So I want to build a computer with no moving parts...

[UHF]

I'm planning to build a custom open air case. It will be Mini ITX, so the graphics card will be exposed to decent natural air convection.

These are my inspirations;
http://www.fanlesstech.com/2015/01/nofa ... build.html
viewtopic.php?f=14&t=67691

Basically I want the best performance I can get for the video card. And No Fan.

If I'm lazy I can look at a few GTX 750 TIs.

However, I'm aware that I can go fanless for a GTX 960, but I'm wondering if anyone has considered working with a GTX 970? What about just removing the fans from something like the Asus Strix? What about down clocking?

The Arctic Accelero S3 can cool (mostly) a GTX 960, but the Arctic Extreme III looks like a much bigger radiator could it handle a 970?

I don't plan to stress test the card I settle on, but I would like to play games, like BF3, and Crysis 3.

Any ideas folks?

[CA_Steve]

Welcome to SPCR.

The practical limit for passively cooling a gfx card with made-to-fit coolers has been 50-75W. A 60W passive GTX 750 Ti may require directed airflow to keep from throttling with an OEM cooler. Check out the Palit KalmXreviews in the 750Ti thread. aerial went with a large aftermarket cooler and is seeing slightly better results. The GTX 960 has twice the power load of the 750 Ti at 120W. I think it would be a near impossible task to get twice the cooling efficiency out of a cooler made for a gfx card. You'd have to channel your inner MacGyver and modify a tower CPU cooler (a good google search for you)...and would probably still need to under clock/undervolt it...as well as still provide heatsinks for RAM and VRM circuitry.

145W GTX 970 stretches this further.

[quest_for_silence]

I'm planning to build a custom open air case. It will be Mini ITX, so the graphics card will be exposed to decent natural air convection.

What would be the difference from a readily available bench table/chassis?

What about just removing the fans from something like the Asus Strix?

The fins geometry is not up to the task (running fanless): compare the ASUS cooler with the Accelero S1 Plus/S3.

What about down clocking?

It will be not enough unless it will be really huge, and very likely it should require undervolting (which threats stability): so more probably that not it's not advisable.

The Arctic Accelero S3 can cool (mostly) a GTX 960, but the Arctic Extreme III looks like a much bigger radiator could it handle a 970?

The Extreme geometry is not suited to run fanless and above all you need to cool the power circuitry on high end cards, and you can't do it with fanless heatsinks. I'm neither sure the S3 can handle a full bore GTX 960.

Any ideas folks?

It could be doable, but IMVHO it's expensive, it's pointless (unless you want just to show off) and it likely shorten the build's lifespan (at least of the GPU).

[edh]

The Arctic Accellero S1 Plus or S3 can cool 125W. Just to put it in perspective the original S1 was factory fitted to some 8800GTs which had a TDP of 126W. If you can get it to fit, the Accelero S1 Plus or S3 would therefore be fine for a GTX 960 if you can physically get everything to fit your chosen card correctly - not all cards are the same layout.

MiniITX might make things worse. It does mean that there is no space between the CPU and graphics card which will be harder for passive cooling. MicroATX with the graphics card in the second expansion slot would allow more space for the CPU cooler.

[xan_user]

why exactly does it need to have no moving parts? coolness factor , or is there some other reason?

for me, I much prefer the sounds of one quite fan to the chirping and squealing that often pops up from coil-whine in fan less systems.

[CA_Steve]

The Arctic Accellero S1 Plus or S3 can cool 125W.

Passively with no other airflow?

[UHF]

I'm planning to build a custom open air case. It will be Mini ITX, so the graphics card will be exposed to decent natural air convection.

What would be the difference from a readily available bench table/chassis?

It should be effectively the same for air circulation as the examples I showed. (More like the Dimastech example in that the GPU would be 3 inches off the table or so.)

What about down clocking?

It will be not enough unless it will be really huge, and very likely it should require undervolting (which threats stability): so more probably that not it's not advisable.

I just played with this using one of my GTX 980s in my desktop. Not passive (fans at 25%), but I down clocked it by 500Mhz for core and RAM. Afterburner said it only used 60% of the power the whole time, but its Unigen Valley benchmark showed about 40 frames a second. Which is not much better than a 960. So I guess I'm now leaning to a GTX 960 with 4GB of RAM.
http://www.hitechlegion.com/reviews/gra ... 960-review

The Arctic Accelero S3 can cool (mostly) a GTX 960, but the Arctic Extreme III looks like a much bigger radiator could it handle a 970?

The Extreme geometry is not suited to run fanless and above all you need to cool the power circuitry on high end cards, and you can't do it with fanless heatsinks. I'm neither sure the S3 can handle a full bore GTX 960.

