Aces High Bulletin Board
General Forums => Hardware and Software => Topic started by: TDeacon on January 28, 2017, 05:46:16 PM
-
Comments?
My planned first PC build: (https://pcpartpicker.com/user/markhinds/saved/x3tQ7P). Intend to order parts in next few days, except perhaps for the monitor, in which case I'd use my current 60Hz one.
Notes:
1. Intended for immediate general use, and gaming the MMO WWII flight simulator “Aces High”.
2. Intended for future use when (if) the MMO game Star Citizen is released (perhaps in 2-3 years, with with graphics card update). That's the reason for the 1000W power supply.
3. Won't over-clock initially. If I do later on, will explore getting a better CPU cooler.
4. Will be a dual Windows / Linux installation, with each OS on a separate hard drive, and GRUB boot loader to choose between them.
5. Willing to accept slower load times, so no SSDs initially.
6. External ODD needed only temporarily, per ASUS motherboard instructions. Apparently needed so my SP1 Windows 7 installation CD can access up-to-date USB 3.0 drivers.
7. Although most on-line sources say fast DDR4 RAM has no practical gaming effect, bought it anyway. At least it is CL-14, and one source claimed significance for the arithmetic product of the clock period and the CAS latency, which in this case is 4.4 nanoseconds.
8. I already own an unused OEM Windows 7 license, and so will be postponing my use of Windows 10 until at least January of 2020 when extended support ends.
-
Looks pretty good.
I would seriously consider replacing one of the 1TB HDDs with an SSD,
-
Looks like a nice build, and should more than max out AH3 on your 1080p monitor. My approach would be (and was) save money on CPU, ram, and possibly vid card, unless you're thinking of going to a 4k monitor or VR.... and add the SSD as Iron suggested, then upgrade to more power for less $ in a couple years or so when your new more-demanding game comes out....but that's me. When I'm home I often turn my puter on and off many times during the day for different reasons, work, check email, weather, Aces High, whatever, and love the very fast load time the SSD gives me. My puter runs AH3 very well graphics close to maxed out, on a I3 6100, 8GB and a 3 GB 1060 on a 1440p monitor....I'd rather use the saved $ for something like scuba diving.....Whatever you choose, I also suggest taking your time and enjoy the building process, good luck!
You may want to check this out, appears to be a newer version of the optical drive you have listed: https://www.amazon.com/Samsung-DVD-Writer-Internal-SH-224FB-BSBE/dp/B00WAG0WIM/ref=dp_ob_title_ce
-
Looks like a killer with some overkill...
I agree with the slower loading times not being an issue, I've never been able to store the saved seconds for future use anyway. That said, there's been situations when I would have liked my system to boot a little faster like when someone calls me asking for advice right when I'm about to leave the house.
As for the overkill, a 1000W power supply might be way too much unless you're planning to SLI several video cards together. The power consumption of video cards has decreased hugely in the last few years while the efficiency rate of power supplies has increased. Your plan says 400W, so some 650 W would have a decent amount of overhead. If your PSU is too powerful, the excess will cause issues.
-
Thanks guys. I am now considering a 258 GB M.2 SSD for the 2 OSs, and have modified the parts list. I would need 1 Linux partition on the SSD, plus however many Win7 requires. The remaining partitions could be on a single HDD, comprising Windows data, Linux data, and "linux-swap". The latter wouldn't be needed much with 16 GB RAM, so I could set "vm.swappiness=1".
(test removed)
As for the overkill, a 1000W power supply might be way too much unless you're planning to SLI several video cards together. The power consumption of video cards has decreased hugely in the last few years while the efficiency rate of power supplies has increased. Your plan says 400W, so some 650 W would have a decent amount of overhead. If your PSU is too powerful, the excess will cause issues.
Could you please elaborate on the "issues" you mention above? Only disadvantage I was aware of is that I might be operating below the 1000W unit's peak efficiency during "normal" (non-gaming) computer use. I will look to see if a power efficiency curve is published somewhere. Ahha; here's one (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story4&reid=494). Note that the efficiency is ~90% over the entire range, or at least down to 100W.
Remember that I plan to update the graphics card when Star Citizen is released, in 2 or 3 years, and that I don't want to be constrained by the PS. The Developers claim that their intention is to push the limits of available hardware, although I assume it will also run at lesser graphics settings.
Mark H.
-
Thanks guys. I am now considering a 258 GB M.2 SSD for the 2 OSs, and have modified the parts list. I would need 1 Linux partition on the SSD, plus however many Win7 requires. The remaining partitions could be on a single HDD, comprising Windows data, Linux data, and "linux-swap". The latter wouldn't be needed much with 16 GB RAM, so I could set "vm.swappiness=1".
