• The site has now migrated to Xenforo 2. If you see any issues with the forum operation, please post them in the feedback thread.
  • Due to issues with external spam filters, QQ is currently unable to send any mail to Microsoft E-mail addresses. This includes any account at live.com, hotmail.com or msn.com. Signing up to the forum with one of these addresses will result in your verification E-mail never arriving. For best results, please use a different E-mail provider for your QQ address.
  • For prospective new members, a word of warning: don't use common names like Dennis, Simon, or Kenny if you decide to create an account. Spammers have used them all before you and gotten those names flagged in the anti-spam databases. Your account registration will be rejected because of it.
  • Since it has happened MULTIPLE times now, I want to be very clear about this. You do not get to abandon an account and create a new one. You do not get to pass an account to someone else and create a new one. If you do so anyway, you will be banned for creating sockpuppets.
  • Due to the actions of particularly persistent spammers and trolls, we will be banning disposable email addresses from today onward.
  • The rules regarding NSFW links have been updated. See here for details.

Tech advice requested

I'll do a more in depth discussion after work.

Also as far as liquid cooling goes, I was meaning good quality thermal paste for the best heat transfer as well as a compression tower to actually pump cooled liquid through the system, hence why I made the mention of the noise and power use. Anything less is obviously just a fancy method of air-cooling because that's what the end medium to dispose of the heat is and that's the bottleneck for what it can actually do.

I am considering it because 90+ and sometimes 100+ degrees F is what can be expected in the heat of summer here. I basically don't play games during those times because when I try i tend to hit my automatic temperature shutoff point. Damn glad by MB came with that, probably would have fried by CPU several times by now.
 
In terms of graphics cards, AMD has recently shown very promising performance on the first DirectX 12 assessments, even with their older cards. Apparently all those parallel processing units that made them popular for bitcoin mining a while back are actually being put to use by the new standard. nVidia's serialized GPU processing reigns supreme for DX11 and lower games though. Not by nearly enough to push AMD out of the mid-level market though. If you want future-proofed I'd go with AMD right now. Perhaps something from their 300 series for affordability.
Yeah I've been Following this since the Ashes benchmark started getting reported.

The big issue is Asynchronous Compute performance.

Contrary to a lot of gloom and dooming online, The Maxwell 2 Architecture that Nvidia used with their 900 series cards, is fully capable of doing it, but Nvidia asked Oxide, the guys developing the game Ashes of the Singularity, an RTS that is currently in Alpha level development, to turn The feature OFF for Nvidia Hardware.

This is because Nvidia hardware was posting very weird and detrimental performance numbers when running the feature.

Why is that?

Because AMD implemented the feature at the hardware level, their GPU's themselves schedule task for each ACE subsystem, and they can Preempt tasks at any time for any ACE.

Nvidia implemented a Driver level Scheduler, meaning if something goes wrong or something needs to be Preempted on the GPU, they have to wait till all tasks complete for all Shader Engines for the entire Queue before they can assign new tasks.

This means, Nvidia hardware spends more time Waiting, whereas AMD hardware can Preempt tasks and switch what the various ACE's are doing at any time. Saving them precious milliseconds.

This feature is new for DX12, so it was completely worthless until new to have this capability. It's highly expected for Nvidia to remedy this problem with a driver update, not completely mind you, but also with new Hardware coming next year with their Pascal core architecture.

So it's not that Nvidia cant use the feature, just that they kind of suck at it without a lot of very careful, fiddly, programming.

CPUs...well who do you want? Intel comes with a significant premium and they tend to strip out useful features in all but their top-of-the-line chips, but they have better serialized performance and power efficiency. AMD's performance isn't as good, but their feature set is much more consistent across each chipset in a family. I recommend the FX-Series if you go that direction. With Intel you have to worry about whether one i5 is merely slower than another or if they actually stripped out Virtualization support, HyperThreading, or more.

AMD chips are several years old at this point, barring some edge cases with low to midrange parts. The motherboard Tech they are using for even their "New" products is laughably old and will be replaced with ZEN cores start rolling, eventually.

But AMD did have a thing for heavily threaded performance. Which benefits some tasks, like Video rendering, most. But Intel's superior Serial performance really has dominated the consumer marketplace for some time, relegating AMD to budget brand status for some time.

For a power supply I'd recommend at least 600W, possibly 800, depending on your graphics card, desire to overclock, number of drives, etc. Especially if you anticipate using all of them at once.

