TR Forums

Hello,

Many of the new cards, both AMD and nVidia, are blower designs and I have been wondering what the effect of case pressure would be on their performance.

I was wondering, would it be possible to measure thermals and noise in 2-3 different situations:
1. No case fan, only PSU fan ("neutral pressure")
2. Negative pressure (more outwards than inwards pressure, ex 2 fans blowing out and only PSU sucking air)
3. Positive pressure (more inwards than outwards pressure, ex 2 fans sucking air).

The test would could be limited to 2-3 cards (1070, 1080 and RX480 for example) and a single case (something reasonable, say corsair 450D or whatever you have in hand). Don't need 20 benchmarks. Two games (+ maybe furmark) should be enough to see how the blower copes with different situations.

What do you think?

I don't think most cases are airtight enough for it to make much difference one way or the other to performance. On the other hand balancing airflow in and out is a good idea as it stops dust getting pulled into every nook and cranny. If your case has a dust filter on the intakes then this is even more useful.

Yep, air pressure differences in a case make very little difference to fan output or air density. We're talking about fractions of a percent.

The only thing positive vs negative pressure determines in a desktop case is where the dust ends up.

Positive pressure gently pushes air out of all the many small vents and gaps in a case, keeping those gaps from getting clogged with dust. The dust gets gathered with the intake air, which should in theory be filtered to stop the worst of it.

Negative pressure means that dust will get in through every tiny gap. These gaps do not have dust filters, and by their nature are likely to trap the dust in the gap. Expect to vacuum your case every few months if you don't want it to look like the contents of a vacuum cleaner.

Wow. Feels like 1999 all over again

TheEmrys wrote:

Wow. Feels like 1999 all over again

The scary thing is how many people building computers now weren't even born in 1999.

That and there's less need to worry about dust building up inside your floppy drive.

cheesyking wrote:

I don't think most cases are airtight enough for it to make much difference one way or the other to performance. On the other hand balancing airflow in and out is a good idea as it stops dust getting pulled into every nook and cranny. If your case has a dust filter on the intakes then this is even more useful.

You are probably right, but there is a difference between "probably" right and actually testing it. My hypothesis is that to overcome a slight pressure delta, the blower will have to work much harder and produce more noise. Maybe this pressure delta is negligible in real conditions, but I don't know if anyone has really measured. By the way, not to be pedantic, but airflow is by definition balanced since the case is not expanding or contracting.

The only reason most of the available new GPUs are blower only is because that's what AMD and Nvidia do for their reference designs. And since production has been slow to ramp up, reference cards are still the majority of the market. If you look, most of the custom cards (aka not a reference blower with the board partner's sticker on the fan) are axial fans because they tend to be quieter and cool better. Albeit, the downside is they don't blow the hot air out of the case like a blower design.

ptsant wrote:

By the way, not to be pedantic, but airflow is by definition balanced since the case is not expanding or contracting.

Yeah, that's pretty darned pedantic!

ptsant wrote:

By the way, not to be pedantic, but airflow is by definition balanced since the case is not expanding or contracting.

Just as well you're not pedantic since you started the thread off talking about "negative pressure"

ptsant wrote:

cheesyking wrote:

I don't think most cases are airtight enough for it to make much difference one way or the other to performance. On the other hand balancing airflow in and out is a good idea as it stops dust getting pulled into every nook and cranny. If your case has a dust filter on the intakes then this is even more useful.

You are probably right, but there is a difference between "probably" right and actually testing it. My hypothesis is that to overcome a slight pressure delta, the blower will have to work much harder and produce more noise. Maybe this pressure delta is negligible in real conditions, but I don't know if anyone has really measured. By the way, not to be pedantic, but airflow is by definition balanced since the case is not expanding or contracting.

I'll put it to you this way: I champion positive case pressure simply for keeping the dust on the intake filters, and using larger fans to keep smaller blower fans from spinning up higher than necessary. This also helps when you're using an IWC with the radiator on the exhaust.

However, finding GPUs with good blowers is a rare feat: usually only the best Nvidia cards have them, and they *need* them. You're not likely to find those coolers on the next tier down parts, and if that's where you live (like I do) then it's all a wash. My current EVGA GTX970's have their ACX2.0+ open-air coolers in an intake-biased Define 3 system, and neither temperatures nor noise are concerns, and my motherboard has one slot of space between the two dual-slot cards.

Note that that's two higher-end cards in SLi, with open-air coolers, no issues.

I'd only be truly insistent on using blowers if the motherboard had the cards right next to each other.

