Tuesday, July 5, 2022

What to do if a PC does not POST, comprehensive guide

Nothing beats the feeling of having worked for hours to build a PC from scratch, then powering on, and it just working like a charm on the first try. Nothing is more disheartening and frustrating if, instead, the PC does not POST.

"POST" stands for Power-on self-test, which is a battery of tests that the BIOS of the computer performs immediately after power-on in order to determine that everything works as it should. The expression "does not POST" is used in general to mean that the startup process of the PC does not advance to trying to load the operating system from a boot device. (Some people may use the expression to cover cases where the OS starts to load but fails at some point, or crashes during the process, but this is technically speaking a misuse of the term because it's not part of POST anymore. In general, if you get a picture and can get to the BIOS configuration screen, it means that the system did pass POST.)

A PC failing POST may present itself in many different ways, but some of the most common ways are:

  • The PC seems to power up alright (leds on the motherboard light up, fans spin) but there is no picture (ie. the GPU or iGPU sends no video signal of any kind) and all indications are that the OS is not launching (although this might be hard to determine without an image).
  • The PC enters an infinite power-cycling loop, sometimes almost immediately after power-on (ie. it stays on for just a second or so, and then resets), sometimes with the power cycle lasting several seconds. It may or may not send out a video signal.
  • The BIOS gives a critical error message about a failed test that stops it from proceeding (either to the display, or using some status indicators on the motherboard, or with some motherboards even through its integrated sound chip).

NOTE: The PC power-cycling once after power-on, especially after having been turned completely off, ie. the PSU has been switched off or disconnected from mains, is usually not an indication of a failed POST, especially with modern motherboards. This is usually just an indication that the BIOS "trained" the RAM, and needs to reboot in order for it to function properly with the new settings. This usually happens when XMP has been enabled, and is normal intended behavior and nothing to worry about. Many BIOSes have an alternative XMP mode that can disable this behavior.

What to do if a new PC build does not POST?

Before anything else, a piece of advice: When building a new PC, always test it first with a minimal build on the table, in other words, just the motherboard, CPU (plus cooler), RAM, PSU and GPU (if a discrete GPU is being used).

Many people, even experienced PC builders, make the mistake of building a new PC completely, up to the last screw and zip-tie, all cable management, all bells and whistles... before trying to turn it on for the first time. If it then fails to POST, it may cause a lot of extra work (the worst possible scenario being that you need to replace the motherboard). You will save yourself a lot of unneeded work and frustration if you test it with a minimal setup before you even install the motherboard in the PC case.

Note, though, that even though it's rare, it's not impossible that the PC did work just fine with the minimal setup, but does not POST after the PC has been fully built. In either case all the things below to check apply.

Also note that some motherboards will have some kind of segmented display, POST indicator leds, or even a full-on mini-display to indicate the test that failed. While this can give an indication of the problem, surprisingly often it does not. This is because this is essentially just a "guess" from the part of the BIOS of what's wrong. Sometimes it will be accurate, sometimes it will not.

Also note first and foremost:

  • If you actually seem to be receiving some kind of video signal (your display is not saying "no video" or similar and is not going to sleep mode) but when you start up the system nothing seems to be happening (you only get a black picture or, perhaps, at most with some kind of small cursor on the top left or similar): The very first thing you should do is try going to BIOS: During launch keep pressing the 'del' key to see if BIOS shows up. If it's not showing up (but you are still getting video signal), try waiting for a minute or two (literally, no exaggeration). If you are getting a video signal but nothing seems to be happening it's possible that BIOS is taking a very long time probing hardware. One of the main causes for this may be a broken drive (hard drive or SSD). Some BIOSes will get stuck for a long time trying to probe a broken mass storage drive. If you eventually get to BIOS in this way, try removing all drives from the system to see if the delay disappears. (Then try connecting them back one by one until the delay appears again, revealing the culprit.)

