Not sure I'd bother on a 1080ti unless you were getting it used on the cheap. A quick look into the comparisons between it and a 2070-2080 super seems to indicate it's in about the same range for performance. I say 2070-2080 super because I don't know the accuracy of the results from the quick look. It gets further convoluted depending on the type of each card you're comparing.
The 2070S and 2080S are $500-600 and $700-800 cards, respectively (for my region of the world at this time, anyway). $700-900 for the older, equal or less performing card isn't a great deal. If you could get the 1080ti for less used it might be worth it. There you're taking a risk on the used card though.
The performance part is why I made the 2060 super getting console FPS jab. Granted, it was meant to be a joke.
RT is undoubtedly an interesting concept. I can't help but think it was pushed out a bit too early though. Even though there may be some logic behind pushing it out before hardware is fully ready to leverage it as a testing phase. Most people are likely aware this type of improvement doesn't happen over night. Odds are by the time you hear about anything new it's been worked on for years or months prior to being announced.
Getting to the point, I can't help but wonder if this improved iteration of DLSS isn't a method to mitigate the performance issues RT was going to lay on the hardware. After all, if you slap RT on a card and hype it up people expect to use it. If that RTX 2060-2070 user enables it and watches their FPS tank, when they were paying for the feature, it's not a good look. Toss in the shortcomings of earlier iterations of DLSS and connecting the feature to RTX cards and, three birds meet one stone (more likely at least half a dozen birds but this is beside the point).
The 2070S and 2080S are $500-600 and $700-800 cards, respectively (for my region of the world at this time, anyway). $700-900 for the older, equal or less performing card isn't a great deal. If you could get the 1080ti for less used it might be worth it. There you're taking a risk on the used card though.
The performance part is why I made the 2060 super getting console FPS jab. Granted, it was meant to be a joke.
RT is undoubtedly an interesting concept. I can't help but think it was pushed out a bit too early though. Even though there may be some logic behind pushing it out before hardware is fully ready to leverage it as a testing phase. Most people are likely aware this type of improvement doesn't happen over night. Odds are by the time you hear about anything new it's been worked on for years or months prior to being announced.
Getting to the point, I can't help but wonder if this improved iteration of DLSS isn't a method to mitigate the performance issues RT was going to lay on the hardware. After all, if you slap RT on a card and hype it up people expect to use it. If that RTX 2060-2070 user enables it and watches their FPS tank, when they were paying for the feature, it's not a good look. Toss in the shortcomings of earlier iterations of DLSS and connecting the feature to RTX cards and, three birds meet one stone (more likely at least half a dozen birds but this is beside the point).
More likely - more hardware used for ray tracing, less space for regular compute units. Naturally you'd pay for that with lower performance of everything else. That could prompt Nvidia to start resorting to tricks to mitigate it.
I think also you have to at some point release it in to the wild to know what the next step in refining the performance is and for Dev's to get to work on their side.
the "subtle" nature of the current implementations is largely to do with most being back engineered solutions rather than being built with it in mind from the beginning. it is very early days, and we haven't gotten AMDs version of this yet, but i don't think this is going anywhere, change a lot? sure, much like Phys-X went from being extra cards to being integrated on to the GPU. buying for it now? probably not needed, specially with Nvidia's pricing.
the "subtle" nature of the current implementations is largely to do with most being back engineered solutions rather than being built with it in mind from the beginning. it is very early days, and we haven't gotten AMDs version of this yet, but i don't think this is going anywhere, change a lot? sure, much like Phys-X went from being extra cards to being integrated on to the GPU. buying for it now? probably not needed, specially with Nvidia's pricing.
So anyway.... Back to building a gaming PC. Anyone have recommendations for air coolers to go with ITX builds? I'm thinking a D15 might be too much of a big boy. I'd doubt the second fan would be workable even if it did fit. The case specs list 168mm clearance for the CPU cooler. The board in the system is a Asus Strix z390-I. I'll probably look myself but was wondering if anyone already faced a similar dilemma.
I was using a x62 Kraken but since CAM won't play well with others I decided I want to switch the cooler (Noctua fans off the kraken won't spin unless set to 100%, no joke). It was a consideration previously just because CAM is garbage but this behavior was the tipping point. For the time being the kraken was kicked over to the AIO header on the board so I can marvel at how it cuts off 700 RPM on the pump for not using bloatware.
I was using a x62 Kraken but since CAM won't play well with others I decided I want to switch the cooler (Noctua fans off the kraken won't spin unless set to 100%, no joke). It was a consideration previously just because CAM is garbage but this behavior was the tipping point. For the time being the kraken was kicked over to the AIO header on the board so I can marvel at how it cuts off 700 RPM on the pump for not using bloatware.
