deciding what Computer to get

[Arcamax]

so then i should add an extra dap of thermal paste on the heat sink then right.  will the stuff i have do or should i get this:
Arctic Silver 5 High-Density Polysynthetic Silver 3.5g Thermal Paste

[NiHaoMike]

4GHz seems like a low clock for a 3930k. I think the average clock for SB-E at safe voltages was 4.9-5GHz. Average for IB-E was 4.7. Average for Haswell-E is 4.5. Try your thermal tests at the highest clock you can get at safe 24/7 voltages. You will find that the evo will either limit your max oc or you will have to turn fans up to the point where the system is too annoying.

I'm going for reliability and long lifetime, not hardcore benching. Just the fact that a fairly cheap cooler works great on a lightly overclocked 3930k is quite amazing in itself. Also surprising is how the servers get away with much smaller heatsinks even on the really hot 18 core Haswell-Es.

Nowadays, I do my Folding with my GPU.

[nctnico]

Just the fact that a fairly cheap cooler works great on a lightly overclocked 3930k is quite amazing in itself. Also surprising is how the servers get away with much smaller heatsinks even on the really hot 18 core Haswell-Es.

That is what I have been writing all along: a standard case has no proper airflow so the cooler is recycling it's own hot air which gets hotter and hotter. In my Dell the CPU doesn't even have it's own fan and when running it at full power (for example a compile job which takes about an hour with 10 parallel jobs) the fans don't even increase speed. You don't need much airflow to move a lot of heat but you need proper ducting to get cool air in and the hot air out.

[Armxnian]

so then i should add an extra dap of thermal paste on the heat sink then right.  will the stuff i have do or should i get this:
Arctic Silver 5 High-Density Polysynthetic Silver 3.5g Thermal Paste

The evo already has a layer of paste that is applied by a machine. You don't need to add more paste to anything for the first install. If you remove the heatsink then clean the old paste off the heatsink and the cpu with some alcohol and apply a small dot in the middle of the cpu. Your stuff most likely work but I can't comment on performance since I've never seen or used it. AS5 is a bit outdated, there are pastes that perform better and apply as easily so might as well get that if buying new. I use GC extreme. If you can't get that in Canada then just get AS5 since everything performs close to each other, except chocolate.

4GHz seems like a low clock for a 3930k. I think the average clock for SB-E at safe voltages was 4.9-5GHz. Average for IB-E was 4.7. Average for Haswell-E is 4.5. Try your thermal tests at the highest clock you can get at safe 24/7 voltages. You will find that the evo will either limit your max oc or you will have to turn fans up to the point where the system is too annoying.

I'm going for reliability and long lifetime, not hardcore benching. Just the fact that a fairly cheap cooler works great on a lightly overclocked 3930k is quite amazing in itself. Also surprising is how the servers get away with much smaller heatsinks even on the really hot 18 core Haswell-Es.

Nowadays, I do my Folding with my GPU.

Hardcore benching involves increasing the voltage above the 24/7 safe point. Take Haswell-E for example where the safe voltage is about 1.3v. Run at that and your cpu will last until it's time to replace it. For benching you can go 1.35v or more if you want but you might lower the life span. Go more and you can degrade it, but you can always buy the intel tuning plan.

A xeon on an evo is different than a consumer chip on an evo. They are designed for low power and low voltage operation. They are much higher grade chips. For example the 18 core Haswell-E has basically the same TDP as the consumer Haswell-E chips like the 5820k, yet it has 12 more cores. They also cannot be overclocked much. Here is the 18 core xeon:
It runs at a max of 51c under load with an AIO cpu cooler and on a test bench where there is basically no air flow.

Just the fact that a fairly cheap cooler works great on a lightly overclocked 3930k is quite amazing in itself. Also surprising is how the servers get away with much smaller heatsinks even on the really hot 18 core Haswell-Es.