I agree. The convetional wisdom right now is you dont "Need" anything over a 500w, but I like having a good margin to account for things like Multi-GPU configs, large drive arrays, and Zero Fan speed modes. The less load you put on your PSU the less active cooling it needs. With enough over capacity you can have your PSU kick into a mode where it stops it's fan because it doesn't need the active airflow.

Liquid Cooling is an excellent choice for all occasions, but you might be surprised that quality thermal paste can have a significant impact as well. Arctic Silver is always a good standby, but just about anything is better than stock. I've had this $8 purchase decrease my CPU temperature by nearly 10°C before.

Liquid cooling is expensive and has maintenance requirements. But a good Thermal paste is a GREAT investment. Just stay away from those "Liquid Metal" type pastes, as they are really really annoying if some gets away from where you want it.

Now Motherboards...Stay away from Gigabyte. Every single one I've seen in the past year or two had something (usually more than one something) major go wrong with them. Not all that familiar with others at the moment, sorry.

I've also heard bad things about Gigabyte boards. MSI, or ASUS tend to be more reliable.

I am considering it because 90+ and sometimes 100+ degrees F is what can be expected in the heat of summer here. I basically don't play games during those times because when I try i tend to hit my automatic temperature shutoff point. Damn glad by MB came with that, probably would have fried by CPU several times by now.

Your basically talking about a thermoelectric cooler. A Peltier basically.

They can be real fiddly if you dont know exactly what your doing. You have to insulate your motherboard with putty because they put off so much cold, that the condensation alone can kill your system.

Also, they are loud. Because all that cold comes at a price, the compressor is noisy and you need sufficient air movement to cool the heat generated by the system because of the noble gas law.
 
Yeah I've been Following this since the Ashes benchmark started getting reported.

The big issue is Asynchronous Compute performance.

Contrary to a lot of gloom and dooming online, The Maxwell 2 Architecture that Nvidia used with their 900 series cards, is fully capable of doing it, but Nvidia asked Oxide, the guys developing the game Ashes of the Singularity, an RTS that is currently in Alpha level development, to turn The feature OFF for Nvidia Hardware.

This is because Nvidia hardware was posting very weird and detrimental performance numbers when running the feature.

Why is that?

Because AMD implemented the feature at the hardware level, their GPU's themselves schedule task for each ACE subsystem, and they can Preempt tasks at any time for any ACE.

Nvidia implemented a Driver level Scheduler, meaning if something goes wrong or something needs to be Preempted on the GPU, they have to wait till all tasks complete for all Shader Engines for the entire Queue before they can assign new tasks.

This means, Nvidia hardware spends more time Waiting, whereas AMD hardware can Preempt tasks and switch what the various ACE's are doing at any time. Saving them precious milliseconds.

This feature is new for DX12, so it was completely worthless until new to have this capability. It's highly expected for Nvidia to remedy this problem with a driver update, not completely mind you, but also with new Hardware coming next year with their Pascal core architecture.

So it's not that Nvidia cant use the feature, just that they kind of suck at it without a lot of very careful, fiddly, programming.
Even when nVidia comes out with cards that take advantage of it AMD has years of a head start and its already implemented in hardware in all but their most budget cards. Your assertion that it can be implemented driver-side is probable, but having to communicate all the way to the CPU and back will have a significant impact on performance no matter what they do and I'm not confident they'll do so retroactively for their midrange cards now. If you want to buy a card you know is future-proofed now, you buy AMD.
AMD chips are several years old at this point, barring some edge cases with low to midrange parts. The motherboard Tech they are using for even their "New" products is laughably old and will be replaced with ZEN cores start rolling, eventually.

But AMD did have a thing for heavily threaded performance. Which benefits some tasks, like Video rendering, most. But Intel's superior Serial performance really has dominated the consumer marketplace for some time, relegating AMD to budget brand status for some time.
I'm more concerned about how Intel markets a line of chips as having a feature, but then notes in the release notes that certain chips in that line don't actually have it. I've been burned by that before and those missing features can kill performance. Also, I've had luck getting AMD's high-end chips for around the same price as Intel's mid-high range ones. If you're satisfied with that level of performance you can eke out more performance from the AMD chips than the Intel ones of comparable price. Though they do have a higher wattage.
Liquid cooling is expensive and has maintenance requirements.
Some companies like Corsair and CoolerMaster are now offering pre-filled designs that they claim will never need refilling for ~$50-60.
But a good Thermal paste is a GREAT investment. Just stay away from those "Liquid Metal" type pastes, as they are really really annoying if some gets away from where you want it.
I knew I forgot something. I've always tried to stick with the non-electrically conductive types, myself. That way it doesn't cause a short if you mess up.
I've also heard bad things about Gigabyte boards. MSI, or ASUS tend to be more reliable.
Good to know, thanks.
 