I'm a big fan of Cooler Master. Good price, long-lasting, they tend to run fairly cool without additional "tweaks". The can be a bit louder than some, and I've also gotten myself in trouble a couple of times when the cooling tower was...like...hairs too tall for the case to close. Give yourself at extra inch or so on all sides.
Well, the reason I asked is I'm not sure if I trust the compatibility list for the popular large tower coolers when it comes to ITX boards. There is a huge heatsink/IO cover over the VRM/IO on the board and ram nudged right up against the socket to contend with. My concern was one or the other would interfere with the tower part in the tower coolers or the heatpipes. After measuring a few places I believe they can fit. Some of those CM tower coolers do appear to have a slimmer profile though.
In the end I'll probably just stick with the Kraken for now. I ended up moving the pump tach to the AIO header, fans on CPU/chassis headers and made CAM disappear. It costs 700 pump RPM but this may not be very relevant. Besides, I'm fairly sure the CPU cooler is bottlenecked by the die > toothpaste tim > IHS (non-delidded 8700k).
At some point I'm upgrading to a ryzen setup anyway. This will require a new case because tiny boards are a pain in the ass to deal with. So, back to atx..... Plus, I may just wait until ryzen 4000 releases and has been out for a month or two.
Horizontal-based heatsink Noctua, i.e. Noctua L-something and you're all set.
Aren't those the ones that blow straight down? I always had the impression that style of cooler was.... sub-par. Intuitively, I wouldn't think blowing air straight at a fin stack on top of a CPU would be all that effective. For a lot of reasons. If you had major space limitations it might be the only viable option. Otherwise....
For clarification, the situation is as follows...
Current: ITX board, MATX case, AIO cooler
This current situation of using a larger case with an ITX board came about to provide room to grow, so to speak. The option to swap to a MATX board eventually if a tiny build was deemed to be undesirable going forward. The AIO cooler was selected because they tend to be lowish profile around the CPU socket compared to large air coolers.
Future: ATX board, smallish ATX case (cheaper, more space efficient, less necessary work to dictate airflow), one of the best available air coolers.
This future plan is based on a couple considerations. As noted, ITX is a pain in the ass to work around for a long list of reasons. Of equal importance, I've noticed AMD doesn't seem to place much emphasis into their MATX offerings (would have to dig deeper to fully confirm this but at initial glance it looks to be the case). So I'm leaning toward ATX for this future upgrade. This is the price I pay for not using AMD hardware for a decade.
Additional PCIE slots, more IO and similar functionality isn't of concern. Most boards have way more of this then I need in the first place. The robustness of the VRM is a consideration. Oddly, some of the current AMD ITX boards are better in this department compared to many of their MATX offerings. Toss in apprehension toward older chipset boards with the new Ryzen chips and it points me toward going ATX.
All of the above brings me to large tower coolers on ITX boards, assuming the case permits the tower height (it does). Granted, part of it was based on in the moment rage at a popular AIO failing to work properly with a fan upgrade and trying to force it's bloatware upon those using it (software must read the fan RPM or something and be like, nuh uh...). #nzxtcangofuckthemselves
The initial question was posed because any type of online search gravitates toward links to the motherboard, links to ITX builds in tiny cases, reviews of both, links to random youtube videos of "content creators" with very little clue (so, most of them), etc. Take note, "large tower cooler compatibility with ITX boards" isn't on the list. Incidentally, this prompted an adventure to physically measure potential problem areas of the board in question and compare it to tower cooler dimensions. Fortunately, the tower coolers actually provide useful documentation.
The end result is it would appear the answer is yes, even some of the largest tower air coolers can function with an ITX board. Provided the case dimensions support it. You might not be able to see the board once their installed but... who cares?
The only other consideration at this point is finding a suitable smallish ATX case to support the future. Ideally one designed around cooling the things inside of it properly. So, there goes 90% of the cases right there (more glass, more RGB, more choked off front panels... case designers are evidently good at making dressed up boxes). Preferably one with an inverted layout because it's like heaven for GPU cooling. Unfortunately, those all appear to be too small, huge and/or expensive. So I'll probably have to concede on the last bit.
Apologies for the long post. I'm not intending to be aggressive toward you or anyone else, for the record. In the end all the above text is probably pointless. On the off chance someone else has a similar dilemma....