If that's not the case, then the easiest things to check first:

  • Check that the RAM modules are properly seated. It's surprisingly easy to think that you installed them properly when in fact you didn't. A RAM module being even slightly out of alignment will cause a POST failure. When inserting a RAM module in its slot, there should be a clear noticeable click on each end of the slot.
  • Check that you have installed the RAM modules in the correct slots (this should be checked even if the PC POSTs just fine, because them being in the wrong slots may cause performance degradation). The instruction manual of the motherboard will have instructions on the order in which the RAM slots should be populated.
  • If you are using a discrete GPU card, check that it's properly seated (you'd be surprised how easy it is to make a mistake here). The vast majority of motherboards will have a clamp, a "lock", at the end of the top PCI-Express x16 slot that holds the card in place. This should make a clear noticeable click when you insert the card. Also check that it's fully inserted in the other end (ie. the end that's closest to the back of the PC).
  • Check that you have properly connected the additional GPU power cables to the GPU, if it needs them (the GPU card will have 6x2 or 8x2 sockets for these). Note that some modern GPUs will have indicator leds near these sockets which will light up if it's not receiving power to that socket, which is a good indication.
  • Check that your display is actually connected to the GPU (or iGPU if you are not using a discrete card) with a DisplayPort or HDMI cable. (You'd be surprised how often even experienced people make this newbie mistake. The more experienced you are, the more likely you are to dismiss such trivial mistakes as the possible cause.) If you can, try using another cable to check that the current one isn't broken.
  • If you are using a discrete GPU card and your CPU also has an integrated GPU, try removing the discrete card and connecting the display to the motherboard's own DisplayPort or HDMI slot. (If in this situation it does POST normally, it may be very difficult to find out what the actual problem is. It does not necessarily mean that the GPU itself is broken (if you can, test it with another PC). It may also indicate a problem with the motherboard, the CPU, or sometimes even BIOS.)
  • Check that you don't have anything connected to any USB port. (You'd be surprised how often this can cause problems.)
  • Try clearing the CMOS (the motherboard instruction manual will have instructions on how to do this). Do this even if you think it cannot possibly be the cause (for example because a minimal setup on the table worked just fine, and now that you have installed it inside the case it's not POSTing anymore). You'd be surprised how often this resolves the issue.
  • If you are using more than one RAM module, try using only one. Start with inserting it in the slot recommended by the motherboard's instruction manual. If the PC still doesn't POST, you can try the other slots as well. (If it does POST in one of these situations, but not when all the RAM modules are in their proper slots, then it's an indication of a problem that may be too difficult to solve. You may need to consult the shop or manufacturer on what to do in this case.)
  • If you are in a situation where the minimal build worked but the full build does not, try removing non-essential components (such as storage drives, front panel I/O, case fans and so on) to see if it starts working at some point. This may reveal the offending component, if there's a problem with one of them (or if you have somehow connected it incorrectly).  You may want to clear CMOS after removing some components, like storage devices (the existence of the storage device may be corrupting CMOS in some manner.)
  • One reason for the previous may be a broken SATA cable (if you are using a SATA drive). SATA cables are relatively fragile and can get broken if mishandled (eg. bent to extreme angles). This can easily happen when cable-managing the PC. If the problem disappears when the SATA drive is disconnected (or if BIOS outright states that it can't find the drive even when it's connected), try replacing the SATA cable.

Hardware compatibility issues to check (you should check these when purchasing the components, but re-check these anyway):

  • Check that the motherboard actually supports the particular CPU model you are using. Motherboard manufacturers will have on their websites compatibility lists. Check that the exact model you are using, with every single number and letter, appears there.
  • Even if that exact CPU model does appear in the compatibility list, check if the list entry specifies that it only works with a certain version of BIOS, and thus the motherboard may need a BIOS upgrade to support that particular model. If you have made the mistake of purchasing a motherboard that requires a BIOS upgrade to support that particular CPU, consult the motherboard's instruction manual and the manufacturer's website on whether this upgrade can be done without a display or not. (If it's not possible you'll either need another CPU that's supported out-of-the-box, or you need to return that motherboard and purchase another model that supports your CPU without a BIOS upgrade.) When making purchasing decisions, it's always a good idea to purchase a motherboard that supports your chosen CPU out-of-the-box, without requiring any BIOS upgrades.
  • Check that the motherboard supports the particular RAM model you are using. These will also be listed in the compatibility list of the motherboard model (every number and letter of the particular RAM brand and model should appear in the list.) While RAM incompatibility is less of an issue in recent years, it can still raise its ugly head sometimes.