Last edited:
Even if/when nVidia comes out with cards that take advantage of it AMD has years of a head start and basically all of their lineup has it implemented in hardware. Your assertion that it can be implemented driver-side is probable, but having to communicate all the way to the CPU and back will have a significant impact on performance no matter what they do and I'm not confident they'll do so retroactively for their midrange cards now. If you want to buy a card you know is future-proofed now, you buy AMD.
Exactly. Nvidia is pretty much hosed when it comes to Async Compute performance because of their shortsightedness.

That being said, both companies are expected to release new product ranges next year with some big improvements.

Nvidia Side - Jumping on the HBM bandwagon with Gen 2 of the tech, providing 8-32gb of Vram in extremely compact form factors. Pascal is the big hope for them for keeping market share in high end performance.

AMD - Continuing their HBM advancement with Gen 2 of the tech, also providing 8-32gb of Vram in extremely compact form factors. They have a lot to prove here and they are planning on 3 ranges of cards in their new "Artic Islands" series of cores. They have a lot to prove to show that they can compete with Nvidia and boost their thermal perfomance as well as their gaming performance. They have a big lead with Async Compute performance thanks to extremely forward thinking architecture design. They have recently restructured their graphics division and given control over to a very very respected AMD engineer.

I'm more concerned about how Intel markets a line of chips as having a feature, but then notes in the release notes that certain chips in that line don't actually have it. I've been burned by that before and those missing features can kill performance. Also, I've had luck getting AMD's high-end chips for around the same price as Intel's mid-high range ones. If you're satisfied with that level of performance you can eke out more performance from the AMD chips than the Intel ones and save money at the same time. Though they do have a higher wattage.
Yeah. I find both companies product lines a bit confusing in some ways. Intel is my preferred MFG here, simply in a price performance aspect. AMD's gamble with Bulldozer didn't really pan out because software devs didn't take advantage of the multithreading capability, also Windows had a bug where it didn't know how to properly assign tasks to the Bulldozer cores without overloading the execution units.

Intel has had a bit of a hard time lately, with multiple generations of chips being lackluster in performance gains over previous hardware, namely everything after Sandy Bridge came out has been a sub 10% performance boost at best, If at all.

And then there was the abortion that was Broadwell, where their fabrication tech just wasn't up to the task of 14nm production and they had to push back an expected new generation by an entire year, and even now you cant get the chips because they didn't make many of them.

And then you also have the rise of the -E brand for Intel. Further fragmenting their market. Although it's a smart business move because the -E class chips are all salvaged dies from Server chips. Now marketed and sold at huge markup to "Enthusiasts"

Liquid cooling is not that expensive and some companies like CoolerMaster are now offering pre-filled designs that they claim will never need refilling for ~$60-70.
AIO coolers, like the corsair and coolermaster brands, dont need maintenance because they are completely sealed systems.

When I heard "Water Cooling" I hear "Custom Loop." Which does need maintenance. And custom loops cost a lot in materials and fittings.

But those AIO coolers are perfectly usable
I knew I forgot something. I've always tried to stick with the non-electrically conductive types, myself. That way it doesn't cause a short if you mess up.
Let me tell you, the liquid metal "Pastes" I've seen are amazing for cooling performance, but if something goes wrong, you can get boned. Really really fast.

I had to help some guy wash some out of his socket, because it dripped into the socket. At least he had the presence of mind to not try booting his system like that.


.
 
CPUs...well who do you want? Intel comes with a significant premium and tends to strip out useful features in all but their top-of-the-line chips, but they have better serialized performance and power efficiency. AMD's performance isn't as good, but their feature set is much more consistent across each chipset in a family. I recommend the FX-Series if you go that direction. With Intel you have to worry about whether one i5 is merely slower than another or if they actually stripped out Virtualization support, HyperThreading, or more.