Could you please elaborate on the "issues" you mention above? Only disadvantage I was aware of is that I might be operating below the 1000W unit's peak efficiency during "normal" (non-gaming) computer use. I will look to see if a power efficiency curve is published somewhere. Ahha; here's one (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story4&reid=494). Note that the efficiency is ~90% over the entire range, or at least down to 100W.
Remember that I plan to update the graphics card when Star Citizen is released, in 2 or 3 years, and that I don't want to be constrained by the PS. The Developers claim that their intention is to push the limits of available hardware, although I assume it will also run at lesser graphics settings.
Mark H.
You do not need a 1000W PSU. Tri SLI? Yes. Two card sin SLI? Even that would be too much. But you should need a 750W max. I have an EVGA SuperNova G2 750 myself. Seasonic is excellent as well.
-
TDeacon, I'm not an engineer so this is just what I've read and tried to understand. For me it makes sense by thinking about The law of conservation of energy. If you only need 400W, where will the excess go? It doesn't sound to be a viable way to warm up your house, and heat is anyway an enemy to capacitors and other electrical components. A system constantly being 10c warmer than it could be will fail in half the time.
More technically speaking, the excess can/will cause instability in the current, ripple etc.
Skuzzy knows more about electrickery.
-
Switching power supplies (every computer power supply on the planet) work best when they are under 80 to 85% load. This load is a magic number which keeps the supply electronics working at peak efficiency. The supply runs cooler and produces less noise in the current when running at that load.
It is the very nature of how a switching supply works. Too small of a supply is bad, too big of a supply is bad as both introduce noise and generate more heat.
The closer you can get to an 80 to 85% load, the longer the supply will last and the power quality it will produce will be much better as well. Noise in the power will cause every DC motor (fans, HD's,...) to run hotter and louder. Noise also taxes the power regulators for the CPU.
In short, it is not a good thing to run a switching supply too far outside its 80 to 85% load.
-
Well, guys, I gave up on SSDs for my initial build, as the (very new) one I wanted (http://www.tomshardware.com/reviews/mydigitalssd-bpx-nvme-ssd,4780.html) doesn't seem to have drivers for Win7 or Linux yet, if ever. Other brands are older or out of stock. People on Amazon also had issues getting Win7 to boot from it when they used another SSD's drivers. So I will wait on this, and use a single 1TB hard drive. Anyone know how many disk partitions you can have when using Win7? I would need 3 for Linux Mint, and want the Win7 data separate from the OS, which totals 5 at least.
TDeacon, I'm not an engineer so this is just what I've read and tried to understand. For me it makes sense by thinking about The law of conservation of energy. If you only need 400W, where will the excess go? It doesn't sound to be a viable way to warm up your house, and heat is anyway an enemy to capacitors and other electrical components. A system constantly being 10c warmer than it could be will fail in half the time.
More technically speaking, the excess can/will cause instability in the current, ripple etc.
Skuzzy knows more about electrickery.
Bizman, the 1000W on the PS is just a maximum load rating, which you are not supposed to exceed. What power it actually uses (wall outlet), and delivers (PS output), is a function of the load you attach to it, which varies depending on hardware, which program you are running, and what you are doing at a particular moment in that program. Also, the PPP "400W" figure for my build is low for a worst-case figure, IMHO. For example, compare the ~150W figure they give for the graphics card with worst-case load graphs in various on-line reviews. AND, I still want leeway to put in better graphics cards in 3 years.
Switching power supplies (every computer power supply on the planet) work best when they are under 80 to 85% load. This load is a magic number which keeps the supply electronics working at peak efficiency. The supply runs cooler and produces less noise in the current when running at that load.
It is the very nature of how a switching supply works. Too small of a supply is bad, too big of a supply is bad as both introduce noise and generate more heat.
The closer you can get to an 80 to 85% load, the longer the supply will last and the power quality it will produce will be much better as well. Noise in the power will cause every DC motor (fans, HD's,...) to run hotter and louder. Noise also taxes the power regulators for the CPU.
In short, it is not a good thing to run a switching supply too far outside its 80 to 85% load.
Sounds good in theory, and I am no expert on switching power supplies, but look at the scope traces in the review: http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story4&reid=494 (bottom of page, corresponding the the table just above). Measured at (I assume) the PS outputs, the peak-to-peak AC ripple seems smaller on row 1 (100W) than row 4 (800W). And the power efficiency is almost flat across the range at ~ 90%. So unless I am misunderstanding you, both "efficiency" and "noise" to use your terminology wouldn't seem to be significantly worse at the low end.
And, even if the PS performance weren't this good, I can't see how anyone could keep their PC operating at 80-85% load. I would expect the load to vary hugely between, say, a word processor (low end) and gaming at max settings (high end).
-
In a previous life I helped design switching power supplies, along with other tidbits.
80 to 85% is the ideal and you are not going to run at that all the time., but you certainly do not want to run at 30% all the time either. If you run 75 to 90%, then that is fine.
