce to be more of a U shape... effectively an inverse of the efficiency curve) seems to imply that a lower current flow produces lower power loss and so less heat... which seems to imply that the VRMs "should" actually run cooler with the 3950x than they do with the 2700x... www.ekwb.com/blog/what-is-mosfet-and-why-should-you-keep-it-cool/ So if that is the case, I'm struggling to understand why all the pro's are recommending the 3950x needs a much better motherboard due to VRM temperatures.... UNLESS motherboard manufacturers are purposefully programming the BIOS to use the MOSFETs in a way to actively generate more heat in the VRMs, because they do not want people being able to use newer CPU's with older and cheaper motherboards, ebcause they want you to buy their newer more expensive motherboard which "works betetr with new tech"......... Do we think this is actually what is happening??? Also, water cooling seems like the best way to go to easily cool the MOSFETs but for the added cost you may as well just buy a higher end motherboard in the first place.... but I'm assuming the current passive heatsinks can be made more thermally efficient by replacing the thermal paste "bonding" them to the MOSFETs for a higher efficiency thermal paste... does anyone know how much cooler the MOSFETs will operate by doing this? Have just purchased an IR thermometer so might have some fun with this tomorrow :) PS: I really like the more scientific approach you have in your videos. Perhaps you can do a video on this if it helps people out... although if that is what motherboard manufacturers are doing then it's probably quite easy for them to program into the BIOS to have a VRM temperature setpoint for whichever CPU is being used instead of a voltage setpoint so as to effectively bypass the additional efforts made by anyone trying to improve their thermal efficiencies :(

Years after this video and things have only gotten a lot worse, I just had to give up on a Z490 motherboard because Asus in all their wisdom decided that the Maximus Hero XII(a board that retails for over $500) should have a plastic box over their chipset heatsink. It forces the heat built up in there to escape right out under the two NVME slots there. Brilliant!!!!! If you want to slow cook your chipset and force the customer to buy a new part in a year or two this is a brilliant tactic. I can't think of any other really logical reason to design the board this way. They literally put a heatsink on a chip and then put a box over it with no design for airflow. Its terrible. It seems like so far only Gigabyte is putting a fan on their VRM and a handful of AIOs like those from Arctic have fans on them to blow at the VRMs.

Trying to stabilize my FX-9590 for overnight Handbrake encoding, and I think I'm overheating my VRMs. As usual, you've got an excellent video on the topic. Thanks a bunch!

And people wonder why I still use top down cpu coolers. My cpu may be a couple of degrees hotter, but the overall system temps are lower and the stability is higher.

Based on my experience with my 7700k this was a problem with the Z270 boards as well.

My threadripper monoblock is coming today, sure wish I would've had the chance to buy it stock like that. I reached out to ek and they gave me a quick response via email. Great customer service EK.

You should give your video series about thermals a name? What better name than *OUT OF THE BOX THERMALS*

I see something similar in tech yes city when he use a b350m-a which have no heatsink on the mosfet, and show the huge difference of performance with a ryzen 7 using heatsinks with a fan over it; I think at this point Mobo manufacturers should start using active cooling solution over mosfets like you see on CPU laptops...

Check out the Heat on the back plate of the CPU. Now that is hot. It seem like there should be a heat sink on the back plate of the CPU.

a bit late but anyway, change thermal pads and see a huge drop in temps, even top motherboards have mid range thermal pads installed. No point to have huge heatsinks if heat can't be transferred on them fast.

I used to have an Asus TUF Sabertooth Z77 and I had one of the down draft air coolers by Noctua. I still have the cooler but not the mobo. Conclusion it was the best cooler for that mobo.

If an aircooler - like noctua - is placed to blow UPwards and in top of the case would be a case fan, the vrm would benefit. But seriously... For the price these mainboards are sold, they better be solid copper with much more fins. Or many more vrms to spread temps.

You should test the Cryorig A40 liquid cooler. It has a fan for VRMs.

Glad this was covered. I've been having this problem with this issue for years through the few motherboards that I have owned. Mobo mosfet would always reach its thermal limit before CPU would reach its limits when OCing, even on my now current Ryzen setup with a Gigabyte AB350 Gaming 3. Been trying to find an elegant solution for this. Been thinking of doing a custom Liquid loop with a mosfet block when I can but the top mount intake fan sounds like something to try.

Great video guys !!! Keep up good work !