At the moment (and it has been this way for a few years now) the only reason to go AMD is if you're on a tight budget and want a really cheap CPU/GPU combo with a cheap motherboard to go with it. That definitely isn't the case here, so the sane choices are Intel, Intel or Intel. Yes, that's despite any market fragmentation they've got with their models, low per-gen improvements, and the 14nm issues. The performance gap between Intel and AMD is just that big now.
 
Liquid Cooling really isn't all that different from traditional Air cooling.

There is this misconception that it can magically cool your system no matter what.

It cant.

You can not go beyond Ambient Temperatures, without some kind of compression cooling system, which is a whole nother ball of bees best left to professionals and the insane.

So if your room is 80 Deg F, You can never get your system below that on air or water without some extreme shenanigans.

A water cooling rig can be prohibitively expensive, because you have all kinds of fittings and hardware you have to buy to make it work. It also has a lot of maintenance you need to do year round to ensure the loop works correctly, and when it breaks, it can be catastrophic.

This is why things like the AIO water coolers that different MFG's put out, are more popular than building a custom loop.

They are as simple as normal Air coolers to install, no maintenance required, and give you a significant fraction of the performance of a custom loop.

What's the advantage to using a Water system at all?

Your moving the heat up off the Die, and dissipating it into the air more efficiently than a traditional Air cooler.
All I know is the absolute BEST cooled (without resorting to liquid-hydrogen, which I have seen used on some supercomputers) computer I've ever seen was built with all the electronics exposed, inside a fish-tank filled with mineral oil (everything was connected together before the mineral oil was added).

This results in essentially perfect instantaneous heat dispersion. The only crevat is the system must be kept COMPLETELY dust free. Not a spec of dust on any of the components before immersion in the oil, and no dust can get into the oil once the system is completed.

It's not good for if you plan overclocking, since the oil shouldn't get hotter than 50o​ C (122o​ F) and the radiators typically can't keep up with an overclocked CPU/GPU. But if you don't plan on overclocking then you're looking at a computer with decades of lifespan, rather than years (barring mechanical failure).
Oh, and my CPU fan fell off a month or two ago, but that was just because I didn't screw it in right back in October. It could happen to anyone and I'm frankly impressed it stayed on that long. :p I just stuck it back on with some leftover thermal paste and twisted the fasteners to the right this time, problem solved.

Or was it left...?
Please tell me you scraped the old thermal paste off before applying the new thermal paste...

If you didn't then right now turn off your computer, open it up, pull the CPU fan off, scrape off ALL of the paste currently on there (use a q-tip and/or cotton ball and 90% rubbing alcohol, and a NEW razor blade[male SURE it's a perfectly straight blade with no nicks, and be VERY careful not to leave any scratches] to remove the old paste), and re-apply some new paste.

If you just put a new dab on and re-applied the CPU fan you are 100% guaranteed to have air bubbles in there, and those will kill the Thermo-conductivity of the paste.

As a note, when applying the thermal paste, place a dab in the center of the CPU and then place the CPU fan/heat sink, allowing the Heat sink to spread the paste out. Spreading the paste out with a credit card or something similar is another way to introduce air bubbles (and I WILL prove this if anyone challenges me on this; better yet, I'll get you to prove this to yourselves, with SCIENCE!!! *laughs maniacally*)
...shit. I have a copy of Windows 7 but it's on a disc and I don't have an optic drive.

Can a Windows installation disc be loaded onto an external hard drive and booted from there, or would a new system's USB ports not work without drivers installed?
http://www.microsoft.com/en-us/download/windows-usb-dvd-download-tool

That right there is everything you need, well baring a computer with an optical drive and a USB port.

Basically:
-Install that little tool on the computer with the optical drive.
-Pop the Win 7 install disk into the optical drive.
-Pop a 4 GB or larger flash drive in the USB port. (it will be formatted, so nothing on there you want to keep)
-Run the tool. Point it at the USB drive, and the optical drive, and sStart the program.

After a few minutes the program will finish, and your flash drive will now be a bootable Windows install disk (of whichever version the disk was). As a bonus, install from a flash drive is considerably faster than installing from the disk.

There will be a slight issue if your disk is x64 and the computer you're creating the flash drive installer on is x86, or visa-versa; but this is easily fixed, just a matter of putting a specific file in the tool's root folder before running it. If this is an issue with you, just say so and I'll dig up a link to the file you need (not that hard to find if you know what you're looking for, it's not exactly a rare problem).

