Announcement

Collapse
No announcement yet.

Nearly 26,000 kWh Was Used On Linux Benchmarking In 2016

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    Originally posted by Michael View Post

    Unfortunately rather a big upfront investment of 50+ water blocks and all the time involved in setting it up, plus the huge hassle of needing to take down the entire system and power off everything when it comes to pulling a system out of the rack for any maintenance or changing components.
    Yeah I suppose that would be less than ideal. What kind of heating do you have in your building? Forced air? Electric resistance heating? Hot water baseboard?

    You could get a small packaged air conditioner for a few hundred dollars and install it in that room. If it's a window unit, you'll have to cut a hole in the wall (where it will reject heat to an adjacent room). If it has a ducted exhaust (better), you could install a small round duct or reuse existing ducts to send hot exhaust air to another part of the house. I spotted a previous article about the duct silencer, so it looks like you have a good exhaust duct candidate already installed. Could you reverse the airflow through the duct to exhaust heated (from a packaged AC unit) back up to your front office? It might be a very easy and convenient solution that doesn't require knocking holes in your walls. Then you could regulate the temperature in your server room and the AC unit would only use electricity when it needs to move a bit of heat. That may end up actually being more energy efficient than a ventilation fan.

    A few years ago I was going to build my own refrigerator. Without really analysing the situation I figured that dumping heat outside during cold Canadian winters would be like free refrigeration. Like an idiot I bought a bunch of copper pipe and some other supplies, then decided to actually look at the situation more closely. When I ran some numbers and actually looked at the energy balance (I'm an engineer) it turns out that it's WAY more cost effective to run the compressor in the fridge to pump that heat back into the living space. It's just one of those things that at first glace seems to make sense, but when you dig into it more deeply a very different situation is presented. I ended up building my own super efficient fridge anyway, but that's another story.

    26000 kWh is about 71.2 kWh per day, which is about 3 kW of heat rejection into that room. 3 kW works out to be about 10000 BTU in air conditioning freedom units, so you'll probably want to add a bit extra to that, say 12000 - 15000 BTU? This little critter would do the trick, and it would probably even bolt up to your existing duct without much trouble. https://www.danby.com/products/porta...rs/dpa140b1wb/ Really any AC unit would work, I just picked one from the website of a very common big box store. I'm also having a hard time wraping my head around how that little 530 cfm duct and fan is maintaining the temperature in that room. Really that shouldn't be anywhere near enough airflow to maintain a temperature less than sauna with 3 kW of heat input. Is there other heat leakage out of that room somehow?

    Comment


    • #22
      Originally posted by CMDRZOD View Post

      Thanks for the compliment, I'm also a long time lurker and just began posting today for this article. When I had a large amount of PC's in my home, and being an electrician, running 240v to them is the first thing I did to try to reduce power use as much as possible. As another poster said, every watt saved is less cooling and AC cost.

      As far as watercooling PC's goes, I have a custom water loop with 3x GTX980Ti and 5960x processor at 4.5Ghz. I would never recommend running normal water through these copper cooling blocks. The water at minimum needs to be ran through a reverse osmosis. There's just too much particulates and sediments in tap water to ever run through a PC cooling loop.
      There is a solution to this, rather like cooling the engine in a boat used in salt water: indirect water cooling. Run the distilled water in the cooling loops through a water-cooled heat exchanger, which in turn is cooled by the water being warmed. A water tank with a radiator dunked in it or just a coil of tubing will work.

      Comment


      • #23
        In the UK, that's about £2600, or $3200

        Comment


        • #24
          Did anyone mention a solar roof? This is the only final solution. I have one, it helps alot.
          ## VGA ##
          AMD: X1950XTX, HD3870, HD5870
          Intel: GMA45, HD3000 (Core i5 2500K)

          Comment


          • #25
            Originally posted by darkbasic View Post
            Did anyone mention a solar roof? This is the only final solution. I have one, it helps alot.
            I have solar PV as well as a large evacuated tube array for hot water. A solar PV installation capable of driving 26000 kWh/year is a serious, serious piece of kit. That's also assume grid-tie. If you wanted something like batteries then you're talking an undertaking on the scale of the pyramids.

            Comment


            • #26
              Originally posted by Michael View Post
              for just a 1~5% increase in efficiency, not yet worth calling an electrician out (I prefer not opening up the breaker box nor experienced with 240V requirements) until I have an electrician out here for other work in the future.
              Just out of curiosity, have you done any calculations about the savings in $$$ for each suggested improvement? 1-5% sounds reasonable if you use power in a multi megawatt range. Also good power supplies can be swapped from old to new system so you don't necessarily need tons of them every now and then. I've usually saved money by buying good quality PSUs. Many bad PSUs fail in 1-3 years. Of course there's some warranty. But good PSUs can last 10+ years like my old Fortrons.

