|
If they're hybrid coolers, I assume you mean that there are radiators with case fans mounted on them where the case fans alone normally would be? That seems fine to me, if you have 3x120mm exhaust and 1x160mm intake. They're still going to ensure proper airflow throughout the case; it might not be quite as much as unobstructed case fans but 4 is a good number. If you mean that they're not contributing to case airflow somehow then yeah, you need to do something to keep airflow on the motherboard and DIMMs and HDDs if any too.
|
#
¿
Nov 21, 2016 05:50
|
|
|
|
| # ¿ Apr 26, 2024 20:03 |
|
|
Chuu posted:just a nice switch connected between two standard NEMA plugs would be fine Like this?
|
|
#
¿
Dec 6, 2016 15:46
|
|
|
Hard drives typically can't hit SATA III speeds except with the cache anyway, so it's not a huge loss to use SATA II ports instead. I've never really seen many professional reviews of consumer SATA cards, so it's kind of hard to know what to expect except by skimming consumer reviews and trying to pay attention only to the ones that seem to know what they're talking about. If anyone else has better sources of information I would also be curious to know. Depending on how many ports you need, a used server card might make more sense.
|
|
#
¿
Dec 9, 2016 01:08
|
|
|
^ Looks fine except is there a reason you're only going for the 2-port model? ^Furism posted:Different people will have different opinion about RAID but I'm against it. The problem is that if your RAID card dies, you'll need to have a spare one with the exact same chipset. Otherwise you can't rebuild your RAID array. Sometimes even within the same brand they are not compatible. Just saying. Some RAID expert will probably contradict me here. I don't know about more professional solutions but Intel advertises being able to move any ICH* based RAID to any other Intel system that supports the same type of array (as in, you can't put RAID5 in ICH5R because it only supports 0 and 1) and it should work. If you're going to use a dedicated controller instead of the Southbridge, I'd recommend either having a model that you know you'll be able to get a replacement for if needed or just using a software RAID. Windows Storage Spaces is pretty simple to set up and also advertises being able to move the array to any system that supports the same feature, and for Linux you have a couple options including Greyhole and mdadm.
|
|
#
¿
Dec 9, 2016 16:38
|
|
|
Furism posted:Yes software based is what I would recommend (on a budget - enterprises have way more options of course, but they are all costly), but not using soft RAID itself but rather newer filesystems like ZFS or btrfs. These file systems have built-in RAID-like functions and actually fix a couple of problems RAID has (write hole). I've heard that btrfs still has a few pitfalls out there that can cause data loss if you don't know how to avoid them, and in my own experience it can be a bit trickier than something traditional like ext4 to set up mounts and if you have issues to do filesystem repair. However, I'll freely admit that I'm just knowledgeable enough in Linux to cause problems that I can't fix so it might be simpler to use for someone less ignorant. For ZFS, I've heard great things but it all but requires a lot of ECC memory right? I guess what I'm wondering is, for a home user with a low-end dedicated NAS box what are the issues with soft RAID that make these solutions superior? Right now I'm using Storage Spaces' RAID 5 on my main desktop for bulk file storage with 3x3TB Seagate NAS drives, but I've been toying with the idea of using my N3150 mini-ITX box with something like a Node 304 for a lot of 3.5" bays the next time I need to upgrade. I had been thinking I'd probably use mdadm but I'm curious now. Eletriarnation fucked around with this message at 18:25 on Dec 10, 2016 |
|
#
¿
Dec 10, 2016 18:22
|
|
|
Ah yeah, I see what you mean. I agree that there's a potential pitfall if you use a NAS device or similar to do software RAID, since you might have issues if you have a hardware failure in the device and can't procure a replacement. I meant more the kind of software RAID that you get from OSes that can be installed on generic hardware, like mdadm or Storage Spaces. That seems like it would be more similar to ZFS in practice, right?
