Using a PC as a heater

[samcool55]

If you want to use your pc as a space heater, let it at least do something useful.

Like F&H or something

It would be sad that a lot of the power would be wasted by a random benchmark while you can do something useful with it.

[Litargirio]

I have been running F@H 24/7 since Saturday, and it is keeping my room very warm, to the point where I open the windows every now and then to cool it down, even though it's freezing cold outside. 

 

6 hours ago, djdwosk97 said:

That's so much work to set up.

It isn't. All you have to do is run the installer, and it works. No further effort required.

[porina]

I run my own little home processing farm to keep my house, not just a room warm in winter. It is equally efficient to a simple heater, and both are inferior to air source heat pumps (air conditioner in reverse) which is about the only way to do much better.

 

On the positive side, it does actual work in producing that heat. I run assorted distributed computing projects. 

 

I have found some downsides, apart from the obvious upfront hardware cost and space required. One is that I don't have a way to regulate air temperature. As outdoor temperature varies, I have to manually adjust how much work is going on so that it doesn't get too hot. I run CPU only as standard, and turn on GPUs when a boost is needed. The other, I found out last night, for some unknown reason my internet connection went down. During that time the computers could get no work, so were idle and not producing much heat.

[Velvet Revolver]

9 hours ago, djdwosk97 said:

The heat in my apartment just stopped working and so I started to wonder.....how efficient is a computer at converting electricity into heat? Is it less efficient than a traditional electric space heater? Would 300w of computer put out the same amount of heat as a 300w heater (I know space heaters are generally >1 kw)?

 

also, how well would 300w heat a room that's 10'x10'x8' with no windows?

Wow you got 16k posts and you wonder if a computer can be used as a space heater and have the same efficiency. Is this a joke?

[79wjd]

5 hours ago, Velvet Revolver said:

Wow you got 16k posts and you wonder if a computer can be used as a space heater and have the same efficiency. Is this a joke?

Why is it a joke......

[Mooshi]

Grab an old Pentium 4 if you wanna set your room on fire. 

[ProjectBox153]

Just get a Pentium 4 Prescott and you'll be boiling in an instant!

[KeltonDSMer]

12 hours ago, manikyath said:

you're actually quite wrong there.

 

energy cannot "disappear" into nothing, it always has to go into something, while the power is used to "power the computer", in the case of a computer, "powering it" means that it's driving the billions of transistors, which all.. kick out the energy that went into them as heat.

 

essentially, aside from heat, this is where the energy in your computer can go to:

- making things move (fans, hard drive)

- lights

- I/O signals (ethernet, audio, etc. all use power as well, but in the end they honestly end up as either movement, lights, or heat as well.)

 

energy cannot be lost, and if there's nothing else for it to go into, it'll go into heat.

 

This is correct, every joule pulled from the wall for a PC (or any electrical device really) is converted to heat. Even the energy for the fans/lights is converted to heat; the air from a fan slows down immediately and the friction that slows the air down is now...heat!

 

7 hours ago, porina said:

I run my own little home processing farm to keep my house, not just a room warm in winter. It is equally efficient to a simple heater, and both are inferior to air source heat pumps (air conditioner in reverse) which is about the only way to do much better.

 

 

Yep, if the goal is to heat/cool a specific area, moving energy with a heat pump could be considered more "efficient" compared to a resistor.

 

 

@djdwosk97 , using your PC to warm your room is exactly the same efficiency as a conventional space heater (100% electrical->thermal conversion).

[corrado33]

7 hours ago, porina said:

I run my own little home processing farm to keep my house, not just a room warm in winter. It is equally efficient to a simple heater, and both are inferior to air source heat pumps (air conditioner in reverse) which is about the only way to do much better.

 

On the positive side, it does actual work in producing that heat. I run assorted distributed computing projects. 

 

I have found some downsides, apart from the obvious upfront hardware cost and space required. One is that I don't have a way to regulate air temperature. As outdoor temperature varies, I have to manually adjust how much work is going on so that it doesn't get too hot. I run CPU only as standard, and turn on GPUs when a boost is needed. The other, I found out last night, for some unknown reason my internet connection went down. During that time the computers could get no work, so were idle and not producing much heat.

You SHOULD be able to do this relatively easily. 

 

Something like this usb thermometer from newegg.

 

http://www.newegg.com/Product/Product.aspx?Item=9SIA6703MH1334&ignorebbr=1&nm_mc=KNC-GoogleMKP-PC&cm_mmc=KNC-GoogleMKP-PC-_-pla-_-Cell+Phone+–+USB+Chargers-_-9SIA6703MH1334&gclid=CNmcpKiD5dACFUVafgodXKAIKQ&gclsrc=aw.ds

 

And some nifty scripting in autohotkey (hit the "stop processing" key on a few of the computers when the temp gets too hot.)