Arctic Accelero S3 playing video games open air, with no fan.
https://www.youtube.com/watch?v=tCvhmISioJw

By the way, can you explain what you mean by (Arctic?) Extreme geometry is not suited to run fanless? To my eyes, the vertical fins will convect naturally much better than the Accelero S3, and there are more fins. That is, I think it will have better air flow.

Any ideas folks?

It could be doable, but IMVHO it's expensive, it's pointless (unless you want just to show off) and it likely shorten the build's lifespan (at least of the GPU).

I disagree with a lot of what you think here. CPUs are beefier these days for lower power. Why do I need to put fans on, so I can get almost nothing more. I have an i7-4790k, and an older i7-2600k, and its not much faster. Further more, over the years I've run my 2600k balls to the walls for weeks on end doing Reality/Lux rendering. Its not happy machine any more, but it is water cooled.

why exactly does it need to have no moving parts? coolness factor , or is there some other reason?

for me, I much prefer the sounds of one quite fan to the chirping and squealing that often pops up from coil-whine in fan less systems.

Because I want to!

The Arctic Accellero S1 Plus or S3 can cool 125W.

Passively with no other airflow?

Mostly... Stress testing it will cook it. But it can game, like so;
https://www.youtube.com/watch?v=tCvhmISioJw

If I was to do something like this in a build, I'd be configuring throttling based on temperature, and carefully adjust my games/PC to reduce heat.

[UHF]

The Arctic Accellero S1 Plus or S3 can cool 125W. Just to put it in perspective the original S1 was factory fitted to some 8800GTs which had a TDP of 126W. If you can get it to fit, the Accelero S1 Plus or S3 would therefore be fine for a GTX 960 if you can physically get everything to fit your chosen card correctly - not all cards are the same layout.

MiniITX might make things worse. It does mean that there is no space between the CPU and graphics card which will be harder for passive cooling. MicroATX with the graphics card in the second expansion slot would allow more space for the CPU cooler.

I'm curious... but you're using a "Scythe Zipang 2" fanless? Would that work open air?

[quest_for_silence]

The Arctic Accellero S1 Plus or S3 can cool 125W.

Passively with no other airflow?

I don't think so, but actually I didn't check twice.

[CA_Steve]

Because I want to!

Don't let us get in the way of your inner MacGyver/Ben Heck. Under clocking the video card is useful - undervolting is, too. Power is proportional to freq x voltage^2. So, a 10% drop in core voltage can lead to a ~20% drop in core power use.

[xan_user]

Because I want to!

totally understand that..


(...but it is my journey into fanless builds, that taught me one fan was better than none.)


Edit. PS, If your going custom, i would suggest a mobo and GPU orientation that is vertical for both. ie with I/O ports facing either up or down. in fanless rigs one of the main keys is also keeping the mobo temps within specs...vertical orientation really helps in that regard.

[quest_for_silence]

Arctic Accelero S3 playing video games open air, with no fan.
https://www.youtube.com/watch?v=tCvhmISioJw

I'm sorry but I'm failing to see what you mean: can you kindly point out the relevant timing?

By the way, can you explain what you mean by (Arctic?) Extreme geometry is not suited to run fanless? To my eyes, the vertical fins will convect naturally much better than the Accelero S3, and there are more fins. That is, I think it will have better air flow.

First of all, the airflow is not given by the numbers of fins (but those fins "use" the airflow), while fin spacing matters more.
Set aside that, what you mean might (might!) be a reasonable guess (a somewhat cooling advantage) if there were a chimney or stack effect but, in an open air scenario, that's not the case, you have no flow above or below the heatsink. Besides, don't you think AC would have adopted the same scheme on the passive S1/S3, if they thought it were more effective?

It could be doable, but IMVHO it's expensive, it's pointless (unless you want just to show off) and it likely shorten the build's lifespan (at least of the GPU).

I disagree with a lot of what you think here. CPUs are beefier these days for lower power.

So what? I don't understand what you mean.

[UHF]

Arctic Accelero S3 playing video games open air, with no fan.
https://www.youtube.com/watch?v=tCvhmISioJw

I'm sorry but I'm failing to see what you mean: can you kindly point out the relevant timing?

?? What timing are you looking for? Near as I can tell he produces no details about how well he ran games.

By the way, can you explain what you mean by (Arctic?) Extreme geometry is not suited to run fanless? To my eyes, the vertical fins will convect naturally much better than the Accelero S3, and there are more fins. That is, I think it will have better air flow.

First of all, the airflow is not given by the numbers of fins (but those fins "use" the airflow), while fin spacing matters more.
Set aside that, what you mean might (might!) be a reasonable guess (a somewhat cooling advantage) if there were a chimney or stack effect but, in an open air scenario, that's not the case, you have no flow above or below the heatsink. Besides, don't you think AC would have adopted the same scheme on the passive S1/S3, if they thought it were more effective?