Even in the site you linked it was clear the 75% load was running better than any other load. Pretty much what I expected, although the voltage regulation wandered more than would like to see.
Sun Flower makes that power supply and they make good supplies.
Is there a reason you think you need to use a 1000W supply? Are you thinking it will last longer than a 750W, or is there some other motivation?
-
(text removed)
Is there a reason you think you need to use a 1000W supply? Are you thinking it will last longer than a 750W, or is there some other motivation?
I have no idea what sort of graphics card(s) Star Citizen will need in 2-3 years when they finish it. I am hoping my other components will still be good, so it's mainly an anticipated unknown graphics card update. Otherwise, I would have bought a 750W or 850W at most. Ideally, I should wait and see, but I can no longer put up with my 15-year-old Dell running XP / Linux. To give you some perspective, that Dell still has its original 350W power supply, so I am hoping this one will last for awhile too.
MH
-
Still don't need 1000 watts of power haha.
-
On a budget I'd recommend 650W, bang for the buck and possibly expecting normal upgrades I'd take 750W, expecting to go SLI and possibly even tri-SLI I'd bump to 850W and all as high amperage single rail designs. No need for 1000W.
-
Thanks for the good responses so far.
On a budget I'd recommend 650W, bang for the buck and possibly expecting normal upgrades I'd take 750W, expecting to go SLI and possibly even tri-SLI I'd bump to 850W and all as high amperage single rail designs. No need for 1000W.
Reasonable, but again look at the scope traces from my link above. Since at 100W the efficiency is pretty much the same as at 800W, and the ripple is actually less, I am not seeing the evidence that my build will be injured with THIS power supply, if I initially run it at 20-50%. Here's a quote from Tom's Hardware: "Myth #1 is that using an oversized power supply will consume significantly more power at idle than a smaller unit since PSUs are less efficient at low loads and high loads, and are at their best between 40% load and 80% load".
And that's for now. My thinking was that what if 3 years from now I wanted to use (for example) 2 AMD Vega cards in Crossfire? I don't know what their TDPs would be. What if 6 years from now I wanted 2 of the following generation?
Not saying that my choice is for everybody, since at the moment it is overkill, but it still seems logical for my situation given that I want at least the case + power supply to be reasonably future proof. Remember that 15-year-old Dell.
Mark H.
-
<snip>
"Myth #1 is that using an oversized power supply will consume significantly more power at idle than a smaller unit since PSUs are less efficient at low loads and high loads, and are at their best between 40% load and 80% load".
</snip>
That is quite correct. The size of the power supply has nothing to do with the amount of power being consumed.
You and I are looking at two different images of the scope as it is quite clear the power at the 70-80% load is cleaner than at any other load. Still noisier than I would prefer, overall. I wonder if they used a ballast resistor in that load test?
Sure you can run a switcher between 10 and 95% loads, today, but the best power they output is going to be between 80 and 85%. From 50 to 80% the power will be a bit more noisier. As to whether or not components will be damaged foes, it depends on a lot of other factors. Anything with a DC motor is going to run hotter as the power gets noisier. Many motors will make more noise as well.
Intel put regulation into the CPU os it helps smooth things out there. If the motherboard has regulation it will also help, but that can also depend on the type of regulation. Electrolytic capacitors being used for the filter will drift with temperature and will fail sooner than ceramic capacitors. Digital filtration will be rock solid at the expense of being more sensitive to wider noise levels.
Motherboard manufacturers rarely talk about the type of power filtration used on the motherboard as it is something most people do not understand or care about.
You may be perfectly fine with an over-sized power supply, then again you may not. My personal preference is not to gamble with it.
-
All this made me think about the reasons why, and since I'm a simple person, my thoughts are those of a layman...
So, one thing I've learned is that heat is the enemy of many electrical components like capacitors. Another thing I've learned is that a system can get accustomed to a higher temperature and fail if it gets cooler all of the sudden. And I know that there's a fan speed controller in power supplies to control the cooling. And that bearings of rotating things such as fans have an optimal temperature where the lubrication works at its best to reduce friction and smooth rattling.
Now if only a small percentage of power is taken from the PSU, it supposedly runs relatively cool, the fan spinning only occasionally. Not only would that keep the bearings cold, also the variation in temperature would vary between the thresholds where the fan starts and stops. Based on the threshold levels for processors in the bios settings of motherboards the difference between hot and cold can be several tens of degrees. Wouldn't that cause variations in the output as well?
Likewise, if the PSU runs at a high percentage level, the fan would keep spinning to keep the temperature at an optimal level without major highs and lows, which again would reflect to the output.
I may be speaking about the egg rather than the hen, or I may be totally off the track here. As I said, this was just some logical thinking based on limited knowledge and no education. Please correct me if needed.