|
|
#
¿
Dec 10, 2016 19:06
|
|
|
Paul MaudDib posted:Or alternately, a Zotac ZBox/Liva X/some other little micro-PC running Windows on a laptop processor, roughly $100-200 depending on what you want. This would be my recommendation. I know that Intel is about to come out with a new "Apollo Lake" quad-core (Atom) Celeron NUC with HDMI 2.0, and something like that would seem like a good forward-looking option if you want a cheap, compact system for Steam streaming and video watching. You don't need to run Windows either if you'd rather save money on the license, Steam streaming works on Linux.
|
|
#
¿
Dec 14, 2016 02:48
|
|
|
The one that's spinning at 500rpm, does it work as expected if not plugged into the 7V adapter? Does the other one still spin at 1200rpm if you swap adapters? Try to isolate the anomaly to a specific fan, adapter or header.
Eletriarnation fucked around with this message at 16:39 on Jan 3, 2017 |
|
#
¿
Jan 3, 2017 16:34
|
|
|
Get one that claims a sine wave or "pure" (not simulated) sine wave output. I use this one. Notice how it says 1000VA/600W; this is because power supplies without Active PFC (as well as some other appliances, I think laser printers are infamous for this) will draw more power from the lines than they actually consume, so the VA measurement indicates how much power can be drawn and the W measurement is how much can be consumed. With Active PFC, you don't really have to care about that because the numbers are within 1% or so of each other. Also, realize that getting a higher-wattage power supply is not going to increase the power draw of your computer unless you're moving from a high-efficiency supply to a lower efficiency one. The DC power draw will be determined by the load - that is, what components you're using - and the AC power draw from the wall or UPS will be equal to that plus some small proportion for conversion inefficiency. Most supplies worth having these days are at least 80% efficient, so the AC power drawn from your UPS will be at most 125% of the DC drawn from the power supply by the PC. Exceptional ones are 90+%. Also, efficiency tends to peak in the middle and drop off for loads near the PSU's limit or very small loads. For that reason, you'll get better efficiency under load from a power supply that has capacity to spare as long as you don't go nuts and get an 850W supply for your HTPC or something like that. Eletriarnation fucked around with this message at 03:31 on Jan 15, 2017 |
|
#
¿
Jan 15, 2017 03:25
|
|
|
Something like a card reader is tricky to predict precisely if it's taking power from the USB header, although I would hope and expect that it won't use a substantial amount unless actually accessing a card. Ports shouldn't use any power at all unless they are in use, and even a passive hub from a front bay that turns 2 ports into 4 wouldn't take much. If you start daisy-chaining unpowered hubs I expect that the voltage drop would cause issues before the passive power draw from the hub takes enough current to be noticeable. A lot of desktops and some more recent laptops have a feature to deliver more power from some or all of the ports than the 500mA allowed by USB1/2 spec or 900mA allowed by USB3, so I would definitely check into that if this is a practical and not academic query. For example, some recent NUCs have a yellow USB3 port that can provide up to 1500mA.
|
|
#
¿
Jan 16, 2017 08:07
|
|
|
Yeah, the 750W is a limit and not a static value. If you have a 275W 980Ti, a 100W processor, and 75W of other poo poo in your computer on a 90% efficient supply then it will only draw 500W from the wall/UPS even if it's rated for twice that. If your current UPS is a simulated sine model and you plan on buying a new power supply with active PFC, then you should get something like that. You could maybe get away with a 600W model, but 900W will be less likely to crap out on you while you're redlining it playing games or something - especially since I assume you have at least one monitor plugged into it too. (e: That 810W model should be totally fine too, although it'll only save you $20.) Eletriarnation fucked around with this message at 05:52 on Jan 20, 2017 |
|
#
¿
Jan 20, 2017 05:45
|
|
|
Deviant posted:Is it normal that my i5-6600k clocked at 4.2Ghz fluctuates from 4200mhz to 4195mhz from time to time? Yeah, small BCLK fluctuations and resulting small CPU clock fluctuations are normal. I just checked on my 2500K/Z68 setup and HWInfo says that BCLK has been anything from 98.7MHz to 99.8MHz while it's been running.
