VehiclesFashionRecipesBlogsHuntTravelsSportFunHandmadeITEducation
Mini-Games
x

x
zakruti.com » IT - Software » Gamers Nexus
AMD Ryzen Heatpipe Orientation Benchmark & CPU Cooler Myths

AMD Ryzen Heatpipe Orientation Benchmark & CPU Cooler Myths

FBTwitterReddit

video description

Rating: 2.5; Vote: 2
we benchmark a series of myths relating to CPU cooler orientation on AMD Ryzen CPUs, featuring Noctua air coolers. Sponsor: MSI GTX 1660 Super Gaming X (Amazon ready to publish, will go live after video Noctua NH-U14S Noctua NH-D15 A while back, people talked about how heatpipe orientation might affect CPU thermal performance for asymmetrical die layouts on non-monolithic CPUs. Given that AMD Ryzen takes the multi-chip approach to CPUs (chiplets, and that its lower-end SKUs only have a single chiplet and single I/O die, the heat sources for these chips would be off-center and imbalanced toward one side. That's where the initial idea arose. The question is whether that has any relevance once you get to the heatsink level, or whether any differences are meaningful (if even measurable. In this benchmark, we'll be looking at how the Noctua NH-U12S, Noctua NH-U14S, and Noctua NH-D15 air coolers perform on AMD Ryzen chips when rotated in their two different directions. The GPU was accounted for and eliminated as a variable, as discussed in the content, and airflow was run from the EPS side in to the GPU side out (with the GPU in the lowest possible slot and as a half-height card. Buy the Disappointment PC Shirt in cotton or Tri-Blend to support us, our intros/in-depth testing, and to commemorate a year of disappointment! These will be a limited run due to the nature of the shirt. We have a new GN store:
Date: 2020-05-06

Comments and reviews: 10


If you guys want to do some more cooling testing, it seems like there are a lot myths out there about what types of fans to use (i. e. static pressure vs airflow. You'll hear things like for a heatsink or radiator you want a static pressure fan, or if you're blowing through a hard drive cage. I couldn't find any real data on this out there, just a bunch of conventional wisdom, so I did some quick tests myself about a year ago. The funny thing is in my own testing a lot of the conventional wisdom doesn't actually pan out. Even blowing through an AIO radiator I found airflow oriented fans to work just about as well on the PUSH side (this is important. In fact, there were very few instances where static pressure fans were clearly better than airflow fans. The one place static pressure did work better was when PULLing through a radiator, where airflow fans (thin, widely spaced blades) had trouble generating enough suction and made a horrible turbulence noise (sounded sort of like a grinding noise, the best way I can describe it. I think the problem is a lot of static pressure fans are too static pressure oriented. There just isn't enough resistance in most cases, even blowing through an AIO radiator, for them to reach their full potential (pull orientation, once again, is the exception where it makes a big difference, at least compared to very airflow oriented fans. I think this is why the Corsair ML fans are a little less static pressure oriented than the SP series fans, since it's just a better design in general for most situations. Blowing through a really thick radiator obviously could be different, but for most radiators and I'd guess heatsinks, you don't appear to need that much static pressure. In any case, it seems like this is one area where few people have done any concerted testing, and it could help a lot to set things straight about what situations you'd actually want a static pressure fan, vs an airflow fan. I think the results would probably surprise a lot of people, seeing as my own limited testing certainly surprised me.
reply

So I have a D15, 3700x, on a MSI B450 tomahawk max inside a Fractal Design Focus G case. And the VRM heatsink and Ram placement paired with the case constraints means the only way I can orient the D15 and use 2 fans, is to have the cooler exhaust up or down. Out the back, I can only use one fan. However, running it that way (my D15 didn't come with hardware to mount vertically out the box) I haven't seen temps that warrant the second fan for my use cases yet. Edit: very much worth noting, for anyone looking at the NH-D15. If you mount the fans as indicated in noctua specs, the D15 could be 165mm tall. However, if you have to contend with RAM sticks and a VRM heatsink, the D15 very easily ends up a cm taller than noctua spec. Pick a case accordingly, or end up with my predicament
reply