Airmantharp wrote:

I'll put it to you this way: I champion positive case pressure simply for keeping the dust on the intake filters, and using larger fans to keep smaller blower fans from spinning up higher than necessary. This also helps when you're using an IWC with the radiator on the exhaust.

However, finding GPUs with good blowers is a rare feat: usually only the best Nvidia cards have them, and they *need* them. You're not likely to find those coolers on the next tier down parts, and if that's where you live (like I do) then it's all a wash. My current EVGA GTX970's have their ACX2.0+ open-air coolers in an intake-biased Define 3 system, and neither temperatures nor noise are concerns, and my motherboard has one slot of space between the two dual-slot cards.

Note that that's two higher-end cards in SLi, with open-air coolers, no issues.

I'd only be truly insistent on using blowers if the motherboard had the cards right next to each other.

I agree completely. Never owned a blower-style GPU.

However, if someone does end up with a blower-type GPU, should he take extra care in ensuring positive pressure or not? Is it important for the performance of the GPU? Both the 1080 and the RX480 are running close to their thermal limits, so it would be interesting to examine if there is something that can modify the performance of their cooling systems. For example: slow down exhaust fans or accelerate intake fans? Does it matter?

I would run this test myself out of curiosity, but as I said I don't have a blower-style card.

I'd want positive pressure either way.

As mentioned repeatedly above, you want dust to be caught in intake filters, not degrading performance by clogging up everything inside.

Airmantharp wrote:

I'd want positive pressure either way.

As mentioned repeatedly above, you want dust to be caught in intake filters, not degrading performance by clogging up everything inside.

Yes, I get that. What I'm saying is that the mere fact of adding a blower can change the equilibrium in your case from positive to negative pressure. If you are exchanging an axial fan GPU with a blower GPU, you are adding an exhaust fan. Should you add another intake fan to compensate, for example?

ptsant wrote:

You are probably right, but there is a difference between "probably" right and actually testing it. My hypothesis is that to overcome a slight pressure delta, the blower will have to work much harder and produce more noise. Maybe this pressure delta is negligible in real conditions, but I don't know if anyone has really measured. By the way, not to be pedantic, but airflow is by definition balanced since the case is not expanding or contracting.

Whilst air pressure has a significant effect on which way air flows and how fast it flows, the actual cooling performance is linearly related to air density; If pressure is at two atmospheres, a heatsink has twice as many air molecules to interact with and will be twice as effective, given the same airflow. You can eliminate all kinds of turbulence and fluid dynamics from the equation because it affects both sides of the equation equally. Trust me, as an engineer with more background in fluid dynamics than I ever wanted to know, it's one of those blissfully simple bits to learn.

So, does an increased case pressure with denser air lead to increased cooling performance or heatsink efficiency?
Defintely, yes.

How much extra performance do you get?
So little, it might as well be zero.

What people forget is that atmospheric pressure is IMMENSE. A typical 1200rpm, 120mm fan has the ability to produce a difference in static pressure of about 1.5mm/H₂O. Let's say you have a really imbalanced case with three intakes but only one exhaust - that's a pressure differential between inside and outside of 3mm/H₂O. Do you want to know what one atmospheric pressure is, using the same units?

10300mm/H₂O. Yes. That's the pressure caused by the weight of 1280km of increasingly thinning atmosphere on top of you. Air is not weighless, it just seems like it is because our bodies have equally immense pressure inside every cell. Pressure so great that a human would explode in the vacuum of space (probably, or perhaps this is just Hollywood. Ask XKCD, it's probably something which Randall has covered at some point!).

So, 3mm from a surplus of two 120mm fans, divided by 10300mm of atmospheric pressure means that the air in a positive pressure case of 3-intake, 1-exhaust is 1.00029x more dense than the air outside your case, which means that your cooling solutions will work 0.029% more effectively. Likewise, it'll be 0.029% less effective if you have 'strong' negative pressure with your fans.

People make the mistake of thinking "two intakes for one exhaust must mean double the air pressure!"
Nope.jpg.
The actual difference is 0.000 to three decimal places. If you have to look at the fourth decimal place to see anything, it's probably best described as "negligible" when everything else of relevance to cooling performance is more directly, linearly proportional.

ptsant wrote:

Airmantharp wrote:

I'd want positive pressure either way.

As mentioned repeatedly above, you want dust to be caught in intake filters, not degrading performance by clogging up everything inside.

Yes, I get that. What I'm saying is that the mere fact of adding a blower can change the equilibrium in your case from positive to negative pressure. If you are exchanging an axial fan GPU with a blower GPU, you are adding an exhaust fan. Should you add another intake fan to compensate, for example?