 

Honestly, it'd take me a few minutes to set it up probably. The "hardest" thing would be to use 1 usb thermometer to control many different computers. But you could probably do it over e-mail or unix mail. You can easily run scripts off of e-mails with most mail programs (thunderbird, outlook, etc.) Or, I'm not sure if F@H lets you set up your own "distributed computing" node, where you have one pc control what jobs get sent to the others. If that were the case, it'd be easier, just stop sending jobs out when the room gets too hot!

 

 

Alternatively, you could have all of the fans output to a common pipe, That pipe would usually vent to the room, but when it gets hot enough have  a small door (which leads to the outside) redirect the flow to the outside instead of inside. That'd be a bit more complicated, having to control motors and what not. Still not hard though. Ventilation stuff is easily found at home depot.

[manikyath]

47 minutes ago, KeltonDSMer said:

 

@djdwosk97 , using your PC to warm your room is exactly the same efficiency as a conventional space heater (100% electrical->thermal conversion).

in a sense, using a computer is more efficient, because in transitioning electricity to heat you're also doing something useful

[porina]

26 minutes ago, corrado33 said:

You SHOULD be able to do this relatively easily. 

I'm not sure your definition of easy is same as my definition  

26 minutes ago, corrado33 said:

Honestly, it'd take me a few minutes to set it up probably. The "hardest" thing would be to use 1 usb thermometer to control many different computers. But you could probably do it over e-mail or unix mail. You can easily run scripts off of e-mails with most mail programs (thunderbird, outlook, etc.) Or, I'm not sure if F@H lets you set up your own "distributed computing" node, where you have one pc control what jobs get sent to the others. If that were the case, it'd be easier, just stop sending jobs out when the room gets too hot!

I'm actually using BOINC on Windows. I believe BOINC clients allows a degree of remote connectivity and control, but I never looked in detail to how this works.

 

There could be two levels to this, one is "no new work" which does what it says. Once existing work is done, it doesn't get more until it is enabled again. Another signal is to suspend immediately. These could be used to set different thresholds as there will obviously be some lag depending on work unit length if NNW is set. 

26 minutes ago, corrado33 said:

Alternatively, you could have all of the fans output to a common pipe, That pipe would usually vent to the room, but when it gets hot enough have  a small door (which leads to the outside) redirect the flow to the outside instead of inside. That'd be a bit more complicated, having to control motors and what not. Still not hard though. Ventilation stuff is easily found at home depot.

Not practical at all... also not ideal  as I would only want to work like a normal heating system and to only heat as required, and not regulate by dumping excess.

[corrado33]

11 minutes ago, porina said:

I'm not sure your definition of easy is same as my definition  

I'm actually using BOINC on Windows. I believe BOINC clients allows a degree of remote connectivity and control, but I never looked in detail to how this works.

 -snip-

I'd be interested to see if you actually try it out. A few hours of tinkering may net you a fully working thermostat in the room! 

 

Apparently it's not "difficult" to control a bunch of boinc clients remotely. You just need to install a slightly different program and do a bit of setup. 

 

http://boinc.berkeley.edu/wiki/Controlling_BOINC_remotely

 

I dunno, seems like a fun project if you had the time.

 

(Looks like the program you'd need is BOINC Tasks, linked below.)

 

http://boinc.berkeley.edu/addon_item.php?platform=win&item=http%3A%2F%2Fwww.efmer.eu%2Fboinc%2Fboinc_tasks%2Fdownload.html

 

 

If you hadn't noticed, I love doing home automation stuff.  Especially hacked together home automation stuff. 

 

 

EDIT: I mean heck, you may not even need an external thermometer/thermister. You may be able to correlate one of the computer's motherboard temp sensors with the room temp (it'd be slightly warmer obviously) and use one of those instead...

[corrado33]

20 minutes ago, porina said:

Not practical at all... also not ideal  as I would only want to work like a normal heating system and to only heat as required, and not regulate by dumping excess.

Sorry for the double quote. I guess it depends on what you value more, the boinc points, or your electricity bill.  (The vent to the outside would allow you to run boinc 24/7 and never worry about the temp, netting you the most amount of points... credit?? whatever it's called on boinc.)

 

The other way works more like a normal heating system, like you said.  

 

 

God dangit. Now you've gotten me interested in folding/seti again...gah

[PianoPlayer88Key]

So what happens if you use a system with dual (or quad) Xeons, OC them each to match an FX-9590's TDP (using bclk if necessary), and run Prime95 small FFTs?  As for the GPUs, what about two R9 390X2s in crossfire, running Furmark & each overclocked to match the performance of Titan XP SLI?  Would that heat a typical small to medium room much or at all?

 

Otoh, would the thermal throttle point on various parts (like 100°C on my 4790K & 6700K) limit how much heat they could put out?  Also what difference would air vs watercooling the parts make for heat output into the room?