|
|
#
¿
Jan 22, 2017 19:51
|
|
|
Converting an HDMI output to go into a VGA input will require some kind of boutique active converter because HDMI is a digital signal and VGA is analog; they're totally different and incompatible. There are probably boxes that do it but they may not be cheap.
|
|
#
¿
Feb 16, 2017 23:32
|
|
|
Yeah, recently I've checked into replacement batteries for an X220 and an E6230 (Ivy Bridge) and from what I saw, you can still get OEM batteries for these older business systems if you look around but they're 2-3x the cost of similarly sized off-brand models. Of course, resilience of the off-brand models varies widely.
|
|
#
¿
Feb 23, 2017 16:23
|
|
|
katkillad2 posted:I don't know if this should be posted in the monitor thread, tv thread or GPU thread so I'll try here... AMD cards - at least older ones - generally default to 8% overscan when using HDMI (or DVI->HDMI cables on the DVI port) because it was needed to compensate for inherent underscan on some old displays and someone at AMD decided it would be better to have to fix black bars on normal screens than to be unable to see the edges of the desktop on an underscanning one. I've only ever seen this with 1080p and 720p screens (haven't tried 4k over HDMI with AMD) but hopefully your issue has the same fix. You should be able to find an overscan slider in the GUI for GPU settings - whatever replaced Catalyst Control Center, I don't remember the name - and drop it all the way down to fix your problem. If you're saying that slider's not there though... It can also be fixed in the registry (HKEY_LOCAL_MACHINE/SYSTEM/CurrentControlSet/Video/<hex identifier>/0000/<something like DALR6DFPI[resolution][refresh rate]>) by finding the byte that's an 08 by default and changing it to 00. This needs to be done for all resolution/refresh rate combinations you wish to fix. This is the only way to fix this issue for cards older than the 5000-series in Windows 8.1 or 10, which is why I know about it - I have a 3450 and a 4850 which still work great but without this fix they kind of suck if you want to use HDMI. Eletriarnation fucked around with this message at 00:54 on Apr 9, 2017 |
|
#
¿
Apr 9, 2017 00:23
|
|
|
The market for people looking to comprehensively upgrade their old systems isn't really large enough to be worth the effort of designing special versions of new processors that work in old sockets. The memory controller would have to have multiple versions to accommodate different versions of DDR, and iGPUs would have to be designed with old chipset compatibility in mind as well. Motherboard manufacturers would have to put out BIOS updates (for free?) with new generations of CPU for motherboards they sold years ago. There are a lot of considerations. You could make a case that sockets should only change when the memory technology changes - e.g. Haswell/Broadwell should have been on the same socket as Sandy/Ivy Bridge - but I feel like going any farther than that is of questionable value, and even then you have to make sure your motherboards are going to be designed with voltage regulation that can handle whatever range of voltages/currents will be demanded over several generations of CPU. I think Socket 775 was probably the longest-lived platform spanning about 5 years from Prescott P4s to Yorkfield Core 2 Quads, but even that isn't as unified as it might appear since older systems don't have the chipsets and software support for the highest-end CPUs. It was only reasonable then because the memory controller was off-die and you didn't have to worry about popping a DDR1 processor into a DDR3 board or vice versa. Eletriarnation fucked around with this message at 15:56 on May 5, 2017 |
|
#
¿
May 5, 2017 15:49
|
|
|
Hipster_Doofus posted:Yes, but not for long. Keep a finger on it and if/when it gets too hot to touch, turn off using the power supply switch.