In my experience, it's pretty rare for a cooling solution from a reputable company to have seating issues with a mo-bo from an equally reputable company. That's mostly going to happen with new companies or "budget" boards and stuff like that. What I was talking about was the case not able to close properly because the CM heatsinks can be pretty tall. And one definitely does not want to any part of the mo-bo to be pressed against the case in weird ways. That added pressure with the constant heat fluctuations can pretty quickly result in physical stress and damage to the board. (And nothing pleases more than putting the finishing touches on the system...then realizing I can't close the case...
)Nothing wrong with the Krakens, though. I've never used one myself, but I've not heard any complaints about them. Also, I've not ever personally built a system with liquid cooling. Can't say anything there.
To avoid clearing issues, I just stick to big tower cases. It helps ventilation too.
Well, I'd assume the boards with the same socket type would be roughly the same in terms of positioning, for the most part. Standards are a thing. ITX boards are always kind of squished together though. I wanted to be sure.
There is nothing wrong with the performance. The problem with them is you have to run the CAM software for the pump to run at it's maximum RPM. Otherwise it runs slower. The second problem, and the reason it drove me up a tree, is I tried to run Noctua fans off the Kraken and they wouldn't spin at all unless pushed to 100% RPM. Initially I did a double take to ensure fans were plugged in correctly. Then I thought I bought duds or something. It took a couple minutes to figure out what was up.
It's as if the CAM software is only able to run fans with RPM values equal to it's included fans. It's almost as if they intentionally try to force people into using the CAM software. This kind of... behavior... is getting really old. Boards come with bloatware. GPU's come with bloatware. CPU coolers come with it.... All of this on top of the garbage Windows runs. At this point I've decided any hardware trying to force bloatware isn't even going to get a look.
There is nothing wrong with the performance. The problem with them is you have to run the CAM software for the pump to run at it's maximum RPM. Otherwise it runs slower. The second problem, and the reason it drove me up a tree, is I tried to run Noctua fans off the Kraken and they wouldn't spin at all unless pushed to 100% RPM. Initially I did a double take to ensure fans were plugged in correctly. Then I thought I bought duds or something. It took a couple minutes to figure out what was up.
It's as if the CAM software is only able to run fans with RPM values equal to it's included fans. It's almost as if they intentionally try to force people into using the CAM software. This kind of... behavior... is getting really old. Boards come with bloatware. GPU's come with bloatware. CPU coolers come with it.... All of this on top of the garbage Windows runs. At this point I've decided any hardware trying to force bloatware isn't even going to get a look
.
Hm. That doesn't seem right. Are you setting them with software, or did you try managing them using the speed curve settings through your BIOS / UEFI? Frankly, if I couldn't manage them through even the BIOS, that would be enough for me to contact the manufacturer and troubleshoot. Or just get a refund. I mean, they're still just system fans, so they should be managable like any other fan. (If they're physically built into the unit in an "on/off" way...that's stupid. I'd call that a design error.)
I did that only once when I was planning to build an "uber-system that I could just upgrade forever". (
Oh...what a silly boy I was.) I found the case being too big meant poorer air flow. It ran hotter than I expected, and it got dirty really quickly. All of that combined built up a lot of heat in short order, I think. (Plus, I only updgraded the video card, RAM, and the processor once before the entire thing was more or less obsolete. I never took any advantage of that case...)
Depends. Noctua coolers tend to be better than even most budget vertical-based heatsink coolers.
Hmm, I'll try to explain. So the Kraken unit itself has two connectors coming off of it. One runs to a USB 2 board header. I believe this is so it can communicate with the CAM software. The other is a sata connector with three sets of wired connectors split off of it. One goes to a sata cable off the PSU. Presumably to power the pump and fans. The other splits off into four 4 pin fan connections for the Kraken radiator fans and two additional. The third is a pump tach. I suspect this is for providing the pump rpm reporting to the board (it's a 3 pin connector but only has two wires, so one is blank... I suspect those two wires are for PWM and RPM).
The Kraken setup directions specify to plug the pump tach into the CPU fan header. Again, presumably this passes PWM to the fans, since they're split off the same cable, the pump (it's PWM too) and pump RPM reporting. The fans and pump are powered by sata, or directly from the PSU.
Originally, this is how I had it installed. The fans on the Kraken rad were running off the Kraken fan splitter and a few case fans, with another fan off a chassis fan header. Those other case fans were nzxt fans since it's also a nzxt case, unfortunately. This worked fine with the CAM software. Although, I'm not keen on needing to have software connected to the internet, requiring a login and somewhat consistent updates to it for a CPU cooler (what are the odds an update breaks something?). In this config the pump runs at it's rated 2700 RPM and fans are adjustable like any other PWM fan.