You should check and make sure that the pressure situation isn't reversed, and if it is, then take action.

Positive vs negative pressure isn't strictly a count of fans.

It can also be achieved through fan management. Fewer intake fans that always push a much greater volume of air versus the exhaust fans will also create positive pressure.

https://www.youtube.com/watch?v=a12aDCxrcts

Filtered fan inlets on "positive pressure" setups create noise.

My quietest computers use a PSU with a large diameter fan as an exhaust, a large diameter tower style CPU cooler and no filters. That's a "negative pressure" setup.

I try to set it up so that heat rises out of the case to encourage convection.

That setup works well enough if the temperatures stay reasonable in an extended stress test.

Adding a hot GPU might screw up the temperatures so a blower cooler that directs most of the GPU heat out of the case can be the right move.

As mentioned, the pressures involved are very low so a negative pressure setup is not going to appreciably change the air flow out of the blower cooler.

If you have more fans and a reasonable layout, your GPU can run quieter and cooler with an axial fan cooler but at the expense of case fan noise.

If you try to pick high efficiency parts, cooling can be pretty simple. If you are dumping lots of heat in to your case then you need to start adding on to take care of it.

I'll be damned if this isn't an interesting thread. I've maintained Positive Pressure (or at least I think I do) for about two decades now and since case makers are getting better at shoring up leaks I find this, relatively, good. However, Chrispy_ and Ryu have got me thinking. Maybe there really isnt a right or wrong way. "Is my stuff overheating during intense gaming?" Nope. Then I'm good. "Am I getting anomalies, artifacts, weird BSODs, or even random reboots and shutdowns?" Nope. Then I'm good.

If I had Blower GPUs in my SLI setup I would set up as positive pressure but with my current 2 EVGA classified 4gb 770's that throw out tons of heat into my case it is setup with more exhaust than intake fans. My 2 side 120mms are setup as exhaust and the amount of heat expelled from the side of the case is substantial. I had them as intakes figuring they would help cool the cards but the GPU's fin setup made dead spots with no airflow and temps increased by 20c vs having the sides as exhaust.

Noise is not a problem since my case is elevated on a stand to the left of me so the side fans are on the opposite side of the case to me.

I've generally run negative pressure. I rely on air being pulled in through the front panel by the exhaust fans to cool the drive bays. Yes, I do get a bit more dust buildup. But it is quieter, and the drives seem to get cooled more evenly.

An interesting side effect of most case fans is that their airflow is related to the relative static pressure.

To simplify the example, all fans are the same size and RPM - but if you had three intake and two exhaust, the exhausts would be at their max airflow because they wouldn't be working against static pressure, meanwhile the three intakes would be working against the backpressure in the case and their airflow performance would decrease to match the static pressure in the case. The net result is that the intakes and the exhaust would match airflow (almost, a bit is going to escape through gaps in between panels etc) but because there is more work being done by the intakes, the static pressure would increase.

To calculate your airflow, balance exhaust CFM against intake CFM according to manufacturer ratings at zero static pressure. Whichever has the lowest airflow is your airflow for the case, and whichever is higher is what will create the pressure differential.

If your intake airflow is lower than exhaust, the intakes will be running at rated airflow and the exhausts will be matching that airflow to work against atmospheric pressure, creating slightly negative pressure in the case.
If your exhaust airflow is lower than intake, the exhausts will be running at rated airflow and the intakes will be matching that airflow to work against internal pressure, creating slightly positive pressure in the case.

Things like dust filters hinder a fan's CFM so if you have intakes with dust filters and you want positive case pressure, you may need to crank the RPMs higher or just have more fans to compensate.
Sadly, with variable fan controls the CFM graph is quite a complex 3D curve - you have to plot airflow vs static pressure vs RPM which is bad enough even when all the fans are identical and running on the same profile. This complexity basically mandates the need for trial and error

I've never really concerned myself much with the pressure differential -- as Chrispy mentioned, the actual pressure difference either way is going to be extremely small.

What I have attempted to do as much as possible is manage how the air flows through the case, with the idea that I want to be efficiently removing hot air and replacing it with cool air without any risk of "rebreathing". To do this, I make sure that all of my intakes and exhausts are on opposite sides of the case. In most builds this means I have the intakes on the front and the exhausts at the back -- sometimes depending on the case layout this turns into the top, but again towards the back of the case. The intent is to have the air flow in, do its job cooling components, then get the heck out carrying the heat with it. This helps avoid having pockets of hot air getting caught somewhere and continuously flowing over the heat sinks.

