Hard Light Productions Forums
Off-Topic Discussion => General Discussion => Topic started by: Nuke on December 19, 2011, 02:54:41 pm
-
so i built a new computer (il spare you the case, drives and cooling crap):
intel i5 2500 @3.3 ghz (not the more expensive overclocking varient)
asrock z68m/usb3 mobo
gskill sniper 8gb ddr3 1600 dual channel kit
evga gtx560
all powered by an antec bp550+ 550W psu
but thats not what this thread is about. every time i build a computer i note the changes to the systems. some are good, everything is faster, more optimized, and there are fewer parts to worry about than last time. but there are other things, good cases are huge, power requirements skyrocket, fans get noisier. so ive created a long as **** list of all the things i dont like about the way computer hardware is evolving.
first gripe: ram
for the first several days of operation the system was rather unstable, resetting the bios to defaults helped the issue a little, but the system wasnt solid enough to play games. i started to suspect the psu. its 3-rail configuration made me think that no one rail had enough juice to power the card. of course after diffing around throught he documentation on the website (why manufacturers always fail to include complete specifications in the box is beyond me) i found that the stand alone pcix rail was capable of delivering more power than the card could use, and if it didnt i could have just patched another rail into one of the video card power connections (i also had the option of swapping for the 750 watt from my old machine).
so next possible culprit was memory, memtest86 would just crash at around 30%, but without producing actual errors (most situations where memory config is not right generally produces errors instead of just crashing). the bios autoconfigured the ram to spec and i that was a good sign, and so i went with my instinct and gave the voltage another notch and the problem vanished. this problem normally manifests as errors and not by crashing memtest86. furthermore the ram specs said that 1600 was the best the ram could do without having to do anything overclockey, which is something spiking the voltage usually goes with. at least it wasnt as big of a ***** as my last 8gb of ram, which had to be downclocked to 800 to even function.
tldr: ram should be built to more precise specs.
second gripe: fan headers
not enough 3 pin fan connectors on the mobo. granted the mobo i got was a reduced sized microatx, it only had headers for a 2 3-pin fans. which is funny because this is the first time ive even bothered to buy 3 pin fans. most mobos, even small ones give you at least 4. maybe its because i used a brand of mobo ive never used before. so one of my fans just runs full speed all the time. the plan was to make the rig less noisy when not under load, and that plan has gone out the window. they have fan controllers for this kinda thing, but a monkey with a knob is nothing compared to a completely automatic proportional control system with feedback. i got some tiny 8 pin microcontrollers which could probibly drive the fan (2 pulses per rotation is easy to do) and get analog input from a thermistor. then its just a pid algorithm away from being as good as the mobo at controlling fan speed.
tldr: not enough 3-pin fan connectors!
third gripe: cpu sockets
this one pertains to the cpu socket. seems every generation installing a cpu gets scarier and scarier. its not like the totally awesome works every time p4 stock cooler, or the equally awesome p3 chip and cooler on a card. even the spring-loaded strap of doom on older cpus was rather tame in comparison. a cpu socket should never require so much force to close that you feel like your gonna rip something off of the mobo. the c2q setup was bad just because it always seemed kinda weak to me, like there was nothing holding the heatsink on the chip, but the load plate was easy to close. but this i5 slot, the load plate involved flexing a steel bar with an insane amount of force, and it took me several minutes of fiddling around and thumbing through the manual before i figured out that this was normal. heatsink was the same loose as **** kinda thing the c2q chips used. but this one kinda felt more solid, as if they fixed the bugs or something. but ****, how much force does a cpu socket need? i should also point out that if that torsion bar ever decides to break while my computer is on, the cpu will die some 3 seconds later. oh joy.
tldr: make cpu sockets only require a sane amount of force to close.
gripe four: screws
this one is short, so no tldr. half the standoffs that came with my mobo weren't tapped correctly and the screws liked to strip. higher grade fasteners please.