No they wouldn't. The idea of flow depends on the orientation of the motherboard. Very very few cases put the motherboard flat, as in on a surface. In fact, I'd say that most are towers.

If there is heat, there will be air flow. You want to do everything you can to improve that flow. Sideways fins will slow up flow, vertical fins would flow faster.

There will always always be a chimney stack effect. You can see it by holding your hand over a light bulb.

It could be doable, but IMVHO it's expensive, it's pointless (unless you want just to show off) and it likely shorten the build's lifespan (at least of the GPU).

I disagree with a lot of what you think here. CPUs are beefier these days for lower power.

So what? I don't understand what you mean.

There is no evidence that this will shorten the lifespan of a PC. None. There are a lot of options to produce a long lasting PC. If you look around, there is a lot of talk about military grade parts. (Those are high temp, long lasting.) All things being equal, a cheap motherboard would likely have a shorter lifespan, if and only if it got too hot.

Most complex silicon (GPU. CPU, Chipsets) are built using technologies (like SOI) allowing much much higher temps. (FYI, I design high temp electronics for a living. I just don't usually work with air flow or heat sinks.)

[UHF]

Because I want to!

totally understand that..


(...but it is my journey into fanless builds, that taught me one fan was better than none.)


Edit. PS, If your going custom, i would suggest a mobo and GPU orientation that is vertical for both. ie with I/O ports facing either up or down. in fanless rigs one of the main keys is also keeping the mobo temps within specs...vertical orientation really helps in that regard.

I have been thinking about that... a lot. You're talking about a case with board orientations like this;
http://www.nofancomputer.com/eng/products/CS-70.php

That design would have a superior airflow and cooling capability. However, cables also need a fair bit of clearance which you probably need to consider in the design.

Its not really the aesthetic I want. I'm also wanting to make this as small as possible, like 22cmx22cmx22cm. (3rd part heat sinks tend to throw that idea out, which is why I'm trying to figure out what I can now.)

[xan_user]

my current rig. ~110 watts peak

sure It works fanless... But I am just not comfortable running it without a fan long term, especially during norcal summers. The fan is not really needed on the CPU, but it does keep the back of the GPU cool and move air around the mobo, VRMs, SSD, pico-PSU ect...

Plus I needed something to cover up the small amount of squealing/whine from the various components. ( the fan also tells me if my computer is on or not. lol.)

keep in mind too, that often those little electronic noises get worse as temps increase.

AFAIK there are very few members on SPCR that run a truly passive rig, unless of course they have one of the ma$$ive integrated heatsink cases, use integrated GPU or live near the arctic circle.

[quest_for_silence]

Near as I can tell he produces no details about how well he ran games.

So how do you know he can run games safely and without throttling?

There will always always be a chimney stack effect. You can see it by holding your hand over a light bulb.

I'm sorry, but despite what you claim about yourself, maybe you have to be a tiny bit less presumptuous about those things (on a light bulb my hand feels some natural convection, the chimney effect is a tad different phenomenon which requires a difference in pressure).
In my (indeed, little) experience (I ran fanless systems since about a decade, and every time I had to revert to fanned ones), and even nowadays no one can run as you claim: what you think will mandatorily happen, it is just uneffective (quite weak) under load (set aside the heat source is not on the motherboard, but on the graphics, and it is transferred through the heatpipes).
Don't believe me, just watch what happen "if".

There is no evidence that this will shorten the lifespan of a PC. None.

Heath is the bane of any electronics but you should know that.
Just for instance, did you give a look to the SPCR Accelero Extreme review?
I think you know any electronic part has derating curves/function/figures related to the operating temperature, right? Do you think that graphic card will have the same lifespan with its power circuitry running at 147°C instead of 96°C (so 51°C more: and it isn't fanless, just with fans at 5V)? I think you also noted the power draw increase of about 5% just because of (less) cooling, so increasing the heat build up.

Summarizing, you're free to act as you think best, but I'm quite convinced you can't re-invent physics (and I bet you won't find any "military grade" Nvidia GTX 970 graphics who can stand the increased heating of running fanless for the same lifespan of a Strix or Gaming card, at least for any somehow decent amount of money).

AFAIK there are very few members on SPCR that run a truly passive rig, unless of course they have one of the ma$$ive integrated heatsink cases, use integrated GPU or live near the arctic circle.

To the OP: that's the most recent Gallery thread I'm aware of.

[CA_Steve]

As mentioned earlier, I applaud your wish to build a fanless gaming box. Go for it. That said, I've got to jump in on the temperature/lifetime thing.