-
TDeacon, you've understood my thinking correctly. However, a single cold start isn't what I'm thinking about. It's the long term effects of a continuous waving between 20 and 50C instead of a constant 35C that worries me.
-
Looks pretty good.
I would seriously consider replacing one of the 1TB HDDs with an SSD,
OMG yes. He won't regret it.
-
TDeacon, I made a really bad mistake when I went to reply and hit the delete instead. Here is your post.
Now if only a small percentage of power is taken from the PSU, it supposedly runs relatively cool, the fan spinning only occasionally. Not only would that keep the bearings cold, also the variation in temperature would vary between the thresholds where the fan starts and stops. Based on the threshold levels for processors in the bios settings of motherboards the difference between hot and cold can be several tens of degrees. Wouldn't that cause variations in the output as well?
Likewise, if the PSU runs at a high percentage level, the fan would keep spinning to keep the temperature at an optimal level without major highs and lows, which again would reflect to the output.
Bizman; if I understand you correctly, you are wondering whether the quality of the output power would be affected by internal PS temperature, as controlled by its fan? The reviewer sort of covers this on the previous page covering cold testing, although he doesn't show scope traces. Note that the efficiency is almost identical between the cold and hot testing tables. In the absence of information to the contrary, as a purchaser all I can do is assume that the scope traces showing ripple + noise (defined in my comment below) would also be almost identical.
You and I are looking at two different images of the scope as it is quite clear the power at the 70-80% load is cleaner than at any other load. Still noisier than I would prefer, overall. I wonder if they used a ballast resistor in that load test?
Guys, here's how I am reading these scope traces, but keep in mind it's probably been 20 years since I used a digital scope.
Looking at this page (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story4&reid=494), I am comparing the Test 1 (~100W) to the Test 4 (~800W) rows. I am assuming that the reviewer has applied a DC offset to the traces, to allow better evaluation of the portion of the signal which is of interest. IIRC, the deviation from a idealized DC horizontal line is caused by a combination of ripple and noise, and is normally measured peak-to-peak. So at 3.3V, for Test 1 (~100W) I measure 5 mV, and for Test 4 (~800W) I measure 8 mV. At 12V, for Test 1 (~100W) I measure 6 mV, and for Test 4 (~800W) I measure 9mV. So from these scope traces, the ripple + noise is better at 100W (10% of rated capacitiy). Is there a flaw in my reasoning?
MH (EDIT to correct typo)
-
You are reading the scope using absolute comparisons. Assuming they used an analog load, rather than a digital load, you need to extrapolate the percentage of the ripple from load to load. See if that clears it up.
-
You are reading the scope using absolute comparisons. Assuming they used an analog load, rather than a digital load, you need to extrapolate the percentage of the ripple from load to load. See if that clears it up.
I am comparing (ripple + noise) for 2 different 3.3V signals, and comparing (ripple + noise) for 2 different 12V signals. Since the DC component is the same for both 3.3V signals and for both 12V signals, I say absolute comparisons are valid. For example, the reviewer's scope traces for 3.3V show about 5 mV for 100W load and about 8 mV for 800W load, with the 100W load being less (which is better). If I calculate percentages, I get 5 mV / 3.3 V x 100 = .15% for the 100W load. 8 mV / 3.3V x 100 = .24% for the 800W load. Again, the 100W load value is less (which is better). Same conclusion either way.
Unless I have overlooked something, which does occur occasionally, I think we'll have to agree to disagree here. Conclusion is of course based on the reviewer's data for ***this particular power supply***, and does not necessarily invalidate your statement about switching PS in general. Thanks to everyone for their inputs to my questions. :)
MH
-
There is a big difference in an analog load ballast test of a switching power supply and a digital load test of a supply. I do not have the time, right now, to explain exactly why that is. If I get some time later, I will try and hit the highlights and explain why you are not getting the correct picture from what you are seeing.
-
The center-top-of-screen toggle text shows 100 fps offline and 143 fps online with my new 144 Hz monitor, with all settings on max. Given my system (see OP) is this normal?
Thanks in advance,
MH
-
Offline and online differ by the settings for the arena. Online and offline use different settings, unless you have gone through and carefully matched the offline settings with the online arena settings.
Time of day will matter. Clouds matter, and so on.
-
Hi Skuzzy,
What I meant was, is the 100-plus fps number at the top of my screen real? (Yes or No).
(Anything over 100 is good, as I never got better than 60 with my old PC, even when it was new 15 years ago).
MH
-
I went with a pretty similar list for my first build. I went with a 650w supply and it has served me well.
https://pcpartpicker.com/user/eskimojoe4269/saved/CDnLkL
Even with the new 1080 TIs, I don't expect one would need more than 850w unless they're planning on doing an SLI setup later on.
I learned a lot from these guys. I would consider their opinions to be quite valid.