gripe five: cases suck
the case had a really unorthodox way of mounting hard drives. hard drives mount via rubber bushings in a metal plate into the seldom used vertical screw holes in the drive. this plate is then mounted vertically between the 5.25 bay and bottom of the case so that the bottom of the drive faces the inside of the front panel. the drives may be installed so the sata connectors face forward or back. the bushings are a nice touch, and they might be on to something here. there are two slots so you can install 2 drives side by side. this would be nice if the case was slim, but its not. the lower drive slot clips with the video card, and there is something sticking out of the front panel which prevents the drive from being installed there. i initially tried rotating the drive 180 degrees but this did not allow any clearance for sata connectors. i was able to use the top slot though, so i didnt have to do any dremel-fu.
i always complain about case design. the only cases that are even remotely well designed are rather large cases. one big issue ive always had was the fact that we still require 5.25" optical drives. id kill for a 3.5" optical format that was generally accepted as a standard that you could buy games/movies/music on. like the 8cm disc format, but with blue-ray-ish data resolution. hell i already have replaced optical media with flash cards for moving large files that im not willing to tie up my wifi network to transfer. ive never trusted optical media as a backup format, and preferred to back up to another hard drive instead. the need to have all these bigger than needed drives in your computer is what keeps cases unnecessarily large. you could argue that buying games, movies, and music on physical media is going the way of the dinosaur, but it aint something i want to see happen (but i dont want a massive box either). power supplies are also rather big, but as powerful chips reduce their power consumption these might get smaller from here. as for cooling, smaller cases have better cooling characteristics, because they can evacuate the entire air volume more quickly than with a large case.
video cards are also getting uncomfortably large. my gtx560 is at least rather light compared to the gtx260 i bought a couple years ago. it seems to use a light aluminum heat sink instead of a heavy copper one it also seems to be double wide for show instead of for cooling, though the plastic fairing may have better controlled airflow to provide better cooling, it didnt seem to form fit the heat sink like in older models. if cards are going to ever get bigger again, id like to see a forward mounting bracket as well to shore up the strain on the card and mobo slot. i find the modern computer makes less use of the bus slots on the mobo anyway. audio, network, ports, drive controllers, etc are all built into the mobo, and usb devices have taken over for most peripherals. so the "mount expansion cards vertically" mentality is loosing its validity. maybe its time to consider parallel mount video cards with the mobo. use a vertical pass through bus slot that physically bolts to the case and mobo through a system of standoffs, multi card setups would allow for stacking of cards through the pass through interface.
tldr: case standards need to be re-thought out.
gripe five: no! i dont want to overclock!
have i ever stated how much i despise the fact that even the lowest grade computer hardware has overclocking features. i tend to have a preference to run hardware within spec for a decent margin for error. watching the hardware utility that came with the mobo and seeing my cpu randomly overclocking itself to 3.7ghz kinda made me cringe. the chip was meant to run at 3.3 ghz and the fact that the mobo defaults allowed the cpu to run 500 mhz above spec is rather non-desired behavior. turns out this is a "feature" of the cpu. the mobo utility lets me overclock the rig on the fly and with no confirmation at all. i accidentally downclocked my cpu to 1.6 ghz while trying to calibrate my fan settings. this concerns me, this should be something you have to do in the bios. what would have happened had i gone up instead of down? not something i want to see answered. it is not a good idea to keep this thing on the system, im going to un-instal it as soon as the system is working the way i want it to. ive already deleted the icons so you have to launch it from the command line.
the mobo has more overclocking settings than i want to see. and i really dont want to use any of them. i know the way cpu manufacturers set clock rates almost arbitrarily, though cpus are graded after being stamped with their final specs, though some down specing happens to meet quotas for higher demand low in chips. the typical overclocker exploits this, though i find overclocking a waste of power, efficiency and operational life. but does every component you buy, even modest stuff like this build, need to have overclocking features?
tldr: overclocking for everyone is a bad idea
the final tldr: computers suck!