Things that are more difficult to test and/or fix (but if you can, you should):

  • Check that there are no bent pins in the CPU socket (if Intel) or the CPU itself (if AMD). This should be quite visible when looking at the pins against a light at the proper angle, as a bent pin will be quite noticeable against the rows and columns of other pins. (A bent pin is a really bad problem because manufacturers will generally not give you a free replacement even if it's still under warranty, so usually you'll be on your own.) You can carefully try to fix the bent pin with a needle, and see if it helps. (If you can't get a replacement from the manufacturer this is essentially your only option because else the CPU or motherboard is just a useless paperweight.) Always be very careful when installing the CPU (and the CPU cooler). Carelessness easily causes bent pins, hosing the entire system.
  • If you can, test the system with another GPU card (even if your CPU has an integrated one and you already tested that.) This is usually the easiest of these to test because any random GPU card will do, no matter which brand, model or generation (as long as it's a PCI-Express card).
  • If you can, test the system with another PSU. This especially if the PSU you are using is not brand new (but even then you should, as it's not impossible for a PSU to be faulty out of the factory.)
  • If you can, test with another brand or model of RAM (even if it isn't explicitly listed in the compatibility list of the motherboard.)
  • If you can, test the system with another CPU of the same model that's known to work, or another motherboard of the same model that's known to work. (This is an option that's not practically possible for most people, but if you can you should.) If the exact same model of CPU or motherboard is not available but you can use another model that's known to be compatible, this will of course also do.

What to do if an existing PC stops POSTing?

You may have used your PC just fine for years, when suddenly it just stops POSTing (ie. the OS doesn't even start to launch and you may even not get any video signal). Note that this is different from the PC failing to boot (eg. the OS gives an error, a BSOD, or crashes during the bootup process). But even in this latter case it doesn't hurt to try these checks anyway.

These are things you can try to diagnose the problem:

  • Try clearing the CMOS. This is a quick and easy thing to test, and will oftentimes help. Do this even if you have not done anything that may have affected BIOS settings. CMOS may have become corrupted for some reason and cause POST to fail. If it starts working again after clearing CMOS, it may be an indication that the CMOS battery is running out. This especially if this happens again in the near future. If this continues to happen, try replacing the battery. (Consult the instruction manual of the motherboard.) Another symptom of the CMOS battery running out is if BIOS settings keep resetting to defaults even after you change them to something else (eg. fan curves, XMP profiles, etc.)
  • If this starts happening after a BIOS update, the culprit is quite clear. Try clearing CMOS, and if that doesn't help, see if you can restore the original BIOS (by either "downgrading" to it or, if the motherboard supports it, restoring the motherboard to factory settings).
  • Try the same things as listed in the previous check lists above (especially the easiest ones). It's not completely impossible that some component has dislodged over time due to vibrations. Also, if you have recently replaced the CPU cooler, or removed it and placed it again (for re-pasting the CPU), check that there are no bent CPU pins. (While unlikely, it's not impossible for a pin to become bent by merely installing the CPU cooler. This should always be done carefully.)
  • Check that there's nothing connected to the USB ports.
  • If the behavior of the PC seems erratic, ie. the reason for it not launching seems to change from power-on to power-on, it sometimes failing almost immediately, sometimes even getting to loading the OS but then failing (eg. a reboot, crash or BSOD), sometimes with random BIOS error messages, a very common culprit for this is a failing PSU. A failing PSU may still provide enough power to run BIOS but not enough to run the CPU, which may cause the system to not POST. The amount of power provided may also fluctuate, causing erratic behavior. If at all possible, try with another PSU that's known to be good. (You might also just make the gamble and buy a new PSU. It might be a good idea anyway, if your current PSU is many years old.)
  • An outright consistent no-picture POST fail may also be caused by a failing PSU. There's no other way of testing this than by trying (or buying) another PSU.
  • If everything else fails, it may be impossible to diagnose the problem other than by doing the more difficult checks in the previous list, ie. testing with another motherboard or CPU (or RAM). It's not impossible for these to break seemingly out of nowhere (one common cause being a current spike from the PSU or somewhere else).