[J. Hammond]

22 minutes ago, PianoPlayer88Key said:

Would that heat a typical small to medium room much or at all?

 

 Also what difference would air vs watercooling the parts make for heat output into the room?

The quick answer isn't if it will heat a room up, it's whether it will heat a room within a time that you would consider convenient. While it may take half an hour to warm up a medium sized room with a heater, it may take hours with a PC, even though in the end the PC is much more efficient.

 

There is no difference between watercooling and air cooling in the amount of heat that will be eventually be released into the room.

[porina]

10 hours ago, PianoPlayer88Key said:

So what happens if you use a system with dual (or quad) Xeons, OC them each to match an FX-9590's TDP (using bclk if necessary), and run Prime95 small FFTs?  As for the GPUs, what about two R9 390X2s in crossfire, running Furmark & each overclocked to match the performance of Titan XP SLI?  Would that heat a typical small to medium room much or at all?

That is kinda what I do, except I don't run stress loads but distributed computing projects so that work goes into something. If you go to PrimeGrid on BOINC and do the SGS/PPSE/PPS projects, they stress the CPU just like Prime95 small FFT does while giving you a chance to find a Top 5000 size prime number. For GPU it wont quite be furmark levels but there are various projects you can do too. For how much heating, that depends. For practical purposes power taken at wall = power heating your room. A high end CPU let's say is 100W ball park. GPUs vary a lot but the thirstier ones are typically up to 250W each. An actual heater would be 1000W+.

10 hours ago, PianoPlayer88Key said:

Otoh, would the thermal throttle point on various parts (like 100°C on my 4790K & 6700K) limit how much heat they could put out?

At the throttle point they will reduce their power to help keep temps under control. Less power = less heating.

9 hours ago, J. Hammond said:

The quick answer isn't if it will heat a room up, it's whether it will heat a room within a time that you would consider convenient. While it may take half an hour to warm up a medium sized room with a heater, it may take hours with a PC, even though in the end the PC is much more efficient.

Unless you go crazy like me and put PCs everywhere... I'm still not able to run all my CPUs without my house getting too hot. Definitely not cold enough to fire up GPUs yet. A single PC would contribute, but multiples quickly escalate.

[PianoPlayer88Key]

58 minutes ago, porina said:

That is kinda what I do, except I don't run stress loads but distributed computing projects so that work goes into something. If you go to PrimeGrid on BOINC and do the SGS/PPSE/PPS projects, they stress the CPU just like Prime95 small FFT does while giving you a chance to find a Top 5000 size prime number. For GPU it wont quite be furmark levels but there are various projects you can do too. For how much heating, that depends. For practical purposes power taken at wall = power heating your room. A high end CPU let's say is 100W ball park. GPUs vary a lot but the thirstier ones are typically up to 250W each. An actual heater would be 1000W+.

Ahh, so I'm guessing you might need higher wattage parts?  Or just put higher wattage through them with a better cooling solution, or what?  For example, I'd expect a Celeron G3900T wouldn't put out much heat at all, but an i7-6700K would put out some heat, while an FX-9590 would put out more.  Then, what happens if you take that i7, or that FX, delid & repaste with CLU, slap a custom watercooling loop with a 480mm or 560mm radiator on it with it exhausting into the room, and overclock to the point where it thermal throttles?  Would the cooler be dissipating more heat to keep the CPU temps in check, or how does that work? 

Quote

At the throttle point they will reduce their power to help keep temps under control. Less power = less heating.

Ahh that's what I was figuring.  What about putting a better cooler on, like I was saying above?  Would that have it exhaust more heat?

Quote

Unless you go crazy like me and put PCs everywhere... I'm still not able to run all my CPUs without my house getting too hot. Definitely not cold enough to fire up GPUs yet. A single PC would contribute, but multiples quickly escalate.

Ahh, interesting.

 

So what about if you have the opposite issue?

 