In general, higher temperatures derate component lifetimes. For semiconductors, the driver is electromigration. Current moves ions causing connections to fail. Raise the current density, go to a smaller geometry process, have a higher operating temperature, etc.. and it gets worse. The workaround as processes shrink is to come up with new interconnect metallization chemistries and to lower operating voltages to in turn lower current densities. SOI improves transistor leakage current - not lifetimes.

I figure Nvidia and AMD design their gpus to an acceptable failure rate within their warranty lifetimes and within given operating parameters. They are consumer grade products meant to last a couple of years. Thicker interconnect for longer lifetimes blows out the die size lowering $ margins. So, operate above the rated clock / voltage/temps and lower the MTTF. Operate below the rated conditions and increase the MTTF.

In specific, it's the luck of the draw. Process lots/chips/board component quality varies. You might get a board that lasts for many years at high temps (I've got an old passive GT 6600 that ran at 85C for years without issue). Or, you might get a board with a crappy VRM capacitor that fails in a year or two. <shrugs>

(I spent a lot of years in the semiconductor industry).

[UHF]

I'm not sure why you guys are jumping on me. Read my first post again. I'm saying that I want to see how well I can do. I'm not claiming that I'll achieve perfection. I have never claimed all electronics are military spec, and that they last forever.

I typically operate commercial grade electronics at 200C. (The biggest issue I face is that oscillators tend to give up the ghost soon, and many ICs have thermal shut offs. After that its a short life span, but we're more likely to experience mechanical failure first.)

Going fanless is a lot about making sure you don't over heat anything, and thus derate your parts. But there will always be heat.

I also don't plan to continuously stress test the computer. I know what that does, and I don't feel like making a sound proof case. My last desktop was running Lux for months on end, at 120% CPU loading and 70C continuously. I just replaced all the fans and a few other parts. My new desktop makes a racket when GPU rendering, but is otherwise dead quiet.

I don't plan to stress out this little machine. I'm not that concerned about cooking my GPU to 80c for 10-20 hours a year or so. I'll take my chances, and I'll work hard to keep it to that level of heat when stressing it out. For regular operations or even home theater operations this baby should be cold and quiet. Heck, I'm contemplating a 35W i3 on an 80W heat sink.


Hence I 'asked' if anyone had done anything more than simply running a GTX 750TI, or a GTX 960 fanless.

I wasn't expecting the Spanish Inquisition.

[xan_user]

I'm not sure why you guys are jumping on me. Read my first post again.

....Any ideas folks?

we are only giving you our opinions based on our experiences.

we get asked about fanless rigs a lot... but we've all (mostly) come to the same conclusion, that its not worth the headache, time, and money to do it.

with that said, by all means go for it.

[UHF]

Well, I'm most certainly hearing 'don't bother with performance'... and a GTX 960 with passive cooling is probably not as good as it sounds. (If I did that, it would like be getting throttled.)

Interestingly with the Mini ITX you can get rid of the heat pipes for a big block heat sink with a monster heat spreader;
http://www.quick-cool-shop.de/en/heatdi ... gular.html

That's because the video card is right at the edge.

[quest_for_silence]

I wasn't expecting the Spanish Inquisition.

Well, I didn't think I was as rude as Torquemada, in case I just duly commented some apparently bold statements.
Particularly:

There will always always be a chimney stack effect.

and

There is no evidence that this will shorten the lifespan of a PC. None.

In case those things do require some demonstrations by you: if it isn't what you perceived, I beg your pardon.

Set aside that, as a matter of fact I also said "It could be doable", CA_Steve wrote "I applaud your wish to build a fanless gaming box. Go for it.", and xan_user added "totally understand that": I think no one asked you for any abjuration, right?

Any of us (I mean, silencing enthusiasts) already wanted to do what you want to do now, at least once, and our experiences lead us to share some common beliefs about that (that in the past it didn't worked as expected/desired, but that some day it might even work, maybe cutting some corners, so you may try): extremely summarizing that's what collectively we said, I think.

Hence I 'asked' if anyone had done anything more than simply running a GTX 750TI, or a GTX 960 fanless.

Purposedly I linked to you that above quoted Gallery thread: current state of the art for running fanless high performance graphics is to use flat pipes (or perhaps even vapour chambers) and massive heatsinks, nothing small nor inexpensive.

I have been thinking about that... a lot. You're talking about a case with board orientations like this;
http://www.nofancomputer.com/eng/products/CS-70.php

That design would have a superior airflow and cooling capability. However, cables also need a fair bit of clearance which you probably need to consider in the design.

I forgot to comment on that: in the mATX flavour, it didn't prove to cool that well, even when fanned, at least with high performance graphics. So, despite the somehow promising design, such a solution mileage may vary (personally I wouldn't bet on it, but that's just me).