Sunday, July 3, 2022

Why do we still use a 24/60 based time system?

Humanity has had the need to measure things for thousands of years, most importantly because of commercial reasons. Managing large amounts of produce, storing it, transporting it, selling it, buying it and so on becomes really complicated and confusing if there is no way to measure and express quantities. Knowing how much of what you have is extremely useful. Selling produce likewise becomes so much more manageable when prices can be set based on units of measurement (after all, not everything can be sold at a given price per piece, because "pieces" can be too small for this to be feasible, like grains of wheat or rice, or basically impossible to be divided into "units", like liquids.)

Unsurprisingly, over the millenia different peoples came up with different ways to measure stuff, using different units of measurement. Since there was generally very little contact between distant lands, isolated and semi-isolated peoples came up with their own units independently of each other. In addition, measuring and measurement units were heavily driven by custom and tradition, which meant that different things would often use different units of measurement, even though they all measured the same thing (such as weight or volume).

Indeed, for the longest time you could go to eg. an open-air marketplace and encounter a dozen completely different measurement units of weight, used to measure different things (eg. flour could use its own measurement units, wheat could use its own, peas could use their own and so on), and a dozen different units of volume (eg. beer could use different units from wine, which could be different from units of water) and so on.

The Imperial System of measurements succeeded in somewhat standardizing, unifying and replacing all these hundreds and hundreds of different other measurement systems, but many hold-outs and relics still persisted to some extent, and co-existed alongside the Imperial system for quite a long time.

At some point the scientific community got tired of this mess. As science progressed more and more, and measurements became more and more important and universal (and especially the accuracy of measurements became important), a single unified standard was needed. The Imperial system was awkward to use because converting between the different units (eg. inches, feet, yards, miles...) was difficult. On top of that, science was in more and more need for extremely small and extremely large units of measurement, for which the Imperial system was rather inadequate for.

They needed a simpler, easier to use, and more consistent measurement system that also allowed easily going arbitrarily small and arbitrarily large, so they came up with the metric system, which eventually became the standard in science. While the base units in the metric system are rather arbitrary, the conversion between different scales of the units is 10-based and thus very trivial to do. This better system replaced all of the thousands of previous systems (with the exception of the Imperial system in some places, which still refuses to die).

But one has to wonder: Given that the metric system replaces and defines all units of measurement, using a 10-based system... why doesn't it also do so with time? Why are we still using the ancient 24-hour, 60-minute, 60-second time measurement system? Fractions of a second are done using base 10, but other than that we are still stuck with this odd system that makes conversions harder than they could be (eg. how many seconds is 5.21 hours? Not easy at all to do in your head.)

The main reason for this is that by the time when the pressure to create a single unified measurement system was large enough that it was done, time units were already pretty much universally standardized.

While you could find a dozen different measurement systems being used even at the same marketplace, every single person there was nevertheless using the same hours, the same minutes, the same seconds. Nobody was using any other units of time. This form of timekeeping had somehow become universal, and had been so for very long time. Thus there was no impetus to change it. It was just easier to keep the system that everybody was already using anyway.

(It's not like nobody tried. Some countries did try to create a 10-based timekeeping system, with 10-hour days and 10-day weeks. All these attempts failed miserably because everybody simply ignored it. It never caught on.)

So in the same way as the Imperial system still exists today and is widely used in some countries, the 24-hour 60-minute system likewise still exists today and is still ubiquitous. Tradition. There was never any impetus to change it, so it was never changed.