The static pressure drop through a case from the inlet to outlet is pretty small, if you have inlet fans, I would assume that they are moving close to their full CFM rating unless there are filters or grills or a mess of cables to blow through.

A PSU fan has to suck air out of a case and then blow it through a pretty tight box filled with wires and heatsinks. Even a very silent PSU will manage to blow quite a bit of air through it despite an almost worst case static pressure drop.

I wouldn't really worry about static pressure unless I was pushing air through a radiator or heat sink.

On that topic, I saw that Corsair has some new magnetic levitation fans with quite nice specs, good flow, wide static pressure range 4 wire PWM control from 400-2,400 RPM and no motor bearing noise. I'm thinking about buying a pair to swap on to CPU coolers.

Lastly, I always take the time to get my cables tucked away for better air flow. It just makes sense.

My current case followed my traditional approach: stick inlet fans everywhere on the front there was a filtered spot for them, outlet fans everywhere on the top and back there was a spot for them, and turn the RPM on all of them down until I can't hear them. ...And then I leave the side panel off the case half the time anyway because I'm mucking around in there or whatever.

Chrispy_ wrote:

What people forget is that atmospheric pressure is IMMENSE. A typical 1200rpm, 120mm fan has the ability to produce a difference in static pressure of about 1.5mm/H₂O. Let's say you have a really imbalanced case with three intakes but only one exhaust - that's a pressure differential between inside and outside of 3mm/H₂O. Do you want to know what one atmospheric pressure is, using the same units?

10300mm/H₂O. Yes. That's the pressure caused by the weight of 1280km of increasingly thinning atmosphere on top of you. Air is not weighless, it just seems like it is because our bodies have equally immense pressure inside every cell. Pressure so great that a human would explode in the vacuum of space (probably, or perhaps this is just Hollywood. Ask XKCD, it's probably something which Randall has covered at some point!).

So, 3mm from a surplus of two 120mm fans, divided by 10300mm of atmospheric pressure means that the air in a positive pressure case of 3-intake, 1-exhaust is 1.00029x more dense than the air outside your case, which means that your cooling solutions will work 0.029% more effectively. Likewise, it'll be 0.029% less effective if you have 'strong' negative pressure with your fans.

Thanks for the details. Let me put it another way. Given, as you say, that air density is more or less constant and out of our control, this is not the variable that we are looking at. The value we care about is airflow, and airflow depends on pressure difference. Your card needs a pressure difference of, say, 1mm/H20 between the blower and the outside of the caseto achieve a sufficient airflow while gaming. However, and this is the important point, the other fans are also creating a pressure difference. So, what I have been saying is that if you have, say 3 12cm fans pumping air out of the case your blower will still have to achieve the same 1mmH20 to cool the card. Since, as you say, the external pressure is fixed, this means that it will have to counter the negative pressure first then add 1mmH20 more. So, for the same pressure difference the blower has to produce more work, ie rotate more quickly, because it is countering the effect of the other 3 fans. Conversely, imagine a setting in which you have 4 very powerful 14cm intake fans. At that point, the pressure across the GPU is probably already quite positive and you may even get sufficient cooling with the blower turned off (this is what happens in servers: powerful fans ensure flow for all components and you don't need GPU fans, as in this Tesla card: http://www.nvidia.com/object/tesla-servers.html).

Anyway, I might be very wrong here but that's why I'm asking if someone with a blower-type GPU and a decent case can try experimenting with the case fans to see what the effect is on the blower.

Generally you want to balance it as much as you can, erring towards slightly positive for dust reasons.

With the pressure being so close to constant, the airflow of the whole case is limited by the lowest of either "total exhaust airflow" or "total intake airflow". The limiting side will run at optimum airflow and the limited side will have reduced airflow to match.

If you have vastly more exhausts than intakes, the airflow through the exhuasting blower will drop, relative to its fan speed and fan noise, so from the most obvious perspective of keeping a graphics card cool and quiet, you definitely want to avoid a negative pressure fan setup. In an extreme example, you'd have one intake, 3 exhausts AND a blower on your GPU. The airflow of the one intake would be split between the 3 exhausts and the blower, so whilst it would make barely any difference to the pressure, it would still be working against that pressure differential. Two things to remember:

Small pressure differences make significant airflow differences.
Any pressure differential makes a fan less effective, so balance is better than imbalance.

You will be unable to balance it perfectly, so if you want dust control via intake filters, go for a slight intake bias for positive pressure.

One thing to remember is not to forget the fan on your PSU and its orientation.

ptsant wrote:

Anyway, I might be very wrong here but that's why I'm asking if someone with a blower-type GPU and a decent case can try experimenting with the case fans to see what the effect is on the blower.