a few years ago i posted a pic of my zalman cooler blowing upward into the power supply's fan, boy a lot of people tried to set me straight on that one. well tried, i never changed it cause it didn't put out much heat. haters gonna hate. also, have you done a test with putting a fan below the gpu? i've found an 80mm fits there, and without the covers over the unused slots in the back you can get a good bit of heat blown out there. but would it be better to blow cool air in to the gpu? then the rear top fan would be sucking more heat out. unless you reverse the flow completely and have the front fans blowing out? the only fan i have laying around is a 5500 rpm monster, so a noctua is in the mail because it's only 10 bucks for some experimenting
reply

Yeah I've played with this back in the day I used air cooling, very basic heat rises so when top to bottom whilst that should be what works best the bottom heat pipes dont function as intended as obviously heat will not travel down. So even though front to back works best the upper pipes do more work that the lower I've always wondered why they never made a design that had all the heat pipes travelling upwards as this would surly work much better. Because you have to remember if you set it up top bottom then you are really only useing half of the coolers proformance as the lower pipes cant function correctly as heat will never travel down (it cant) physics I'm told. Hope this is addressed.
reply

Good points, Steve. I think a majority of people who have thermal issues are either from improper thermal paste application or poor case temps overall. Case in point? I cool a 3950x on PBO with a Noctua U9S (92mm single fan) in a thermaltake Core V1. Theres a 200mm noctua fan foe case intake and 2 80mm noctua fans for case exhaust. Under full synthetic stress, the CPU levels off at 79-80. I can shave a couple degrees by turning up the case fans. The cooler orientation has just 1-2 pipes cooling both chiplets. Boost is no issue, as I get cores above 4. 7ghz. Oh and I'm not using paste. just an IC Cooling graphite pad.
reply

Cool video, and expected results. That said, I was under the impression that the heat pipe orientation was only a concern on CPUs with two chiplets, and it wasn't about rotating the cooler so so fans went top to bottom, but rather it was that if you buy a cooler with pipes that run vertically in a back to front fan orientation, you effectively only have one heat pipe running over both chiplets, where if you buy a cooler where the pipes run horizontally across the IHS each chiplet would be under a different heat pipe. I don't see how orientation on a single chiplet CPU would make any difference either way.
reply

The only problem with these aftermarket coolers is that they're too big for older mini-towers from what it looks like. Not to mention the fact that there's more things that the CPU fan cools than just the processor (e. g. Hard-Disks) For another thing, air from downward-facing heatsinks (e. g. Amd's stock Wraith coolers) spread across the board through holes in the fins/pins whilst hitting the surface of the sink, hence the fan blowing down. It's sort of the reason why all stock coolers have fans pointing with it's back to the sink.
reply

Reddit is a dumpster fire these days. It's full of this sht on every possible subject, stabbing blindly in a dark room. People who know the least tend to speak the most about something. They like to hear themselves talk. That and the idea that Reddit is chock full of dead, locked threads with outdated information that should be avoided. They want you to start a new thread, but don't like when you base it on a previously closed subject. Kind of like Tom's Hardware with most of their forum posts over 5 years old. Useless.
reply

Kinda a side topic here: The technical aspects of wicking or the flow of the gas/liquid inside the heat pipes is something I havn't seen exposed in detail. Neither of these orientations take into account the effects that gravity might have. Try setting the case where the heatsink tower points straight up vs. straight down maybe? Also are all heatpipes created equal these days? Some of the older ones were different and didn't use the sintered copper design if i'm not mistaken.
reply

First of all: are we sure that AMD uses only one chiplet in those CPUs? The whole architecture has the advantage of literally glueing together the CPU and potentially increase yield. I. e. they can use two chiplets with defective cores to make a working CPU instead of throwing them away. Why wouldn't they do that? I'd guess that physical heatpipe orientation has much more effect - after all heatpipes are affected by gravity - as well as physical fin and fan orientation.
reply
Add a review, comment






Other channel videos