|
|
#
¿
Jun 20, 2017 23:13
|
|
|
Alereon posted:It looks like the voltage is different so they can't be mixed. It looks like yours is standard 1.5v, hers are low voltage 1.35v. It wouldn't hurt anything to try though, just don't use the system if it can't pass the windows memory diagnostic, and be prepared to have to remove the stick if it has problems. Usually in this situation I think both will run at 1.5V and I've never heard of DDR3L that actually had issues running at 1.5V, so it should be fine. You probably don't want to mix 1.35V DIMMs with old desktop DDR3 that runs at 1.65 unless you know the old stuff is stable at 1.5 too and can set it there though, running >20% above spec could be an issue for longevity. I've never heard at 1.65V modules that wouldn't run at 1.5V either but if it's crazy low-latency stuff you might need to loosen timings some. Eletriarnation fucked around with this message at 19:42 on Jul 20, 2017 |
|
#
¿
Jul 20, 2017 19:40
|
|
|
Alereon posted:If I'm understanding he wants to run a 1.5v DIMM at 1.35v, and while that doesn't risk damage it has a very good chance of not working. Yeah, running 1.5 at 1.35 might well not work. I figured the laptop wouldn't let him choose though, it would just force 1.5 and work fine.
|
|
#
¿
Jul 23, 2017 01:27
|
|
|
Interesting, I use one of those to turn my Kill-a-Watt into a dongle so that it doesn't block power strips. I hadn't paid any attention to the gauge before but will have to keep in mind not to test something nutty with it.
|
|
#
¿
Jul 28, 2017 02:21
|
|
|
If powerline isn't practical for whatever reason get a mini-PCIe to desktop x1 bracket and an Intel 7260 half-mini card. You can get them already combined but they cost more that way.
|
|
#
¿
Aug 4, 2017 15:28
|
|
|
apropos man posted:This talk of using the wrong PSU is making be think twice about continuing to run my home server off a EVGA Supernova 650w supply. It's an i3 in a MATX board with 4 drives attached: one of them an SSD. Unfortunately, you'll find that in the standard ATX form factor high efficiency power supplies are rare below 450-500W. I don't know if 200W power supplies are available at all retail, and the only ones that are Gold or higher rated that I know of under 450W are from Seasonic. I have their 400W Platinum fanless model in a machine that I used as a home server for a while and was really happy with it, but if you already have a well-rated supply attached I don't think you'd ever save enough money from the efficiency gain to make up the cost of buying it. Even the best PSUs will have a noticeable efficiency drop below around 10-15% load, but all that means is that instead of 45W you're using 50W or whatever. Eletriarnation fucked around with this message at 01:01 on Aug 7, 2017 |
|
#
¿
Aug 7, 2017 00:56
|
|
|
From what I recall the built-in surge protector on the Cyberpower consumer sine-wave models is not very robust, and some reviews recommend you put them behind a dedicated surge protector if it's a concern at all. I did that with the 1000W model for months and didn't have any problems.
|
|
#
¿
Aug 30, 2017 18:50
|
|
|
You probably want to make sure to get some level of 80+ certification just because any power supply without it is likely going to be a piece of poo poo with efficiency in the 70s and no Active PFC. White (or just '80 Plus') and Bronze are "I am on a budget/this machine won't be used much" tier. Silver is very rare in practice, most supplies that might fall in this range are either rated Bronze or good enough to rate Gold instead. Gold is kind of the standard tier these days and what most people will recommend for any machine that will be running for several hours a day, especially something with a high-wattage CPU or GPU. Enterprise-grade server supplies pretty much start at Gold and go up from there. Platinum is rather expensive but might make sense if you're going to run something 24/7 or have a really noise- or temperature-sensitive application. I bought a Platinum fanless supply to use in a home server, and they're quite common in enterprise servers at this point. Titanium is for crazy poo poo or when cost is no object. I wouldn't look at it with current pricing unless I had a unique need or planned on approaching/exceeding 1000W in daily use. Eletriarnation fucked around with this message at 17:36 on Sep 13, 2017 |
|
#
¿
Sep 13, 2017 17:31
|
|
|
When it freezes, is the OS locking up entirely or just Explorer (the folder/taskbar/start menu GUI) locking up? Can you get any response with a Ctrl-Alt-Delete? I haven't seen issues with moving an old system disk from one machine to serve as a data disk in another, but admittedly I usually format the old disk first thing and don't actually keep the old partitions longer than I need to get the data off of them.