• Let's say you're living in a place where temps regularly hit 100-110°F (and rarely flirt with 120°F) outdoors in summer, you don't have A/C, and 80-85°F is considered somewhat "cool" in the house, with 100°F indoors being seen once in a long while.  (This is assuming the computer is off.)  Or maybe your climate is even hotter.
• Also idk how much difference it makes for the PCs & keeping temps under control, but let's assume humidity is < 20%, like a desert.
• You're playing a a demanding game, like GTA V with everything maxed out (including going crazy with scaling), Rise of the Tomb Raider with HD textures & everything maxed, etc. (For example, my GTX 1060 3GB gets 0.3fps at 1080p with everything maxed at the beginning of the GTA V benchmark.  My iGPU (HD4600) gets 1.5-1.8fps.)
• You're using as many 5K (5120x2880) 60Hz monitors as your 2 (or 4 if they support 4-way SLI/CF) GPUs support (I generally see 3 displayports per card on most gaming GPUs, I've seen 4 on some workstation cards, & the AMD FIrePro W9100 has 6 mini displayports per card & supports 4-way crossfire, which adds up to 24 monitors).  For a 2-way GTX 1080 or Titan XP setup, that'd be 6 5K monitors.  Of course the game is full screen across all of them.
• You're also rendering a HDR video to 4K H.264.
• Assume your GPU & CPU are fast enough so the game never dips below 2x your monitor's refresh rate even in the most intense/demanding scenes, and the render is 4x the original camera framerate.  (Let's say 120fps on both for the sake of argument.)
• If no hardware today will get those FPS in those particular games at that resolution & settings, then if you wait until such hardware exists, you use whatever is the most demanding game released at that time.  Same goes for the render situation - if 8K or higher is a thing, you do that instead of 4K.
• If you have multiple PCs, each one is running the entire above described workload on its own, or more.

 

How would you keep your temps in check?

 

[porina]

2 hours ago, PianoPlayer88Key said:

Ahh, so I'm guessing you might need higher wattage parts?

Nope, I still aim for the most power efficient parts, as I want the most work done for the least power. When the weather is cool, the heating effect is a side benefit.

Quote

So what about if you have the opposite issue?

How would you keep your temps in check?

You're over thinking it. Apart from myself, I don't really expect anyone to do this as an overall heating solution  At the moment I manually adjust it according to the weather.

 

Actually, there was one company doing it commercially. I can't remember the name, but I think they called them computational furnaces. They give you the devices which are essentially sealed computer units in a radiator like form factor, and they pay for the electricity for them to do work. You get the heat produced for free. They benefit by selling the compute resource and not having to pay for cooling that you would in a traditional server environment.

 

Edit: oh, they're data furnaces. example at following link, not necessarily the same as the one I saw before: http://arstechnica.co.uk/business/2015/05/data-furnaces-arrive-in-europe-free-heating-if-you-have-fibre-internet/

 

[corrado33]

Have you thought about controlling the temp yet so you don't have to do it manually? 

 

(PS: I've since setup my own little distributed computing farm with a few PCs at work and my own two PCs here.)

 

PPS: Cosmology@home doesn't run on Macs apparently (due to needing virtualbox.) Although it runs on windows just fine without virtualbox..

 

Oh and to answer a few of the questions above.

 

Using a better cooler would simply allow you to do more work while keeping the temps (of the pc components) lower. So yes, it'd heat your room a bit better. 

[Lord Nicoll]

On 12/8/2016 at 1:00 AM, djdwosk97 said:

The heat in my apartment just stopped working and so I started to wonder.....how efficient is a computer at converting electricity into heat? Is it less efficient than a traditional electric space heater? Would 300w of computer put out the same amount of heat as a 300w heater (I know space heaters are generally >1 kw)?

 

also, how well would 300w heat a room that's 10'x10'x8' with no windows?

I use my rack servers to heat the main interior of my house, on average 10°C-15°C warmer than the room, or in this case hall/landing, not heated or adjacent. 

[flibberdipper]

I'm just chirping in again to give some moderately real-world results.

 

I was playing DOOM on my rig (literally 100% utilization on my 7950 and about 90% on my Pentium), and my rig managed to bring the temp of my room (11x11x9 or something weird like that with two exterior walls and a crappy window) from 70F to 74F in about an hour and a half. Your mileage may vary, but my juice sipper sure seems to work better than a space heater.

[AshleyAshes]

What's Wrong With This Thread: It's not about getting your landlord to fix the broken heat in December, which your landlord would be legally required to do with reasonable expediency in most Canadian provinces and US States.  If the heat in my apartment broke I could legally withhold rent payments until it was fixed.

[imPixelTV]

pentium 4 + 7970

[MANIDY inc]

I have a 390 that runs 85-89c in watch dogs (only game its run above 72c in) and my room is still freezing AF 

[Me1z]

Electrical draw results in a 1;1 production of heat. If your system is pulling 300w from the wall it will exhaust 300w of heat, the same as a 300w heater. Having a more 'efficient' system simply means that you are able to increase productivity while outputting the same amount of heat. An efficient system can't magically consume 300w and not produce an equivalent amount of heat. It has nothing to do with product design or electrical resistance, heat is directly proportional to energy usage regardless of the device using it. That's why a 60w light bulb gets just as hot as a CPU with a 60w TDP. There's another good example; TDP, or Thermal Design Power. It is both an expression of heat output and electrical draw. Its kinda funny, its almost like the two are directly related or something.

 

TLDR: Using a PC as a heater is a great idea. This is coming from someone who is actively heating their basement with a folding/boinc rig. Why spend money powering a simple space heater when you could produce the same heat for the same cost while also curing diseases?