-
Just as a "you should've known better", Assrock sucks donkey balls. The mobos are relatively okay if you dont start doing anything fancy with them. (Including tweaking memory timings)
-
fortunately i didnt need to do anything like that. just a little more voltage and its running great. i thought id try new brands. after asus lost all my favor (the last 5 products i bought from them died untimely deaths) and old stalwart msi didnt have any decent boards for that chip/memory setup (the msi microatx didnt support ddr3 1600 without overclocking), and dfi left something to be desired in the mobo i bought to fix my old rig. never liked biostar, gigabyte, or intel boards, too many problems with those back in my system builder days. i thought it was time to try something new. lesson learned, i aint buying another one of these thats for sure. might try a foxconn next time around, the one i installed in a computer i gave to a brother in law still works, considering the makeshift off grid power systems he uses, and his general abuses of hardware, it still runs.
ive already pointed out that i dont like to overclock, the costs and benefits of that is another can of worms entirely.
-
I've been using gigabyte the last few years, I have had absolutely no problems, you should give them another shot.
-
my asrock is fine.
nuke, the overclocking you saw isn't really quite overclocking, it's the CPU's turbo mode kicking in. mine also drops to 1.6 at idle, but as of yet i've been unable to figure out how to turn that off. disabling thermal throttling doesn't seem to have worked. what did you do?
-
im starting to think that buying computer parts is like playing russian roulette. sometimes you get something solid, and sometimes you get a total peice of ****, and who makes it is no variable in the equation. so far the mobo hasnt done anything wrong, aside from being a little too easy to accidentally overclock. it actually did a pretty good job of configuring the memory. i didnt have to screw with the ram at all the timing was instantly recognized, just needed some juice. thing was the ram was on the mobo's qvl, and logic would say that they should work together without error and without tweaking. but no, a little tweaking was still required. regardless its not a problem anymore, computer runs and doesn't crash.
-
my asrock is fine.
nuke, the overclocking you saw isn't really quite overclocking, it's the CPU's turbo mode kicking in. mine also drops to 1.6 at idle, but as of yet i've been unable to figure out how to turn that off. disabling thermal throttling doesn't seem to have worked. what did you do?
i haven't figured out how to turn it off either. i just set the multiplier back to 3.3 and set the power settings to never throttle back. im not sure if its doing what its told though. im going to have to read up on this turboboost feature, ive always been paranoid about automatic overclocking features like this.
-
power requirements skyrocket
I bet that 2500k draws the same, or less, power then a 3.2ghz P4 with HT. In fact, I bet your whole system can idle under 200w.
As for GPUs... Ya, but they've been at this level of power draw for a while now, and they're moving to a smaller transistor size next gen (7k radeons and 600-series nvidia), so we'll see how power does then
first gripe: ram
All ram over 1333 9-9-9 is overclocked, you'll very likely have to set the board accordingly because dispite being a standard, XMP is supported by like, nothing.
second gripe: fan headers
Its micro-atx, they're expecting an HTPC... They are a fool to expect that, but whatever. That's what molex connectors are for now.
third gripe: cpu sockets
Ah... no, nothing to say, but asrock is not exactly the physicly strongest board on the market. Thinner materials, makes me cringe when they bend.
gripe four: screws
Yes, they do. I'll take them over 90% of the toolless crap though.
gripe five: cases suck
Case standards are just fine if you dont buy mini-tower. (see: same logic as mirco/mini-atx above)
gripe five: no! i dont want to overclock!
Thats called SpeedBoost, or TurboBoost. Basicly, if you need more speed the cpu will automaticly downclock 2 of your cores and ramp up the others. This is not overclocking, and the CPU is designed to do it. Go check the CPU's specs on Intel's site, it should list the max 'turbo'. 3.7 sounds about right. Leave it alone, modifying it to run 3.3 at all times -is- overclocking as far as Intel cares.
im starting to think that buying computer parts is like playing russian roulette. sometimes you get something solid, and sometimes you get a total peice of ****, and who makes it is no variable in the equation.
If you're 'in the game', or spend time doing the reaserch, you can easily judge who is best (asrock is not one of them, anywhere, but their products dont just die). They typically cost more (lots more, like, 75-200% over 'low-end' more), and it is always dependant on features vs needs and price.
EDIT: as for the CPU dropping back to 1.6, thats called SpeedStep. When the CPU is idle, or just doesn't need the speed (web browsing), it'll drop back (there are probably 2 or 3 levels to the drop). It also drops the volts with it, allowing the CPU to run colder, and draw less power. The moment you do anything that requires more then 50% of the CPU while its at that speed, it jumps back to full speed. It's a power saving feature, and Intel has gotten good at it.