              Not mentioned here, but you could possibly also save by trimming the software stack or using network booting via PXE. Maybe not anymore with all the SSDs around, but diskless servers work just fine. One thing I've been wondering is the redundancy in the test suite. How much time is spent redoing the same steps? Is it all optimal? No redundancy at all? All previous results are cached and stored permanently?

              Comment


              • #27
                Originally posted by monkeybutt View Post
                A solar PV installation capable of driving 26000 kWh/year is a serious, serious piece of kit.
                That’s averaging 3kW.

                Comment


                • #28
                  Originally posted by ldo17 View Post

                  That’s averaging 3kW.
                  Yes I'm aware. I mentioned that in the above cooling estimate. But the sun doesn't shine at night, and to offset a 3 kW continuous load it does not take 3 kW of solar. It takes a heck of a lot more.

                  If you look up the insolation tables, Indiana gets an average of about 3 hours a day of sun in the winter, and about 6 hours a day in the summer. From the table it looks like Indiana is actually less than ideal for solar in general. So lets take summer as an example. Lets say you have a grid-tie system with 80% end-to-end efficiency (that's about what I'm getting on my array). All of a sudden it now takes 30 kW of panels just to make up the load. In the winter, assuming you're perfect about cleaning your panels and don't miss a single day of sun, that's a 60 kW array that you'd need.

                  Now lets look at some ballpark dollar figures. To offset this substantial load and make 26000 kWh in a year, you'll probably be able to split the difference and install around 45 panels, assuming you're okay with using the electrical grid as a sink/source and don't want to do something crazy like batteries. Lets say you're using 100 watt panels. That's 45 panels 100 watt panels at about $250 each, with a grid-tie string inverter for each pair of panels (23 each) and all the associated wiring, permits, disconnects, agreement with the utility (you need to register with the utility and provide an outdoor lockable disconnect) and labour to install. You're probably looking at about $25000 - $30000 to install. This is also assuming you have a 200 amp dedicated service to your house, which is the absolute bare minimum to back-feed 45 kW into the grid.

                  45 kW of panels will also require 250 m^2 of South facing roof space. That's a lot. Really a lot. That's like 16m by 16m (or 52 ft by 52 ft) which is an area that most people do not have. I have a medium sized house and I don't have anywhere near 52x52 ft of South facing roof. Not even close. Actually my whole house could very comfortably fit inside a 52x52 ft footprint. If you owned a supermarket or a warehouse you could probably put up that much solar, but I think it's safe to say that the majority of homeowners don't even have the roof real estate to physically install that much solar.

                  Of course you could always do what I did, which is throw away the idea of COMPLETELY offsetting your usage, and just install a smaller system. Even a smaller system will put a dent in your power bill. But don't think it's going to make 26000 kWh per year go away. Also don't ever think that they'll pay for themselves. Barring government incentives, the payback period for solar PV is still piss pour. Like 15 years before it even begins to pay for itself, assuming no repairs during that time. Unfortunately your solar array may not live long enough to ever pay for itself.

                  AND! Also don't do what I did, which is install your solar array without thinking about the roof that's under it. My solar array has been installed since about 2010, and now the roof is ripe and needs to be replaced. I have to tear the whole thing up. So basically I'm screwed. If you're putting up solar you need to make sure you've got a good 50 year roof on your house (metal - not shingles). You don't want to put a solar array on top of a roof that has 20 year shingles, and the shingles are already 12 years into their usable lifespan. Which is basically what I did. Doh.

                  So if you're rich and live in a gigantic mansion and want to put up 45 kW of solar, and can afford $25000 for a monster solar array as well as very likely another $25000 for a new roof under the thing, then yes, this project is feasible. But for most people it is not. I mean hell, I have three Porsches and a private lake and I can't pull off a solar install of this magnitude. For the vast majority of people it would be well beyond the realm of possibility.

                  Comment


                  • #29
                    I just re-read my post from above, and I think I came off a little too negative toward solar. Don't get me wrong, I love solar. I have a decent amount of solar capacity installed, and I have a car that I already have covered in solar panels that I'm (slowly) converting to electric. Solar is awesome, it's the future, and I will continue to install panels all over the place and promote the technology when speaking to people. But I think we also have to be realistic about the capabilities and limitations. If you think purchasing a couple panels and putting them on your roof is going to power your house, you will be disappointed.

                    PS Don't poo-poo the solar panels on my car. Yes I understand the output of the panels. Yes I understand how much power it takes to move a car. I've had this argument a dozen times. The car SITS stationary all day. If it can sit for 8 hours between drives, you can add legitimate mileage during that time because the car is CHARGING when it's OFF. Sorry for my preemptive defensive rant.

                    Comment

                    Working...
                    X