|
|
#
¿
Sep 13, 2017 17:57
|
|
|
Transient People posted:Thanks for the info on PSUs guys! Super helpful stuff. Yeah, get bronze for sure in that case. In the US it's probably 50% or less price increase for gold at similar wattage so it makes more sense to pay the small margin for efficiency. As far as the supply goes I'd say high wattage is over around 650W but most computers don't use anywhere near that, 300-400W at full load is more typical for single CPU single GPU gaming machines.
|
|
#
¿
Sep 14, 2017 18:58
|
|
|
Most DP ports on video cards have an HDMI signal which can be separated out by a passive adapter or cable (DisplayPort++) so you don't need the active adapters, but the DP++ signal is capped at HDMI 1.4. If you need HDMI 2.0, you will have to use an active adapter as I learned when trying to connect an older NUC to a 4K TV.
|
|
#
¿
Oct 11, 2017 05:13
|
|
|
Faith For Two posted:How do I connect a computer to wifi if I only have an ethernet port? Mikrotik, Ubiquiti, and DD-WRT all offer various bridge modes which allow you to use a router as an Ethernet-connected wireless adapter. For UniFi it's called Wireless Uplink and bridges the base station's same subnet seamlessly but you have to have a UniFi node as the base station as well. Mikrotik and DD-WRT I think both have a WDS mode which will bridge the same subnet across from another WDS-supporting devices, as well as client bridge modes which can provide a new inside-NAT subnet to wired clients with any kind of base station.
|
|
#
¿
Nov 19, 2017 04:11
|
|
|
It's a problem specifically with active power factor correction, which is ubiquitous in newer and more efficient PSUs. Older models (esp. before about 10 years ago) or modern bargain-basement models are usually passive PFC, which is easily identifiable on PSUs that have it by the presence of a source voltage selection switch. This also conveniently means that if you see a voltage selection switch on a new power supply, you can be rather confident that it's a cheap piece of poo poo.
Eletriarnation fucked around with this message at 03:40 on Dec 7, 2017 |
|
#
¿
Dec 7, 2017 03:31
|
|
|
Kibbles n Shits posted:Will a modern SATA 3 SSD have any issues on my old rear end Z77 motherboard? Two of my SATA ports say "6gb", I assume it'll work fine but I wanted to make sure before I pulled the trigger, I can always just wait until I upgrade the whole system later This specific case was already addressed but more generally, SATA is backwards- and forwards-compatible with other versions of itself. You can plug a SATA1-2 drive into a SATA3 system or vice versa and they'll work together at the highest speed they both support.
|
|
#
¿
Mar 19, 2018 18:44
|
|
|
Drunk Badger posted:I have a gaming PC with two monitors, and I've found that when I run video on my second monitor at the max 60 fps, it messes with my games running at 144 fps on the main monitor. Adding Chrome to the mix for looking things up screws with this even more. I'd like to get a mini pc to only run video / websites on (no gaming), something larger than a stick PC with the following: Apollo Lake NUCs hit every item on the list.
|
|
#
¿
Mar 21, 2018 19:19
|
|
|
Drunk Badger posted:How about something like this that appears to have everything already in it? It'll work, but it has HDMI 1.4 so 4K output will be limited to 30Hz. Eletriarnation fucked around with this message at 15:57 on Mar 22, 2018 |
|
#
¿
Mar 22, 2018 15:46
|
|
|
Yeah, I have a mini-ITX Celeron N3150 board from ASRock and have been really pleased with it. It's tempting to try to replace my Core 2 NAS/Plex server for power savings but I'm already using it as an HTPC, I'm not sure if it would be fast enough, and I already have the Core 2 system idling around 50W including the drives so I'm not sure how much lower is feasible. The other issue I notice with those prebuilt mini-PCs based on Cherry/Bay Trail in addition to lack of HDMI 2.0 is that a lot of them come with only 32GB of built in storage. 32GB is perfectly fine for Linux but it's pretty much an absolute lower bound for Windows 10 and tighter than I would want to deal with even as someone who usually doesn't mind making such compromises. With 32GB you basically have to keep the system disk entirely clean to even have enough space to install major OS updates, so even 64GB is a huge step up.