-
this is the coldest running computer ive had, everything is under 100f most of the time, 91 on the cpu sensor and 88 on the mobo sensor. not sure what the gpu is running, theres another utility for that i still havent installed yet. air out the back is fast and cold, even under load. this computer is fine now that many of the bugs are hammered out. my complaints are still valid though :P
All ram over 1333 9-9-9 is overclocked, you'll very likely have to set the board accordingly because dispite being a standard, XMP is supported by like, nothing.
thats exactly my point. i dont even know what the spec is, i know the documentation says that 1600 is not overclocked and anything higher is. my problem is that the documentation is wrong. if im technically overclocking at 1600 , then why do they say "runs at x, overclocks to y" when what they give for x is already overclocked. you could argue that if the factory says running at 1600 is ok then go ahead and do it, even though according the the base ddr3 specs this is overclocked (this is actually what i usually do). but this is somewhat confusing and unintuitive.
Its micro-atx, they're expecting an HTPC... They are a fool to expect that, but whatever. That's what molex connectors are for now.
il buy that. last micro atx board i bought came with 6 headers, and i was kinda expecting the same. its my bad for not downloading the manual before buying the thing. rather than just molex the fan (its a molex fan with a 3-pin control cable), il use my electronics skillz and build some kind of automatic fan control doohickey from a microcontroller and a thermistor. i went with microatx because i didnt need all the extra bus ports, i figured id add at most 2 cards to it (video and wifi). all those bus ports require connectors (surprisingly expensive), and support components, and those add to the cost. mobo does everything these days and you dont need to add a whole lot to it to have a fairly good rig.
Case standards are just fine if you dont buy mini-tower. (see: same logic as mirco/mini-atx above)
my problem is that all the decently designed cases are behemoths. small cases have better airflow, and most of the air is blowing across the mobo. big cases blow a lot of air over a lot of nothing. most of it goes in one fan and out another without going anywhere near any hot components, unless those components have fans of their own to bring that air to them. theres something to be learned from the way rack mount servers work. there is little or no clearance between mobo and the top of the case some of the air is rammed under the mobo to pick up any heat there. all intakes and exhaust vents follow the same line of airflow, so there are no losses turning corners. a small case is better because there is a higher stuff to empty space ratio, so more stuff gets cooled. i want to see a case like that in a common desktop form factor.
If you're 'in the game', or spend time doing the reaserch, you can easily judge who is best (asrock is not one of them, anywhere, but their products dont just die). They typically cost more (lots more, like, 75-200% over 'low-end' more), and it is always dependant on features vs needs and price.
EDIT: as for the CPU dropping back to 1.6, thats called SpeedStep. When the CPU is idle, or just doesn't need the speed (web browsing), it'll drop back (there are probably 2 or 3 levels to the drop). It also drops the volts with it, allowing the CPU to run colder, and draw less power. The moment you do anything that requires more then 50% of the CPU while its at that speed, it jumps back to full speed. It's a power saving feature, and Intel has gotten good at it.
ive been out of the game for quite awhile. i was a system builder in the p4 era, and back then i knew through endless rotating inventory that certain brands were more solid than others (where i developed profound respect for msi, never had a failed build with an msi board). some brands i learned to despise (biostar, gigabyte), and some i just put up with (intel, always intermittent "driver" issues on a few of what should have been completely identical systems). i try to build a new rig every 2 or 3 years to keep up with the tech, and i do some research (read specs, examining technical drawings, check qvls, sometimes i read manuals). building one computer really doesnt give the same kind of perspective as building a hundred machines a week, and so in my case the russian roulette metaphore still applies. even good hardware can fail for no determinable reason.
-
silly thought: fill that empty space with foam insulation.
-
ive seen large server cases that use foam or cardboard fairings internally to vector airflow. i still find it more beneficial to use a smaller case and forgo the foam.
-
thats exactly my point. i dont even know what the spec is, i know the documentation says that 1600 is not overclocked and anything higher is. my problem is that the documentation is wrong. if im technically overclocking at 1600 , then why do they say "runs at x, overclocks to y" when what they give for x is already overclocked. you could argue that if the factory says running at 1600 is ok then go ahead and do it, even though according the the base ddr3 specs this is overclocked (this is actually what i usually do). but this is somewhat confusing and unintuitive.