That is what I have been writing all along: a standard case has no proper airflow so the cooler is recycling it's own hot air which gets hotter and hotter. In my Dell the CPU doesn't even have it's own fan and when running it at full power (for example a compile job which takes about an hour with 10 parallel jobs) the fans don't even increase speed. You don't need much airflow to move a lot of heat but you need proper ducting to get cool air in and the hot air out.

Every standard case has airflow. Fans in the front pull cool air in, and fans at the top and back exhaust the hot air. Hot air rises on its own, you don't need fancy air ducts. If they made such a difference, why has no case manufacturer created a case with ducts? It would be easy to have large, medium and small sizes to fit different hardware. My build puts out a lot of heat, to the point where my room becomes an oven combined with the fact that the temperature in Vegas today was 106. My build is silent when stressing both the CPU and GPU. You say there is a problem with airflow in "standard ATX" cases but there isn't.

[nctnico]

Every standard case has airflow.

No they don't. Just blowing air into a box is not a guarantee it goes where it needs to go. A CPU cooler, obstructions, other fans cause all kinds of weird airflows inside a case. The only way is using well planned ducts.

[Armxnian]

Every standard case has airflow.

Just blowing air into a box is not a guarantee it goes where it needs to go.

Yes it does. It's obvious where the air goes, because the fans direct the airflow, not the case.

Air follows a path. Every component that needs to be cooled is on that path.

If anything, ducts actually reduce cooling in such a small system because the fans on the heatsinks, if any, have to now pull air from the side instead of the bottom or top since they are restricted by a piece of sheet metal. You mentioned your Dell cpu heatsink does not have a fan on it. That's a terrible design because the only fan moving air through that heatsink is located far away, meaning there is basically 0 static pressure. If you have a high fin density radiator, which you need for a small heatsink to fit in a small case, there is a very low amount of air moving through the heatsink. Just because it works in your machine doesn't mean it will work in another or increase efficiency. Also, just because your machine is quiet doesn't mean others are not. I already gave an example of my build whixh is silent and puts out more heat than yours. There are countless other examples.

If there was a problem with standard cases then the case manufacturers would have done something about it. I bet any reptuable manufacturer has air chambers and other equipment to test their designs. But so far only you have found a problem, and have no evidence (numbers) to support your claims, so there is no need to further continue the speculation, as it's already off topic from the thread.

[nctnico]

Every standard case has airflow.

Just blowing air into a box is not a guarantee it goes where it needs to go.

Yes it does. It's obvious where the air goes, because the fans direct the airflow, not the case.

Air follows a path. Every component that needs to be cooled is on that path.

Just because someone in the marketing department drew a pretty picture
Hint: turbulence

[Armxnian]

I don't think that's from the marketing department, but even if it is it doesn't mean it's wrong. You can test it yourself with a smoke pencil or the fumes from your soldering iron.

Turbulence is a good thing. If helps push air past obstacles. Hot air rises and there are fans at the top meaning air is going to be exhausted and not recycled.

It's not like Dell rolls their own silicon and makes their builds from scratch. They buy and assemble components that everyone else uses. If air ducts made a significant impact then someone would have done it already. There are cases worth more than entire systems so the only limit is imagination.

[rdl]

...
In my Dell the CPU doesn't even have it's own fan and when running it at full power (for example a compile job which takes about an hour with 10 parallel jobs) the fans don't even increase speed.

Just a guess, but the fact that the cpu doesn't have its own fan is probably the only reason there's a duct.

[nctnico]

...
In my Dell the CPU doesn't even have it's own fan and when running it at full power (for example a compile job which takes about an hour with 10 parallel jobs) the fans don't even increase speed.

Just a guess, but the fact that the cpu doesn't have its own fan is probably the only reason there's a duct.

