There are several reasons why programs come as installers rather than standalone executables:
File Size Concerns
Programs with many, large dependencies can bundle Web-based installers that download the dependencies and place them in a common location, so that they can be shared by multiple pro...
@PatoSáinz well there's that, but you could also just ship the program itself as a statically linked binary, like nirsoft does, or the demoscene (kkrieger, etc)
it sounds like he's basically asking why we need installers instead of just having everything bundled into self-contained EXEs
I've done a lot of diagnosis and I think I have the problem pinned down well enough. I just don't know where to proceed from here.
My parents have my old Dell desktop computer (XPS 410, if you're curious). It works solidly enough, and never has had any issues. I've even done a few upgrades withi...
@allquixotic Haha. I probably have already. I trust you are. :P
I wish IE metro scaling settings and IE desktop scaling settings weren't shared. They could just make the two settings independent of each other. If I want them to be the same, I'll set them both to the same values. :|
you really have two questions: how can I isolate the problem to static/grounding or not, and how can I specifically address the issue of static/grounding
in my opinion, when I see issues like the one you're experiencing, 90% of the time, if it isn't related to any specific USB port, it's a power issue -- for instance, your PSU might be providing an inconsistent amount of power to the USB hub, and the devices requesting power might not be able to get enough -- if a powered USB hub improves the situation, it's definitely that
@BenRichards so you've already tried powered external USB hubs (with an external plug-in adapter to supply extra power from the mains) and they didn't help?
In fact, the keyboard and mouse used to be plugged into an external USB hub that's built into the desk. I bypassed that to eliminate the possibility of a faulty hub.
IIRC, a USB hub that can't supply enough power to satisfy all the connected devices will start shutting them down; I've had a similar issue before where the PSU was not able to always satisfy the power demand of the USB hub, so the devices would go off and on
you can't assume the PSU is providing a perfectly linear and consistent amount of power -- often they don't -- and the amount of power they can provide is related to load and heat as well
One of the USB devices (in fact the first to start failing) was our UPS, which everything is plugged into.
The computer is connected to it via USB to get status/emergency shutdown/etc
When this started happening, USB communication with the UPS just stopped. It couldn't initialize it properly. Then the keyboard/mouse stopped working. It all came back after I did the cleanout and plugged things back in.
this could be a little unnecessarily pricey, but you could also further narrow down the problem by buying a PCI Express USB controller and installing it into the machine -- assuming that the PSU is not at fault, that would completely eliminate the mobo's on-board USB controller(s) and ports from being at fault, and if you still experience the problem, it could very well be static
on the other hand, if you simply take greater steps to ground the box and problem is solved, then there's your answer
you might consider taping a metal wire (you could straighten out a metal coat-hanger into a long, thin piece of metal) to the inside of the chassis, making direct contact with the metal of the interior of the case, and then tape the other end to something that you know is in direct contact with ground (testable by making static on your body and "getting shocked" by that thing)
if you build up static on your own body, and you touch the running PC, and nothing blows up, and you feel a strong shock on your hand, then it's grounded
because the voltage has to go somewhere, and if it's not blowing out components, it's just going through the casing and out to the ground
if your PSU has three prongs, it should be able to use the ground pin of the power supply to discharge static buildup or voltage spikes of any kind... and it won't trip the circuit breaker unless the current/voltage is high enough to make it think that the mains power is being shorted, which static electricity from rubbing your feet on the ground isn't anywhere near that level of power
I know that you're never supposed to bypass the ground prong with those adapters. They're supposed to let you hook up ground to something else, not just leave it unconnected :P
that alone should be enough to keep static from being a problem, unless you directly touch a live component with a ton of static buildup on your body, in which case you'd fry it and get quite a shock
You think the PSU is going out, though? I do know that it should be grounding out static via the ground wire on the power cable. It's the stock Dell brand PSU so it's entirely possible that it's failing to ground properly anymore.
Like I said, the computer is from 2008. So it's 5 years old by now.
on the other hand, static charge / potential buildup on the system (best depicted by the way your hair stands on end when you are strongly statically charged) is probably not going to cause any negative effects whatsoever
the only way that would turn into something dangerous is if the buildup then gets sucked out to ground or to another object by traveling at high current through or over a sensitive component
if there are extra stored charges in the components, even if it doesn't blow anything out, they can cause malfunctions, purely because it puts the digital circuitry into random states that they're not designed to recover from
eh, but I can't really understand from a physics perspective how those charges could get inside the system if the charge buildup would originate from the exterior
the charge would build up on the chassis itself, if anywhere, and then discharge at the earliest opportunity when a conductive route opens up to ground
My years in verification, working with engineers, taught me that they don't error-check on the level software does, because of simplicity, power efficiency, and performance. They simply design it to be able handle a set number of expected states and constrain the input/output to certain known states. Garbage in, garbage out. :P
Well, you have flops, capacitors, and other places where charge can hang around. zero or one is really just a voltage level that's below or above a certain threshold, and the signals have to be clean enough to align with the clock, which also has to be a clean waveform.