Ya, thats pretty much my problem with it too. Basicly, 1333 9-9-9 is the highest JEDEC standard (supported by everything). Companies now just make their ram, bin it, and slap on a label saying 'certified for this!' with an XMP profile to make it easy for people, or at least those with a cpu/mb capable of loading XMP at all. In reality, 99% of people will see no difference at all from 1333 9-9-9 and 1866 9-9-9, but that isn't the point. My solution: Make XMP a real standard and start supporting it already if JEDEC can't keep up with the world.
rather than just molex the fan (its a molex fan with a 3-pin control cable), il use my electronics skillz and build some kind of automatic fan control doohickey from a microcontroller and a thermistor
Go for it, fan contolers are expensive for no reason at all, so if you can make one out of scraps then more power to you. The molex/3-pin is an 'or' option btw, you can run either 100% from the molex, or be able to control it from 3-pin, but the 3-pin is fully capable of powering the fan on its own. (just in case it wasn't clear. Its great that they give you the option, but a better one would be a detachable 3-pin male -> molex passthrough option available for some higher-end fans.)
my problem is that all the decently designed cases are behemoths. small cases have better airflow, and most of the air is blowing across the mobo. big cases blow a lot of air over a lot of nothing. most of it goes in one fan and out another without going anywhere near any hot components, unless those components have fans of their own to bring that air to them. theres something to be learned from the way rack mount servers work. there is little or no clearance between mobo and the top of the case some of the air is rammed under the mobo to pick up any heat there. all intakes and exhaust vents follow the same line of airflow, so there are no losses turning corners. a small case is better because there is a higher stuff to empty space ratio, so more stuff gets cooled. i want to see a case like that in a common desktop form factor.
I would not mind arguing with you about larger cases and where the air goes (http://img822.imageshack.us/img822/4736/img20111208130530.jpg), but I have a full-size ATX board loaded with cards; not what you're looking for at all. As for air direction... it can be controlled, to an extent. As open as my case is, air has one exit: Up and Back. I have a 120mm mounted in the 5.25 bay, and the upper half of the case is a wind tunnel. 85F motherboard and 95F cpu after my overclock (970BE at 4.0, 1.45v, no speedstep) at idle, which I will assume you're at idle based on 'most of the time' (there's no way a 2500k is staying under 100F under load on a stock cooler).
I would like to see some feature-rich mini-towers in the market as I do build a few for other people who have absolutely no need for a massive board (or anything else a mid-tower would offer) and they love that the new little pc takes up half the space of their old Dell while being a ton faster. It would be nice to have a mini-tower that isn't complete aluminum 'what scrap metal do we have today' crap, but the market just isn't there for anyone who cares enough I guess. I can never really like mini-towers though, I tinker with my hardware a lot and it's not very easy to do so in smaller cases. Complete and total lack of cable management options hurt too.
ive seen large server cases that use foam or cardboard fairings internally to vector airflow. i still find it more beneficial to use a smaller case and forgo the foam.
Compared to a 2U, a 1U is complete crap. The 2U will offer better cooling even with the stronger hardware. But everything is going to blades now, so I guess smaller still wins (unless anyone has seen a double-wide blade server to compare or something).
-
kyad what heatsink is that in the pic? i want to get one that doesn't overlap RAM slots.
-
so i built a new computer (il spare you the case, drives and cooling crap):
first gripe: ram
tldr: ram should be built to more precise specs.
second gripe: fan headers
not enough 3 pin fan connectors on the mobo. granted the mobo i got was a reduced sized microatx, it only had headers for a 2 3-pin fans...
tldr: not enough 3-pin fan connectors!
third gripe: cpu sockets
tldr: make cpu sockets only require a sane amount of force to close.
gripe four: screws
this one is short, so no tldr. half the standoffs that came with my mobo weren't tapped correctly and the screws liked to strip. higher grade fasteners please.
gripe five: cases suck
tldr: case standards need to be re-thought out.