Not quite right. The CPU doesn't need a fan because there is a duct which leads air from the front fan to the CPU and from the CPU to the rear of the case. Same goes for the memory. Now imagine the system had a fan on the CPU heatsink... Where does that fan draw air from? The air will probably circle around the heatsink back into the fan. This air circling around obstructs the airflow through the case so the CPU will have less cooling than in the current situation. See how just putting fans in a case without guiding the airflow is such a bad idea? A standard case maker cannot plan for having ducts to guide the airflow since they have no idea where the CPU or other components are going to be.

Also the power supply sucking air out of or pushing air into the case is just bad. In the first situation the PSU gets all the hot air and pulls all the hot air from the system towards the CPU. Reverse the airflow and the CPU gets all the hot air from the PSU and the rest of the systems gets the hot air from the CPU + PSU. My Dell on the other hand has a PSU which pulls air from the front and pushes it out at the rear and carefully avoids to push or pull air from the inside of the case.

[Armxnian]

Let me explain the airflow in a standard case with a top mounted radiator for the cpu cooler. Fresh/cool air is drawn in from the front, preferably by fans. This air gets instantly picked up by the gpu fans and cpu fans at the top. The air then flows over the entire motherboard providing cooling to the components. The gpu also exhausts air, but the air is hot and rises quickly. All of the air then gets exhausted through the radiator. If the air is too hot then you can add a fan in the back to split the air into two exhaust ports. The power supply is not even in the chain in a proper case. The power supply fan draws air in from the bottom of the case and exhausts it internally from the back. It makes no contact with the air inside the case.

The same goes for a heatsink mounted right on the cpu. I think this is less efficient but you will always have a fan pulling or pushing air out the back so nothing gets recycled. Plus you can add fans to the top.

Ducting can help if there is significant leakage of fresh air or significant distribution of hot air. But this is not a problem in a standard computer where the chassis is so small. There is a very short path that the air has to travel. The fresh air that comes in is exhausted in a matter of seconds even with low speed fans.

If you don't believe me then just check the ambient temperature of any system. They are drastically lower than what any component in the system is rated for. As long as your motherboard has decent airflow over it you are fine. The only thing that matters is the temperature of core components like the cpu and gpu. Airflow does not impact the temperature of these components by much, but can reduce efficiency. Once again, with decent airflow you are fine. If the temperature of core components is normal and your build doesn't sound like a vacuum then you are fine. Anyone can build a system that meets these two requirements. There is no problem that needs to be solved with how current cases work.

[nctnico]

Read on the subject called 'laminar flow' (yes this also applies to gasses) and how making small corridors (=ducts) helps to decrease turbulence and thus increase airflow. Just blowing air into a box isn't going to be the best way! Math proves that: https://en.wikipedia.org/wiki/Reynolds_number

[Armxnian]

You're right blowing air into a case would alone be awful. But you're also actively exhausting the air. In a duct, air is forced in one direction, so it escapes in the opposite direction. When you have an intake and exhaust in close proximity, like a computer chassis, you essentially create an air duct. Some air will leak but that is fine, the majority of the air will follow the path to replace the air which is displaced. Air ducts and vents are useful to carry air a long distance, like in HVAC.

[WZOLL]

If you look at any professional (http://www.digitalstorm.com/vanquish-3.asp) /consumer built computer it is just fans blowing air in and out of a case with no fancy ducts or dividers. It is also important to keep in mind air pressure. A positive air pressure in the case (more  air in than out) will push out dust while a negative air pressure (more exhaust than intake will suck in dust. Or just water cool it and keep the rads outside the case.

[nctnico]

If you look at any professional (http://www.digitalstorm.com/vanquish-3.asp) /consumer built computer it is just fans blowing air in and out of a case with no fancy ducts or dividers.

No air flow control = not professional. That digital storm stuff is just like any other consumer crap.

It is also important to keep in mind air pressure. A positive air pressure in the case (more  air in than out) will push out dust while a negative air pressure (more exhaust than intake will suck in dust.