Pretty much all new motherboards natively support booting from USB (even without a HDD attached), and most older ones will support it with a firmware update.



As to silver-based thermal paste (and never bother with a metallic thermal paste that isn't silver). It's one of the best thermal pastes out there, although there are better ones, but the thing is you should only use it if you KNOW what you're doing.

Put 1 ml too much on, and it will bleed over the edges of the CPU and probably get into the sockets, and even if it doesn't it could short the connections going into the sockets. Basically, the tiniest bit too much and you fry your CPU, and probably your MB as well. A drop falling loosely on your MB is BAD as well.

Alternate, put to little on, and you don't get enough contact between the CPU and the heatsink, overheating your CPU.

If you're just a hobbyist, or working off of youtube videos, then stay away from silver-based thermal paste.

http://www.walmart.com/ip/15907612?...81550832&wl4=&wl5=pla&wl6=78811124072&veh=sem

This thermal paste works well if you plan on overclocking, and it relatively cheap.
 
Blargh. Work ran me ragged so no technical thinking right now.

Keep up the discussions though, it's pretty helpful....or it will be once I am able to process it fully.

I'mma go type up some smut for Ack to assemble.
 
All I know is the absolute BEST cooled (without resorting to liquid-hydrogen, which I have seen used on some supercomputers) computer I've ever seen was built with all the electronics exposed, inside a fish-tank filled with mineral oil (everything was connected together before the mineral oil was added).

This results in essentially perfect instantaneous heat dispersion. The only crevat is the system must be kept COMPLETELY dust free. Not a spec of dust on any of the components before immersion in the oil, and no dust can get into the oil once the system is completed.

It's not good for if you plan overclocking, since the oil shouldn't get hotter than 50o C (122o F) and the radiators typically can't keep up with an overclocked CPU/GPU. But if you don't plan on overclocking then you're looking at a computer with decades of lifespan, rather than years (barring mechanical failure).
Yeah, Mineral oil machines are a real bitch to mess with. You have to be super careful just what you put into them, because the oil can Wick its way up the cables you have plugged into it, and the oil can eat into certain types of sleeves used on cables.

Also, the mineral oil itself needs to have some way to dissipate heat, although generally the sheer size of the tank acts as an efficient enough heatsink for the oil.

But those still have the issue with ambient temperature delta. Even a mineral cooled machine cant go below ambient temps.

doomlord9

You have asked about higher performance cooling, given your situation is quite warm with 90-100deg ambient temps.

Really there are few options for trying to go below ambient temp, and all of them are fiddly as heck, and not cost effective for the most part.

You are better off saving some money, by turning your AC up, or getting something like a window box AC unit to cool just that room.
 
Also, the mineral oil itself needs to have some way to dissipate heat, although generally the sheer size of the tank acts as an efficient enough heatsink for the oil.
Check the link in my post. The units pump the oil through radiators to dissipate heat into the ambient air.

And I see zero reason why a refrigeration/compression/ect. unit can't be set up to cool the oil down further; there are such setups for fishtanks after all (for arctic fish), so the tech does exist.
 
And I see zero reason why a refrigeration/compression/ect. unit can't be set up to cool the oil down further; there are such setups for fishtanks after all (for arctic fish), so the tech does exist.
Yeah you could use something like a water chiller, then coil some of the oil in a tube through the chilled water. But this further reduces efficiency of the overall system and will end up costing more than just turning the A/C up for the house, or getting an A/C unit for the room the system will be in.

In absolute terms it's possible to do this, but it will end up costing a butt ton more than just using something more designed for the purpose of cooling the air down. Because the added cooling capacity you need to chill the system down will be like plugging a very large refrigerator into your house and running it pretty hard, all the time. Because that's essentially what you would be doing.

The reason why it's so costly when a fridge is so big, is fridges aren't meant to handle a constant thermal load like what a computer can put out.

Get one of those ultra tiny 3-5w PC's and sure, a fridge could possibly keep that cool. But a full system it can't. And a modern PC like what the OP is looking for is going to need to dissipate 350-400 watts at minimum.
 
Yeah you could use something like a water chiller, then coil some of the oil in a tube through the chilled water. But this further reduces efficiency of the overall system and will end up costing more than just turning the A/C up for the house, or getting an A/C unit for the room the system will be in.

In absolute terms it's possible to do this, but it will end up costing a butt ton more than just using something more designed for the purpose of cooling the air down. Because the added cooling capacity you need to chill the system down will be like plugging a very large refrigerator into your house and running it pretty hard, all the time. Because that's essentially what you would be doing.