gripe five[six]: no! i dont want to overclock!
have i ever stated how much i despise the fact that even the lowest grade computer hardware has overclocking features.
tldr: overclocking for everyone is a bad idea
the final tldr: computers suck!
first: Gah, tell me about it! It's so irritating!
second, third, fourth: dude, didn't you check the specs before you ordered the mobo? Sounds like you ignored the tanstaafl principle and went for a cut-rate product...
fifth: OH HELL YEAH! except what'd that video card arrangement do to airflow... hmmmm...
fifth[sixth]: That's kind of a feature... they're a lot better about implementing it safely than they were ten years ago. I'll just take my 32nm processors and get off your lawn now. ;)
Final tldr; ram is complicated, and my video card is too big, but most of the problems kind of reads like you bought a **** mobo. Sorry.
also...
I would like to see some feature-rich mini-towers in the market as I do build a few for other people who have absolutely no need for a massive board (or anything else a mid-tower would offer) and they love that the new little pc takes up half the space of their old Dell while being a ton faster. It would be nice to have a mini-tower that isn't complete aluminum 'what scrap metal do we have today' crap, but the market just isn't there for anyone who cares enough I guess.
Oh so very yes!
-
kyad what heatsink is that in the pic? i want to get one that doesn't overlap RAM slots.
CM Hyper N520, the stock fans kinda suck, so I replaced them with tricools. I think they're 92mm.
EDIT: The problem isn't that larger cases are bad, or suck (after all, how much can you cram in a mini-tower before that 'optimal' cooling isn't so optimal anymore?), it's all based on feature needs.
And RAM itself isn't complicated (assuming you stick to speed, the 4 main timings, and volts. After that it's insanely complicated), its that motherboards and CPUs won't support the new standard out of the box.
second, third, fourth: dude, didn't you check the specs before you ordered the mobo? Sounds like you ignored the tanstaafl principle and went for a cut-rate product...
2: Is... irrelevent, all things considered, but even cheap boards -should- have atleast 3; CPU/Front/Back (standard, even for the smallest of computers)
3: Is a combination of Intel's socket design, and ASRock's flimsy PCB. One is unavoidable, and the other is hard to find and most people dont even consider how important it is.
4: Has nothing to do with the board. Screws and standoffs are very often neglected, and are never considered in reviews unless they are in the wrong location or something. This isn't a problem for me personally as I literally have a box of the things, but if you aren't prepared for it, bad screws can cripple a build.
-
3: Is a combination of Intel's socket design, and ASRock's flimsy PCB. One is unavoidable, and the other is hard to find and most people dont even consider how important it is.
Fair enough. I've only done amd builds in the last couple years. I defer to your greater experience.
-
3: Is a combination of Intel's socket design, and ASRock's flimsy PCB. One is unavoidable, and the other is hard to find and most people dont even consider how important it is.
Fair enough. I've only done amd builds in the last couple years. I defer to your greater experience.
I mostly do AMD builds too, my only experiance with asrock is what they've got on display at the local Fry's/MicroCenter and a few Llano builds. My feeling on the mater is that, if the board bends putting in a top-down sata plug, something is not as it should be. (and oh do they bend, it scared the crap out of me the first time)
-
I would not mind arguing with you about larger cases and where the air goes (http://img822.imageshack.us/img822/4736/img20111208130530.jpg), but I have a full-size ATX board loaded with cards; not what you're looking for at all. As for air direction... it can be controlled, to an extent. As open as my case is, air has one exit: Up and Back. I have a 120mm mounted in the 5.25 bay, and the upper half of the case is a wind tunnel. 85F motherboard and 95F cpu after my overclock (970BE at 4.0, 1.45v, no speedstep) at idle, which I will assume you're at idle based on 'most of the time' (there's no way a 2500k is staying under 100F under load on a stock cooler).