My personal thought is. Especially on these humongous sockets. AiO liquid coolers should employ a design that would include a 5V fan over the socket itself together with leads to the radiator which would be mounted elsewhere. For custom water cooling you can always get the ones that cover VRM's also and IMO, they look cooler. (no pun intended). PS: I remember back in the day when coolers actually came with their own passive larger heatsinks for other things on the motherboard just for the same reason. Same goes with those old Zalman GPU ones that had the memory heatsinks seperate. Anyway. It's a thing to work on. And it's not easy (if it is to look good also) becasuse coupled with massive sockets you also get a ton of RAM + nvme so pretty crowded space above the first PCIe slot.

I only care about temps when they hit .1% and 1% highs.

Seems like the Cryorig A40 would be a good choice for overclocking here then

IMO it should work. If passive cooling isn't sufficient vendors should be including fans

Steve, I was half expecting you to bust out rapping after reading the title (Ft. Der8auer). The chorus was going to be "Out of Box Thermals, Out of Box Thermals" Hahaha! Keep up the great work.

Enzotech mos-c1 heatsinks work great

16:25 Stick a fan on a spike and put it right over the VRM's. LOL

AIO and get one that lets you put an L bracket on one of the fans rear or top so it cools the VRM directly

I'm gonna call you out on one small detail. A clc* will not fix your problem, not liquid cooling. In a custom loop, you could add a monoblock that covers the vrm's, and 35-40 watts of heat in a custom loop is nothing.

I've been waiting for this video.

NG is amazing im going to use a old GPU only block & need to know vram temps i now know exactly what ill need on ram now

The highest power consumption process I've ever found for my setup with an i7-4790k and a 960 is running FurMark's benchmarks... Even running FurMark (not the benchmark, just the stability test) AT THE SAME TIME as aida 64's CPU torture test is LESS than just running the FurMark benchmarks... In fact, this helped me diagnose that the cheap ass knock off brand supposed 700 watt power supply I've been using for years actually would crackle and die... at only 350 WATTS.... Switched it with a 750 W corsair psu that's like 10 years older than it and has a ton of dust built up inside it, and besides now not crackling and dying when I run the stability benchmark, it also consumes only 300 to 310 watts max when the old power supply was pulling 350 to 360 before it died...

I recall some of my A-Bit motherboards having heatpipes, as well as their Gigabyte competitors. My old EPOX EP9-npa+ sli had active cooling on the northbridge/southbridge (don't recall). That stuff went out of the window fast.

What about replacing stock VRM heatsinks with something that has larger surface area (even stolen from sole older boards)?

I was pleasantly surprised that my Strix-F Z370 came with a little bracket to mount a 50mm fan above the VRM's.

They should go back to cooper heat-sinks. I remember my X58A-UD7 had a add-on for VRM/NB cooling because they ran hot.

Great job guys!

Thanks for another great content

You know, my old A8N32-SLI-Deluxe had a solution to this, they called it the Q-fan.

watching this in 2023 makes me happy about my i7 7700k runing 5GHz for years at 70c core and 83-89c VRM

As ever, really informative and well-done video, thank you GN!!! I am running a Rtzxen 9 5950X on an Asus Crosshair VIII -- OC over 4.5 is getting way too hot ... so I am replacing my Noctua HD15 for an EK custom loop - looking into EK products, I see they have (for more or less the same hi price) a fancy CPU water block and a quite nice Mother Board water block ... designed for the crossHair - I was going for the fancy CPU water block, but after watching this video. I start to question the choice. Will, it better to put a water block that will cool the entire Motherboard? Or will the temperature difference between the CPU and VRM, increase the overall temperature of the entire system? what do you think?

So im guessing my top mounted H100i pulling in is pretty bad for my VRM on a z370-e, going to swap those to push out and mount a 50x50 fan on the mount asus provided and hopefully that will keep me cool enough for for a 5ghz OC.

An excellent video by the way!!!!

I learned a long time ago by using the Asus Sabertooth Z77 mobo that the downdraft cooler would be best and at stock everything I am reaching from the 20's to the 30's in celsius

84 c above ambient.... wwwwwwhat also please make a t-shirt 'GN approved - not terrible'

BUT, sometimes cooling of the VRM is not the case. Been using 7980XE (for many many tests), even if you keepyour CPU about 60 degrees, or even below, AVX512 used functions/apps limits you to 3.9 GHz maximum. This is the naked truth.