The reason why it's so costly when a fridge is so big, is fridges aren't meant to handle a constant thermal load like what a computer can put out.

Get one of those ultra tiny 3-5w PC's and sure, a fridge could possibly keep that cool. But a full system it can't. And a modern PC like what the OP is looking for is going to need to dissipate 350-400 watts at minimum.
I'm just gonna say we're probably thinking of very different types of refrigeration tech. I wasn't talking about sticking the fishtank in a refrigerator; I was talking about a unit that sucks in the oil, cools it down, then puts it back in the tank in a loop; or having some coils in the oil connected to a small condenser, or similar unit, cycling something cold through them to cool the oil down.

There are plenty of ways to pull it off without involving refrigerator-level energy costs. Some of the more energy-efficient methods would use a bit less power than the computer itself.
 
I'm just gonna say we're probably thinking of very different types of refrigeration tech. I wasn't talking about sticking the fishtank in a refrigerator; I was talking about a unit that sucks in the oil, cools it down, then puts it back in the tank in a loop; or having some coils in the oil connected to a small condenser, or similar unit, cycling something cold through them to cool the oil down.

There are plenty of ways to pull it off without involving refrigerator-level energy costs. Some of the more energy-efficient methods would use a bit less power than the computer itself.

That's the problem with refrigeration tech. It's stupidly inefficient.

Say your hypothetical computer only puts out 100 Watts of heat. And your fridge tech injects 50 watts of cooling into the tank on top of the radiator system you have in place to dissipate the other 50 Watts of heat from the tank.

Your fridge system now has to deal with dissipating 150 Watts that has now built up in itself.

So your only going to move the problem from trying to cool your computer, to trying to keep the fridge system cool. And a fridge system that can withstand a constant thermal load like what a realistic computer puts out is way more power hungry than your normal fridge in a home.

A "Normal" computer at load, especially something like what the OP is talking about, will put out on average around 450 Watts when under load. When Idle it will sit somewhere around 200 Watts.

A normal house fridge, only generally has to be able to ramp up to intermittent 25 Watt load.

It's a BIG jump.
 
That's the problem with refrigeration tech. It's stupidly inefficient.

Say your hypothetical computer only puts out 100 Watts of heat. And your fridge tech injects 50 watts of cooling into the tank on top of the radiator system you have in place to dissipate the other 50 Watts of heat from the tank.

Your fridge system now has to deal with dissipating 150 Watts that has now built up in itself.

So your only going to move the problem from trying to cool your computer, to trying to keep the fridge system cool. And a fridge system that can withstand a constant thermal load like what a realistic computer puts out is way more power hungry than your normal fridge in a home.

A "Normal" computer at load, especially something like what the OP is talking about, will put out on average around 450 Watts when under load. When Idle it will sit somewhere around 200 Watts.

A normal house fridge, only generally has to be able to ramp up to intermittent 25 Watt load.

It's a BIG jump.
You are aware that heat isn't measured in Watts... right? And that Watts used isn't a direct analogue to heat produced?

And that in your example, the refrigeration unit would only have to dissipate the 50W of heat, that built up, not 150W.
 
You are aware that heat isn't measured in Watts... right? And that Watts used isn't a direct analogue to heat produced?

And that in your example, the refrigeration unit would only have to dissipate the 50W of heat, that built up, not 150W.
Watts is a used measurement for Energy, Heat = energy. In other words, Joules per second of energy.

A modern desktop CPU produces around 5-150 Watts of waste heat, depending on load.

And in the example, the fridge actually does have to dissipate the excess heat. This is because of the Noble Gas Law.

If you didn't try and cool the oil down with a refridgeration unit, the fridge system actually still produces waste heat on it's own just running. The 50 Watts of cooling.

Then you introduce it to the oil and the cold fluid absorbs the heat from the oil. So when the working fluid of the fridge is compressed again, it heats up even more to account for all the energy it absorbed from the oil when it expanded and was pumped into the oil section of the loop.

So the fridge now has to dissipate the original 50 Watts of heat it generated in the first place, plus all the heat it absorbed from the oil.
 
Still here and watching, just been hit hard by work, even snagging 8 hours on Saturday.

Judging by what I see thus far, cooling like I was thinking really wouldn't be plausible. The oil idea looks interesting but is far outside of my skillset.
 