I would like to see some feature-rich mini-towers in the market as I do build a few for other people who have absolutely no need for a massive board (or anything else a mid-tower would offer) and they love that the new little pc takes up half the space of their old Dell while being a ton faster. It would be nice to have a mini-tower that isn't complete aluminum 'what scrap metal do we have today' crap, but the market just isn't there for anyone who cares enough I guess. I can never really like mini-towers though, I tinker with my hardware a lot and it's not very easy to do so in smaller cases. Complete and total lack of cable management options hurt too.
is that case the dragon-slayer? looks like it. i considered one of them but decided i didnt want to swing the shipping on it (i live on the ass end of nowhere and they gouge our shipping rates).
dont get me started on cable management. the power supply i got had partial modular cabling (24-pin mobo, 2x4 cpu, and an 6+2-pin gpu plug power cables were hard wired in). i add a cable for the video card's second port, at that point everything was somewhat neatly arranged. i plug in a sata bus line that had 3 connectors on it, of course i only needed 2 of them, 3rd one just hangs there (not so bad really). because of my fan debacle i had to attatch the molex line which had 3 molexes and a floppy connector just tp power one fan. granted i only used 2 of the 4 power buses on the thing, but still there are power cables everywhere and loose ends dangling. id like to see a modular cabling system with cascading connectors that let you remove the cable segment past the last device that you are powering. and not leaving any dangling ends to get caught in fans. then again this is no issue that cant be solved with zip ties or electrical tape (i use both).
second, third, fourth: dude, didn't you check the specs before you ordered the mobo? Sounds like you ignored the tanstaafl principle and went for a cut-rate product...
...but most of the problems kind of reads like you bought a **** mobo. Sorry.
"please go on stating the obvious, it fills me with confidence"
of course i bought a ****ty mobo, i had to cut some corners for a couple reasons, first was the price rape on hard drives and second, i decided to get the gtx260 over the gtx250ti. there was a nice $90 msi i wanted to get, but due to budget constraints i got stuck with this piece of ****.
the problems i noted aren't just about this computer (which aside from the now resolved memory voltage glitch, runs perfectly fine) its about computers in general. ive built ****ty computers and ive built rock solid computers, ive built thousands of p4 rigs in dozens of configurations for an oem some years ago. each time i do i notice a whole slew of recurrent problems with a wide range of hardware. some of them are recent, and some of them have been around since the dawn of the pc. most problems can be solved by simply throwing money at them, theres always a better part you can buy. you can do reserch as well, i spent 2 days reading reviews, scanning customer ratings, reading qvls. i over looked a few things (like counting fan headers), but for the most part my research paid off. but once parts arrive and you start building, it aint theory anymore.
-
is that case the dragon-slayer? looks like it.
CM HAF 912.
I may or may not have just shoved all the extra power cables in the spare 5.25 bays on the last mini-tower. :nervous: It was for the best really, and there went 90% of the management right there (the fan's molex was able to reach the 5.25 bays too). Full modular PSUs would be awesome to have, but I'm sure you can imagine the price gouging for such a feature.
At least I can say I have yet to have an ASRock die on me. The PCB might be crap, but I bet it'll last a long time.
-
i have to say that my gigabyte is proving to be a great one, at least when not asked to overclock, but other than that... stability problems = none.
granted the PCB is a bit on the weak side, as there's a significant warping at the point where the cooler bracket mounts.
the mobo is on the oldie side tho.
-
i went over my old rig and removed the dust bunny colonies that have consumed it. the face plate was attached with a hot glue gun (it was damaged one of the times i moved it), so i knifed it off, removed all the shorted out leds that have been disconnected from the power supply for years, thus eliminating a mess of wires. i fixed an issue where the front fan would clip on stuff making an annoying clicking noise, it only required about 8 small washers and clipping the corners off the fan blades to fix. i removed everything from the case that is not essential to its running. some modular cable lines were empty, and i decided that the flash card reader would be better off in the new rig. it took me 2 seconds to pull the thing out of its rack and detach the rails.
an hour later i finally got it installed in the new rig. it had this dual purpose bay which could fit a 5.25" drive or, with an adapter plate and a panel adapter, a 3.5" drive. this in and of itself is ok. it would have been better id they had slotted holes instead of just round ones. i could not get the dcard reader to flush out with the front panel. turns out i attached the adapter plate backwards, it looked symmetrical when i installed it, but i guess it wasnt. its still kinda recessed a bit. but i guess its fine. its sad that how simply the way the holes are stamped out really effects a build.