I've got some old data that may not answer all your questions, but will hopefully given you something to work with.

TLDR version:
Negative vs Positive Airflow makes little difference by itself.
Directed airflow makes a much larger difference.
Dust makes a larger difference still.

Setup:

Thermaltake Xaser 3 (Silver) Chassis
Enermax Liberty 500W PSU
Thermaltake MaxOrb (HSF)
Athlon 64 3200+
2GB RAM
X1900XT

Load provided by Prime95 + ATItool
Temperatures provided by CoreTemp and ATItool.

The Xaser 3 has 7 80mm fans which I've installed in various orientations and the PSU has a fan (not removed) which exhausts air from the inside of the case out.
1 fan exhaust@Top

CPU 50C GPU 93C

2 fans exhaust@Rear

CPU 47C GPU 92C

2 fans intake@Front

CPU 49C GPU 90C

2 fans intake@Side

CPU 48C GPU 88C

3 fans exhaust@Rear/Top

CPU 48C GPU 92C

4 fans intake@Front/Side

CPU 48C GPU 87C

2 fans intake@Front & 2 fans exhaust@Rear

CPU 48C GPU 90C

2 fans intake@Side & 2 fans exhaust@Rear

CPU 47C GPU 89C

2 fans intake@Front & 3 fans exhaust@Rear/Top

CPU 48C GPU 91C

2 fans intake@Side & 3 fans exhaust@Rear/Top

CPU 48C GPU 90C

4 fans intake@Front/Side & 2 fans exhaust@Rear

CPU 47C GPU 87C

4 fans intake@Front/Side & 3 fans exhaust@Rear/Top

CPU 46C GPU 85C

The short summary is that, for the Xaser 3, positive and negative pressure made less difference than where the air is being directed. CPU temperatures benefited most from the rear exhaust, while GPU temperatures benefited most from direct airflow from the side fans. The top fan is generally a detriment to the system. That is until all the other fans are populated. After which it gives a rather unexpected boost to cooling on both major components. Best guess is that the top fan reduced turbulent airflows (air spinning around in pockets and not getting to the right spot) that presented themselves under the six fan setup. I don't have noise measurements, but subjectively, the GPU was the biggest offender and it definitely sounded worse under the setups without the side intake fans present.

Over the course of the next year, with no cleaning and the full complement of fans, this chassis did a very good job of keeping dust out and my temperatures went up by <1C on the CPU and 1C on the GPU. Interestingly, another guy I went to school with had the blue version of the chassis. He had a completely different set of hardware than I had and didn't really want to take the time to test, but he was a big proponent of negative airflow setups. With no intakes and 3 exhaust, he managed to pick up quite a bit of dust over the next year. He didn't have GPU numbers, but his CPU temperatures went up 9C from when we started.

ptsant wrote:

Thanks for the details. Let me put it another way. Given, as you say, that air density is more or less constant and out of our control, this is not the variable that we are looking at. The value we care about is airflow, and airflow depends on pressure difference. Your card needs a pressure difference of, say, 1mm/H20 between the blower and the outside of the caseto achieve a sufficient airflow while gaming. However, and this is the important point, the other fans are also creating a pressure difference. So, what I have been saying is that if you have, say 3 12cm fans pumping air out of the case your blower will still have to achieve the same 1mmH20 to cool the card. Since, as you say, the external pressure is fixed, this means that it will have to counter the negative pressure first then add 1mmH20 more. So, for the same pressure difference the blower has to produce more work, ie rotate more quickly, because it is countering the effect of the other 3 fans. Conversely, imagine a setting in which you have 4 very powerful 14cm intake fans. At that point, the pressure across the GPU is probably already quite positive and you may even get sufficient cooling with the blower turned off (this is what happens in servers: powerful fans ensure flow for all components and you don't need GPU fans, as in this Tesla card: http://www.nvidia.com/object/tesla-servers.html).

Anyway, I might be very wrong here but that's why I'm asking if someone with a blower-type GPU and a decent case can try experimenting with the case fans to see what the effect is on the blower.

Actually the best air flow through you case happens when the differential is zero. I think you are confusing the specs a bit.

Airflow out of your case would be better if you had positive pressure inside but you have to build that pressure with a fan and that is an ineffective path to go down, it just won't do a whole lot.

A "negative pressure" setup has so little negative pressure in it that it is insignificant compared to the pressure required to blow air through the GPU heat sink.

I'd add one thing for a niche case:

If you have a passive PSU, or one that turns its fan off during low draw; then a positive or negative design is preferred over balance, as it'll move air though what might otherwise be a heat trap.