I really don't think liquid cooling is worth it. Look at this. It's very possible to get quiet air based CPU heatsink/fan combos that can keep a 130+W thing to under 40 degrees above ambient. So if you used that level of cooling and instead got a CPU spec'd at, say, 65W TDP, you shouldn't have any problems.
 
For the whole liquid-cooling thing, it all comes down to one simple question.

Do you intend to overclock the computer?

If no, then you should be fine with traditional air-cooling. Maybe throw on an extra fan or two if you're really worried.

If yes, then you should look into alternate cooling methods.
 
For the whole liquid-cooling thing, it all comes down to one simple question.

Do you intend to overclock the computer?

If no, then you should be fine with traditional air-cooling. Maybe throw on an extra fan or two if you're really worried.

If yes, then you should look into alternate cooling methods.
Hes also worried about the near 100deg temps affecting his system.

The ambient temps themselves wont really cause issues with a system unless you have something wrong with your airflow and are not overclocking.

Just something simple like a pair of intake fans, pair of exaust fans should do for keeping the case cool at stock speeds. A good heatsink for the CPU, something like a Noctua NH15 would be wonderful.
 
Hes also worried about the near 100deg temps affecting his system.

The ambient temps themselves wont really cause issues with a system unless you have something wrong with your airflow and are not overclocking.

Just something simple like a pair of intake fans, pair of exaust fans should do for keeping the case cool at stock speeds. A good heatsink for the CPU, something like a Noctua NH15 would be wonderful.
Hmmm... Well to quote the page about the thermal paste I linked to earlier:
It optimally performs between 122 degrees Fahrenheit and 482 degrees Fahrenheit.
So ambient temp would only speed up the time it takes to get to optimal operating temperatures.

And as to that Noctua fan... Never heard of it before, so I googled it...

It... it... it's... beautiful...

I'm tempted to get one just to see if I can get it to work in my Xbox, lol (I do a lot of modding on original Xboxes, and proper cooling does become an issue with some of the case mods).

Also with ambient temps like that, it might be worth it to get a small heatsink for your graphic's card's GPU (you can get small 30mm x 30mm x 10mm heatsinks with self-adhesive thermal tape [basically a form of thermal compound in tape form, rather than liquid, so you just peel the backing off, and stick it on the chip, and you're done] for $2-3; there are ones with small fans attached available for a bit more as well, but they don't always fit), assuming your graphics card doesn't already have it's own heatsinks and/or fans attached already.
 
Ok, I should really stop getting distracted by pointless stuff and actually respond to people answering my request.

That fan looks awesome, will it fit comfortably in the case you sent me? If not, we'll probably need a bigger case.

Haven't gotten around to actually looking at or assembling a potential system yet, this week was surprisingly hard and I've just been vegging in what little free time I am awake. Means I'll have plenty of money once I actually get around to it though.
 
Ok, I should really stop getting distracted by pointless stuff and actually respond to people answering my request.

That fan looks awesome, will it fit comfortably in the case you sent me? If not, we'll probably need a bigger case.

Haven't gotten around to actually looking at or assembling a potential system yet, this week was surprisingly hard and I've just been vegging in what little free time I am awake. Means I'll have plenty of money once I actually get around to it though.
Yeah the Noctua (So beautiful, so efficient) will fit in the Define S case no problem.

Just a note, on the build I came up with earlier, it doesn't include support for an optical drive. If you really want want it internally mounted, I would suggest a case like this.

http://www.newegg.com/Product/Product.aspx?Item=N82E16811352048

Which is from the same company and is practically identical to the other case, except is has 5.25" mounts behind the front door.

The really nice thing about the Fractal Design cases is that they have readily accessible built in fan filters for intakes. So you can keep a lot of dust out of your system by using them and just cleaning them off every once in awhile.
 
This is a good thing and is honestly a major selling point. I never remember to clean the dust out of my computer so keeping it from building up in the first place is the best, and honestly so obvious I am unsure why it isn't standardized, idea I have seen yet for just the case itself.
 
This is a good thing and is honestly a major selling point. I never remember to clean the dust out of my computer so keeping it from building up in the first place is the best, and honestly so obvious I am unsure why it isn't standardized, idea I have seen yet for just the case itself.
Some MFG's use magnetic fan filters but Fractal built them into little frames you can slide out or pop out, clean off, then stick em back in. A bit easier than dealing with a flexible and floppy magnetic filter.