I decided to replace all the fans with Noctua fans. I set them up in the same configuration, booted the PC and none of the fans off the Kraken splitter spinned. This got me wondering if I forgot to plug something up. I checked that and everything was setup right. Here I was wondering if maybe when reinstalling the Kraken (I picked up a new case, returning it though) I was a bit rough with the wiring. To rule out damage to the Kraken connectors I hooked up a fan splitter cable to the chassis fan header and put the fans on it. Sure enough, all of them worked fine.
I think I did a quick search and saw something claiming the CAM software doesn't like off brand fans. For curiosity, I decided to move them back to the Kraken splitter and force them to 100% RPM via the CAM software (the earlier step rules out the fans being bad, not the Kraken connectors). Again, they started up and worked fine. Evidently, the Kraken doesn't play well with off brand fans. I doubt it's sophisticated enough to magically know they're off brand fans. Hence why I said it seems more likely it's related to the fact they run at different RPM from nzxt fans the Kraken comes with by default.
In the end I just moved the Kraken pump tach to the AIO header, the fans off CPU fan and chassis fan headers and physically isolated the fans from the Kraken. They're completely controlled by the bios now. The Kraken fan splitter piece just got stuffed in a corner in the back of the case. Everything works fine this way. With the lone exception being the Kraken pump loses 700 RPM because without the CAM software it won't run at 2700.
Yes, the way this all works out is pretty stupid
. It's why if someone asks me about NZXT AIO's my immediate response is, "Look elsewhere.".
Post automatically merged:
I did that only once when I was planning to build an "uber-system that I could just upgrade forever". (
Yeah, I'm not sure I buy bigger case = better cooling either. The goal is to take ambient air outside, move it to the components where it picks up heat and out of the case. Those components do need room to contact the air coming in but at a certain point I'd expect it to be counter-productive. Where you have air supposed to be cooling spending time in places without components.
I believe some people even setup convoluted duct systems inside the case to give incoming airflow a direct line over components. Everything I've read on the matter indicates the optimal setup is just enough air hitting those components. In a way where the only way it can get inside is via the intake fans. With the exit points depending on fan pressure setup (everything near exhausts open as much as possible for positive, as an example).
It's kind of the same deal with more fans. It can be detrimental because it can make the intake air bypass the components. Granted, a lot of PC builders say things like hot air rises which is marginal to irrelevant when you start pushing it with fans (it behaves more like a fluid at that point). It may want to rise in it's default state but when you move it with fans it tends to go where the fans send it.
Noctua coolers may be better but they aren't competition free. There are plenty of good ones out there. The problem with coolers designed to push air down is the fin stack of the CPU cooler is designed to take it away from the CPU. It's counterproductive to blow this now hot air right back at the CPU. Particularly when some of it's likely to plow into anything near it. Like ram, the VRM, m.2 drives, etc.
Last edited:
Fractal Design - Nano-S works with D-15. (even though it states 160mm clearance it works and ive seen threads & pics)
As for airflow. Fit two noctua a12x25s on the front of case & one on the cooler. The high static pressure will work better than the 140mm fans the d15 comes with.
Here is an example, using other fans:
what the hell is up with posting imgur links?
Last edited:
Now, I think I get you. A "Frankenstein" approach like this will probably wind up with conflicts, eventually. Not to say it shouldn't be done -- especially if you have a good idea of what you're trying to achieve! But it will probably hit snags, and it will likely happen sooner rather than later.
Doubtful. If you had mangled the wiring or mounts, it would likely just not spin (nor maybe even detect) the 3rd-party fans at all.
Most likely, based on everything you've said. In that regard, I'd probably find a way to mount the fans where I wanted them, but connect and configure them completely separately from the Kraken unit. No crossover at all. What you could do there is trigger the fans to spin up based exclusively on the CPU temp as detected by the BIOS -- not by the Kraken software. If you "aim low" (meaning a few degrees below where the Kraken unit would normally start spinning up the fans), it should more or less line up and provide a similar or better result. It's going to be a bit of trial and error though.
So, in brief, try to "hide" the fans from the Kraken software, and just let the PC think that the Kraken stuff is "broken". There should be overrides as part of the software for the purpose of stress-testing and such. Since it doesn't sound like you're intending to push the limits or anything -- just get consistent performance and temp -- it should be pretty low-risk to fiddle. (But I don't know the software at all.)
Yes'm. The older system I'm referring to was (originally) a Pentium III model. I actually upgraded to a better P3 (...at cost...