Well if Asus desktop motherboards are anything like Asus laptop boards, it's not suprising to hear there's issues with the VRM

EVERYONE LAUGHED AT MOTHERBOARD WATER BLOCKS

While I do feel like Asus should have come up with a VRM cooler that was capable of keeping the VRMs cool, I feel like if I were building a PC with a $2000 CPU and a $650 motherboard I would go full custom watercooling with a block on the VRMs and CPU and that would solve all of the problems discussed in this video.

I would be really interested if VRM temps could be solved by using high power horizontal CPU cooler(blowing straight down to the CPU) instead of vertical one(front to back/etc). I can't find anyone who has tested this yet.

EK has mono blocks that cover the VRMs which is the best alternitive

I have a new 8th Gen i7 8700K system with an ASUS Z370 A Prime motherboard. It came with a bracket to hold an added 40mm fan over the top left corner of the motherboard. I used the bracket to hold a 40mm Noctua fan (no room for anything wider) but have not run any benchmarks. I had to ask Noctua support which way the fan should face - up or down (they said down) because ASUS provided no info on that. ASUS also offers a 3d printed bracket "file" to print a bracket for another small fan for the lower left corner area over an M.2 slot that is covered by a metal plate that presses on the M.2 NVMe SSD to radiate heat. But I have not found a source to "print" that bracket and ASUS didn't even provide an illustration what it looks like or where it goes. I use my system in a large case with a front and top fan and Hyper 212 Evo CPU cooler and a back case fan, and temps usually stay around 30-80 C per the Core Temp utility. (System built Dec. 2017) This system is used for periodic video editing and I am not a Gamer, so my GPU is a simple one just to run 3 monitors.

If the core temp of the planet went up 100 degrees would we even see a 1 degree difference on the surface. Extreme example but one worth considering. Which raises the question how much of a temp difference within a given component before you can measure a difference on the outside. This depends I'm sure.

asus with $350 X399 Prime A: "we actively cooled vrm heatsink because Threadripper will draw a lot of power, and we want to make sure it will run it safely and efficiently". asus with $650 x299 Rampage Extreme 69: "lol nope".

oh... as soon as I was typing about top air intakes or exhaust you talked about it.... this is why i love this channel. Way smart.

My ~10 year old Asus boards have heat sink fins, a phase change pipe and active fan.

Someone should make a liquid cooler with like an 80 mm fan above the CPU cooling plate/water pump that has an adjustable slant for aiming at components next to the CPU socket.

Asus just keeps the same (or at least nearly same) vrm for the rampage series. Looks like they just didn't feel like changing it as it woul'd be needed due to the raising power consumtion

Still really wish people would stop calling an overclock that passes a run of Cinebitch "stable". _no, no it is not_

Too much knowledge, I cannot handle it. I will add some exhaust fans above vrm. Game over !😆

They need ducting, I've had computers in the past where i've put a couling over 1/3rd the rear fan and it worked, many servers do the same where they manviour air through the case.

My new mobo (Z370) has a little fan under the IO cover, presumably to cool the VRMs, hopefully it's enough to counter the fact the heatsinks are mostly just there to support the RGB lighting!

You know what? I watched this video in 2017 like almost one year ago and Now I AM WATCHING THIS AGAIN. THE INTRO AD IS "WOW" 750dollar GTX1080Ti. WOW. Where can I get one? ahahaha

Or you could use 4 x 3000rpm pwm gentle typhoons in push pull on your radiator in the front of your case & 6 x 2000rpm nb eloops exhausting out the top rad & 2 more nb eloops 1 exhasting in rear & one intake in the top front, like me (caselabs sm8). Problem solved. Or one could continue to wine about how quiet they want their pc to be & burn up the vrms

just buy the updates boards like the x299-xe gaming from asus (the updated version of the x299-e). everything is running good now - even with a bit oc still no problem

Hmmm... How about reversing the flow? Make a front mounted radiator exhaust air and move the air intake to the top or back? Might be interesting, although I dunno what would happen to the GPU airflow.

So they should be just fine if you are running a 7900x with a mild OC at 4.5GHz and 1.225v. Your results are with the most extreeme situations (and I fully understand why you did that). I would like to see how it works with the above mentioned settings (give or take). @gamersnexus

I wished I had watched this before buying my last motherboard 4 years ago...