I have experimented a lot with that because I have lots of dust and the direction of the airflow makes no difference at all regarding the build up of dust inside the system. I did find out that having a lot of airflow accellerates the build up of dust. My new Dell stays much cleaner than my old systems in which I needed to remove the dust every 6 months otherwise the CPU heatsink would clog up causing it to throttle down the CPU speed.

[rdl]

The best way to reduce the amount of dust that gets inside a computer is with filters. The only advantage to "positive pressure" is that the source of incoming air (that needs to be filtered) is more controllable. Just putting some kind of filter on the intake side of any fans that a pulling air into the case will make a big difference. With "negative pressure", air is pulled in from the outside through every available opening (other than the exhaust), which makes it pretty much impossible to keep dust out.

All filters will reduce air flow, but some are better than others. Most of the time, two 120mm fans in a mid tower size case will still provide plenty of cooling.

[Galenbo]

Read on the subject called 'laminar flow' (yes this also applies to gasses) and how making small corridors (=ducts) helps to decrease turbulence and thus increase airflow...

But is heat transfer between component and air different between laminar and turbulent, asks the professional himself.

[Galenbo]

... A positive air pressure in the case (more  air in than out)

You mean like inflating the case to 8 bar like the tank of a compressor?
That could temporarily cause more air in then out.

Or just water cool it and keep the rads outside the case.

Also watercool the hard drives then.

[Armxnian]

This is what you've been raving about lol?



What fan ducts? Even dell doesn't use a fan duct. That cover they have on the cpu cooler is to force the air from the top mounted cpu cooler to go out the back and not into the case, which isn't a problem for a right angle heatsink or a AIO mounted in the top of the case as the fans push the air out of the case to begin with. And the components they use look pretty poor quality to make it as cheap as possible. We have about 2000 of these desktops at my school.

[Arcamax]

hi armxnian and everyone else out there. armxian i changed a few things from the list http://pcpartpicker.com/p/PsDxJx you gave me, what do think of this in the attachment.

[Armxnian]

hi armxnian and everyone else out there. armxian i changed a few things from the list http://pcpartpicker.com/p/PsDxJx you gave me, what do think of this in the attachment.

Well I can't comment on the prices since I know nothing about the market value of items in Canada, even if they seem a bit high, but the components themselves look good. Read up on overclocking the G3258 while you wait for the parts to arrive, I'm sure there are hundreds of guides.

[WZOLL]

No air flow control = not professional. That digital storm stuff is just like any other consumer crap.

 What about alienware? It is made by dell and does not have any airflow control. the last computer that I've seen with airflow control had a floppy drive. well to each their own cooling technique

[rdl]

Skip the aftermarket cooler, it's unnecessary. Buy the processor with the stock Intel heat sink included and save some money. You can find a case that will work fine for 1/3 the cost of that overpriced fractal design. It's nice, but you're paying a lot for style and brand name.

[Armxnian]

Skip the aftermarket cooler, it's unnecessary. Buy the processor with the stock Intel heat sink included and save some money. You can find a case that will work fine for 1/3 the cost of that overpriced fractal design. It's nice, but you're paying a lot for style and brand name.

I've already explained that once you overclock, the stock cooler will be too loud. The stock cooler is for running microsoft office on a stock cpu, like the G3220 I have in a spare rig, not a G3258 overclocked to the max running an intensive program like Arcamax intends to do. Plus the stock cooler blows all the hot air into your case, creating the scenario nctnico was complaining about.

He is in Canada, everything is overpriced. You can't compare the price of that case in Canada to other cases on a U.S site. Anything less will result in a cheap plastic case. Spending a little more on a cpu cooler and a decent case will result in a quiet build with parts you can reuse.

[rdl]

AutoCad does not require over-clocking. Did I miss where the he wanted to run something more demanding?

I did overlook the Canadian pricing. The Fractal case is more like only double the cost of a decent case. That's probably not worth worrying about, and there could be good reasons to use an ATX size case for a micro ATX motherboard.