) and dropped a DirectX GPU in it, though I don't remember which one now. Either way, I kept the Monster, so I had the option of both Glide and DirectX, and I managed to play through Deus Ex 1 at release without any major issues. But...gods...did the thing make noise and melt ice. Frig.I've found that having "breathing room" is more or less fine, but having gaping, open spaces inside the case actually adds heat. It's strong airflow in and out of the case that provides the best cooling and tends to keep things from really "caking" on dust. My present system is built to create a "swirl" of air in the case. The intake is slightly stronger than the exhaust fans, and I wind up getting about 30°-35°C idle CPU, with peaks at perhaps 60°C. My 980 ti will jump to ~75°C under load, but it mostly hovers around 65°-70°C in most demanding games (at 60 FPS). Normal operating temp for the GPU is about 50°C. All of that is done with one, high-powered intake fan at around 600-800 RPM, two exhaust fans + the PSU stock at no more than 1200 RPM, and the stock eVGA fans on the 980 ti...which will fire up to sound like a jet-engine when it gets going. (I just crank the volume up.
)End result is, I get nowhere close to the 83°-84°C throttle point on the GPU, nor the 90°C on the CPU -- ever. Performance is also smoking. To date, there is not a single game that I can't get smooth FPS with (barring software issues). The only title I own that makes me turn down even minor settings is Monster Hunter World, and the recent patches started giving me 60 FPS SOLID -- everywhere. So it sure looks like it was a massive optimization issue.
But end result is all air, no liquid. (So...less cleaning...
...!)
Yeah, this is basically what I did in the end. There is a Y splitter off the CPU fan header going to the two Noctua 140's mounted to the Kraken. The other fans are running off a splitter from the CHA fan header. The Kraken water pump is connected on the AIO header. CAM software was uninstalled. The only negative with this setup is the pump won't reach it's maximum rated RPM. In the overall scheme this probably makes almost zero difference. Controlling fan curves via bios over software is the preferred configuration anyway.
I'm likely getting a Meshify C for the case replacement (full size version). In all likelihood the one without the glass side panel. Most cases floating around are thermally challenged because the pursuit of aesthetics chokes off the airflow. This is one of the few smallish ATX cases I've been able to find without this problem. Although, I've seen conflicting reports on it's GPU thermals (this is where the earlier comments about inverted layout cases came from....).
There are a few others but most of the cases built for thermal performance tend to be large ATX, only support MATX or can't hold a 280mm rad (Kraken X62). It's a shame so many cases are stuck in the glass panel/RGB bubble.....
Again, this is because a future upgrade is going to be shifting to an ATX form factor. At this point I'm leaning toward waiting for Ryzen 4000 series to make this upgrade (case, board, CPU, CPU cooler). Primarily to let the newer concepts AMD is running with get a bit more maturity before switching to them.
Thanks for the picture though. That clears up a lot of what I was after
. I've found that having "breathing room" is more or less fine, but having gaping, open spaces inside the case actually adds heat. It's strong airflow in and out of the case that provides the best cooling and tends to keep things from really "caking" on dust. My present system is built to create a "swirl" of air in the case. The intake is slightly stronger than the exhaust fans, and I wind up getting about 30°-35°C idle CPU, with peaks at perhaps 60°C. My 980 ti will jump to ~75°C under load, but it mostly hovers around 65°-70°C in most demanding games (at 60 FPS). Normal operating temp for the GPU is about 50°C. All of that is done with one, high-powered intake fan at around 600-800 RPM, two exhaust fans + the PSU stock at no more than 1200 RPM, and the stock eVGA fans on the 980 ti...which will fire up to sound like a jet-engine when it gets going. (I just crank the volume up.
Those temps sound about what I get as well. A bit higher on the CPU because it's a non-delidded 8700k but GPU temps around the same.
The biggest thing I don't like about the 980 ti lightning is it's freaking huge. Hell, I think it's a bigger card then even the newest and greatest of the current GPU's.
Although it's a midi case, perhaps this one is too big for your taste as from a quick search and comparison it's larger -- but I've really liked what I have read about Phanteks P600. When I finally buy a new rig, I'll probably pick this case (I have an ancient one and want to finally replace it). Excellent cable management, good temperatures, and a front that can be adapted for air flow or cooling. But even with the front fully on, the air flow is actually pretty darn good according to tests I read some months back.
Just wish all these darn driver issues for AMD would cease. I'm on Linux as well, and it's the one thing that keeps me from buying. My current GTX 770 performs fine enough, and I've not had any real driver issues. And I'd absolutely hate to fork out 8-900 € for a new rig and then have the blasted thing malfunction.
@SigilFey: For dust prevention I always use Demciflex filters. It works very well.