But most cases with filters dont filter intake and exaust ports, assuming that with the system running, the air being pushed out from the exausts will keep dust out.

Here is a good review of the new case I linked,

 
Well... I definitely know what case I'm using for next desktop build. Very nice.


Oh one thing you definitely need to do. Add up the wattage of every single component in your computer; every HDD, every optical drive, every fan, the MB, your video card, absolutely everything that draws power, even if it's not always active (like optical drives).

Once you've got all that added up, then look at power supplies. you want a power supply that is at least 150W higher than your computer draws.

The thing is, power supplies are rated at their maximum output, not their maximum sustainable output. If your computer consistently draws 635W of power, and your power supply is rated at 650W, then after a year or so your power supply will start malfunctioning, causing power spikes and shortages. I've actually had 3 CPUs and 2 MB fried from this happening (with power supplies that tested perfectly fine with power supply testers, and multimeters, since the problems are intermittent, and usually only show up when the supply;s under strain). Whereas the same system drawing 635W of power supplied from a 800W power supply will have a perfectly functional power supply for it's full projected lifespan (with most power supplies that's 5-10 years).

Further, running a power supply lower than it's max will reduce the power supply's heat production, since the components won't be working as hard, hopefully extending the life of the power supply even further. This is true even of the better power supplies, that can sustain their listed wattage with no problems.

That said, you don't want a power supply with too high a wattage. Putting a 1200W power supply in that 635W system will actually decrease the life of the power supply some, though not as much as the 650W power supply. Don't go higher than 300W past your actual usage, so for that 635W system, the max would be a 950W power supply. This gives you a nice margin, as well as room to upgrade/add components.
 
Well... I definitely know what case I'm using for next desktop build. Very nice.


Oh one thing you definitely need to do. Add up the wattage of every single component in your computer; every HDD, every optical drive, every fan, the MB, your video card, absolutely everything that draws power, even if it's not always active (like optical drives).

Once you've got all that added up, then look at power supplies. you want a power supply that is at least 150W higher than your computer draws.

The thing is, power supplies are rated at their maximum output, not their maximum sustainable output. If your computer consistently draws 635W of power, and your power supply is rated at 650W, then after a year or so your power supply will start malfunctioning, causing power spikes and shortages. I've actually had 3 CPUs and 2 MB fried from this happening (with power supplies that tested perfectly fine with power supply testers, and multimeters, since the problems are intermittent, and usually only show up when the supply;s under strain). Whereas the same system drawing 635W of power supplied from a 800W power supply will have a perfectly functional power supply for it's full projected lifespan (with most power supplies that's 5-10 years).

Further, running a power supply lower than it's max will reduce the power supply's heat production, since the components won't be working as hard, hopefully extending the life of the power supply even further. This is true even of the better power supplies, that can sustain their listed wattage with no problems.

That said, you don't want a power supply with too high a wattage. Putting a 1200W power supply in that 635W system will actually decrease the life of the power supply some, though not as much as the 650W power supply. Don't go higher than 300W past your actual usage, so for that 635W system, the max would be a 950W power supply. This gives you a nice margin, as well as room to upgrade/add components.
Already factored that into the build I posted earlier.

It has a max draw at stock loads of 462 Watts. With a minimum draw of 113 Watts.

The power supply selected can go up to 750 Watts with 80+ Gold efficiency. Giving him plenty of headroom to add extra components like Optical drives, a Second Gfx card, or a butload of HDD's.
 
What is the point of a second graphics card? Its it only really useful if you use multiple monitors or are you able to chain them together for distributed processing?
 
You can chain them together IIRC.
This.

Nvidia uses a technique called SLI to do it, whereas AMD uses a technique they call Crossfire.

You can nominally chain up to four graphics cards together.

You get a BIG boost in overall frame performance with a second card added, and smaller boosts with a three or four card configuration.

In some cases, two cards that total up to the price of a high end flagship, EXCEED the flagships overall performance.

So it's a nice option to have available to at least toss a second card in.

With DX12, the feature will be a lot more emphasized thanks to Split Frame Rendering and enhanced Multi-GPU rendering features, allowing any and every Graphics hardware in your system to contribute.
 
Hmmm....how difficult is this to do? Just flip a few settings in bios or the appropriate driver software or do I have to fiddle with connections and those stupid jumper settings?
 

Users who are viewing this thread

Back
Top