|
|
#
¿
Mar 23, 2018 15:58
|
|
|
I bought an HP Z420 motherboard "as-is" since I was looking for a new project and willing to accept the possibility that it wouldn't work. Upon receiving it today, I noticed that although everything else looks physically OK one of the larger capacitors is hanging from one pin; the other has broken off and is still soldered into the board. If you're looking at the picture in this entry, it's the one a bit below center on the right edge between the 8-pin ATX12V and the weird HP proprietary fan connector. I am betting that this is probably for the fan and won't be needed if I use a different fan header, but it seems like the best course of action would be to try getting another capacitor and attempt to solder it in. The capacitor is labeled as Rubycon brand, 16V 470uf, 105C, and "3M1220" + "ZLH" which I'm betting are proprietary markings. If I get another 105+C rated 470uF 16V capacitor of good quality, is there anything else I need to check before being confident that it will work? Looking at these on Amazon, the picture's totally wrong on voltage/capacitance but the title seems to match what I want. I'm a bit skeptical of most of the random brand names that sell capacitors on Amazon but I don't really want to wait for shipping from DigiKey or whoever. Also, my solder is from over a decade ago and contains lead. Do I need to get some of the new nontoxic stuff to get a good joint here? Eletriarnation fucked around with this message at 19:52 on Apr 27, 2018 |
|
#
¿
Apr 27, 2018 19:44
|
|
|
Yep, old-rear end roll of Radio Shack brand "Standard Rosin-Core Solder". Sounds good, thanks!
|
|
#
¿
Apr 27, 2018 21:03
|
|
|
Yeah, could have done that but I'd like to get the $40 motherboard working instead of sending it back and spending $100+ for a known working one. Anyway, while trying to remove the existing capacitor leads I discovered that HP used some crazy moon solder which I can't melt at all. Even after buying a better soldering iron, letting it heat up for like 10 minutes at full power until the metal is changing colors, and holding it directly on each lead in turn for a while I can't see any difference. Am I doing something wrong/is this soldering iron actually somehow poo poo or is there another way that this is supposed to be done?
|
|
#
¿
May 1, 2018 15:53
|
|
|
There's a chance you damaged the USB controller, although I would expect the mobo to have overcurrent protection on it and a CMOS reset to possibly clear up the issue. If not though, you could try adding in a PCIe x1 card with USB ports to work around having a dead controller on the motherboard for most purposes.
|
|
#
¿
May 7, 2018 06:07
|
|
|
I have a setup like that at home, USB or KVM switches work but the best way to do it is if you have mouse/keyboard with native support for switching between multiple devices. I use an MX Master and the Corsair keyboard that supports wired, USB receiver or BT. e: and yeah, I'd definitely recommend using extra monitor inputs if possible over some kind of actual video switch since they can often have poor performance or behave unpredictably. That way even if you're still switching keyboard/mouse you only need to use USB. Eletriarnation fucked around with this message at 20:12 on May 16, 2018 |
|
#
¿
May 16, 2018 19:25
|
|
|
It sounds like there could be some crazy poo poo going on with the wiring on that circuit at least, do you have any issues with things other than your PC?
|
|
#
¿
May 19, 2018 16:20
|
|
|
|
| # ¿ Apr 26, 2024 20:03 |
|
|
The manual says "simulated" sine wave, which is a new term to me but I'm guessing it means that it's a stepwise approximation of a sine wave instead of the real deal. In the past those have been not ideal with active PFC power supplies, which are basically all power supplies worth a drat these days, since they could potentially cause a brownout when the switchover occurred. It might be good to check around and verify that other users of this UPS haven't had any problems running a recent PC on it if you can't find an affordable model that advertises true sine wave output. e: compare with this model I'm familiar with, where the marketing materials make sure to let you know that it's a pure sine output and active PFC compatible Eletriarnation fucked around with this message at 19:21 on May 21, 2018 |
|
#
¿
May 21, 2018 19:18
|
|