So once you introduce random static charges, things can start behaving erratically.
They don't necessarily have to be strong enough to cause damage, but they will put the hardware into unknown states and cause malfunction
why do you think graphics drivers have to have a software interposer at the kernel level that carefully verifies any commands it sends to the graphics hardware -- basically if you can get the graphics driver to emit any sort of bizarre or non-spec hardware command to the GPU, the GPU will reset
@BenRichards the highest magnitude earthquake to hit this state since they started keeping measurements in 1758 hit in 2011
some geologists opine that the midatlantic Delaware-Maryland-Virginia-DC area will see major, potentially catastrophic seismic activity, possibly even frequent (on the order of several damaging events per century) starting "soonish" (on a geological scale)
it's been very dormant for hundreds of years
btw I was referring to this earlier when I talked about drivers having to police the command stream being emitted to the GPU to make sure that no spec violations occur
the GPU is like an English professor who's an absolute stickler for the language, and if you say anything to them that isn't 100% grammatical, they just stop listening to you
When I did verification, we basically had to make our tests follow the spec. Any behavior resulting from input data that's out of spec (that's "garbage") we wouldn't bother testing, because it doesn't make sense anyway.
We couldn't afford to spend resources debugging garbage anyway. There were real bugs to tease out, and the drivers were to make sure that out of spec behavior never appeared anyway.
It's surprising when you come from a software background, because you're trained to code for every possible behavior, and test for out of spec behavior as much as possible.
Hardware is a different beast. :P
Evergreen! I remember... I think that was the R8xx series GPUs, wasn't it?
it seems like it would be fairly hard to write a graphics API that properly policies all possible function interleavings from the client (the GPU-accelerated program), does what you expect most of the time, tolerates stupid programmers making mistakes, and still doesn't emit anything that the GPU won't chew up, spit out, and crash
@allquixotic Well engineering GPUs and making their drivers is a very complicated and difficult task. At least if you want to be industry leading, like at AMD ;p
It's gotten much better over the past 5 years or so
@BenRichards not in my experience! heh... I get crashes "all the time"
my BSODs don't follow any particular pattern or happen with any particular frequency, so I have to assume it's the drivers... sometimes I can go an entire 4-day holiday-vacation weekend without a single crash and leave my box up 24/7; sometimes I'll play for 5 minutes in an evening and get a BSOD almost immediately
I had a BSOD while playing XCOM: Enemy Unknown during my recent 4-day weekend, but that was the day after playing for 12 hours straight with a break for thanksgiving dinner
in theory the TDR mechanism is supposed to be able to handle that, but when you have a hardware part in an inconsistent state, it's always possible that the only way to get it back to happy is to do a full system reset
especially if a result of the inconsistency is that main memory gets corrupted
My only recurring issue is likely an overheating PSU. It vents down and despite my case having risers, it's on a carpet and likely can't ventilate properly.
Unfortunately it's too big to put on my desk. I need a wooden plank or something
When I had it on the table at the PAX East BYOC area, I left the thing running every day for hours and suffered no issues, yet it would routinely shut down (with no hardware failures in the event log) when I had it upstairs on our medium pile carpet.
they do that for enterprise parts that you pay $2k for, so I figure why pay $2k when I can get a part binned at the HD7970 performance level and then underclock it like enterprise does :P
It works, but needs to be ventilated better. It's not a bad PSU, just bad airflow, I think
Heh
I was using 6970s in crossfire. I now have a single factory overclocked 7970
Works swimmingly
Which, incidentally, I got for $490 (iirc) in May, has since been discontinued, and just yesterday I saw on Ebay going for $750 with 4 bids and 29 days left :P
Asus P8Z77-V here... they're pretty good with supporting firmware updates but I don't like their poor documentation, the Virtu MVP, and several other things about my mobo -- probably won't buy Asus again
next build I'm tempted to go Intel mobo, Corsair mobo and PSU, Nvidia GPU and Intel CPU... spare no expense, raah raah, blast out a bunch of inefficient power usage and heat from the Nvidia chip, SOI etc, but at least the bloody thing will work
There are several reasons why programs come as installers rather than standalone executables:
File Size Concerns
Programs with many, large dependencies can bundle Web-based installers that download the dependencies and place them in a common location, so that they can be shared by multiple pro...
