Underclocking a PC

[dagorkan]

Normally my PC should be able to handle almost anything, it's a Dual Core 6420 Intel (2.13 Ghz), 2 GB of good quality RAM, 512 MB X1950 graphics... though M&B, Stalker and Rainbow 6 can still be jerky and turns in late game Civ4 take half a minute to process.

I don't care too much about that, I can always decrease game resolution and use lower quality texture it makes no difference to me. What I do care about is the constant 70+ degree temperature on start-up, constantly loud fan, huge power consumption and now failures to even start up. System standby no longer works (it just switches off instead), every time I start I have to try rebooting 2-4 times before it manages to start loading Windows ("disk read error"). Every few days Windows will lock up for no apparent reason (just while browsing Explorer or the internet). I try a flight sim at normal settings it can only handle it for about 10 minutes the screen switches off ("no signal") and the computer shortly afterward. I'm sure it's linked to the abnormal temperature and resource use I don't even need.

I spent too much money on this to have to replace components after just one year. I don't think I've been demanding at all, but anyway I need at least to be able to access the internet and word processing when I need to which if my graphics card gives out will not be possible.

So I'm thinking strategy games and RPGs should be still be fine if I reduced my performance by 10-20% and the stability, life expectancy and energy savings will not have me worried about the electric heater aka PC under my desk dying any day now without warning.

Anybody been in the same situation, got experience doing this? I know there are a billion guides on how to overclock to some ridiculous amount, what about the other way around?

 

[Fez]

Sounds like you need to sort out your cooling solutions first. Perhaps you won't need to resort to underclocking then?

 

[dagorkan]

Why, is underclocking risky? I don't think there's anything wrong with my fans, I've cleaned them out recently and software tells me they only run at a fraction of the maximum they're capable of. I think the system just demands too much from the CPU/GPU, if everything ran 10% slower I wouldn't notice but it would mean less heat to get rid of.

 

[Destroid]

Sounds like something in your system is defective. I doubt underclocking will fix your problems.

 

[Deleted member 7219]

Sounds like something in your system is defective. I doubt underclocking will fix your problems.

Indeed. Your PC shouldn't be overheating on startup, unless everything in the case is clogged up with dust.

 

[Marsal]

"70+ degree temperature on start-up" is pretty much impossible if you have a working cooler on the CPU.

For clock "manipulation" hit DEL (or whatever it says in the bottom of the screen) to enter BIOS right after you turn on/restart the PC. Don't go on randomly changing values. Take the manual for the motherboard you own (or find one online) and consult it. If you are not tech savvy or don't want to bother too much with what exactly you are doing, you should be able to find the clock adjustment settings even without a manual easily.

Decrease the multiplier by 0.5 or 1 (this is the easiest way of underclocking). You can later bump the multiplier up to get the standard clock or overclock (6420 goes well with 8x if you have decent cooling, even stock cooling should suffice).

Once again, you should not be required to underclock, as Core2Duo CPUs run pretty cool and the problem is most likely in the cooler (or general case ventilation).

EDIT: well turn the fans up to 100% and see what happens (it'll be noisy, but let's see what happens). Underclocking is not dangerous at all.

 

[uhjghvt]

i'd say get a bigger heatsink for your videocard and make sure it's in full contact with good quality thermal grease. when a card is that hot on startup it usually means either the fan is clogged or broken or the heatsink isn't in full contact

edit: wait what is 70+ degrees the cpu or gpu? same thing either way though

 

[Kaiserin]

Sounds like something in your system is defective. I doubt underclocking will fix your problems.

 

[dagorkan]

i'd say get a bigger heatsink for your videocard and make sure it's in full contact with good quality thermal grease. when a card is that hot on startup it usually means either the fan is clogged or broken or the heatsink isn't in full contact

edit: wait what is 70+ degrees the cpu or gpu? same thing either way though

GPU hovers around 70 as I type this message (just firefox and the diagnostic software are open, I've already trimmed start-up services to a minimum, so about 200 MB of pagefile are taken up with routine Windows operations)

Diagnostic says my GPU fan is at 45% potential at that temperature

CPU hovers around 48-52 degrees centigrade and the motherboard around 45.

I don't want to add thermal paste or mess around with fans because what I have should be enough, thermal paste was added when I first installed the CPU and I know the fan was pretty tricky to fit. I don't want to mess around with hardware unless I really need to.

I cleaned out both the CPU and GPU fans on Monday this week.

And yes I think it's the graphics card more than the CPU that creates heat, I'm trying to find out how I can make it run slower but the diagnostic/settings software included with it has no help file. I don't see why the graphics card needs to be at 70 degrees just to look at internet forum pages, none of the software I run is anywhere near what I was told an X1950 PRO could handle.

 

[dagorkan]

I've found some forums that say the default ATI diagnostic software is a piece of shit and is well known to badly optimize the fans so I've downloaded a 3rd party thing called ATI tools. Looks like that will solve the problem.

 

[RadioGnome Invisible]

dagorkan, X1950's normal operating temperature lies somewhere in 70-110 zone under full load, so 70 degrees is pretty much OK. For Core 2 CPUs 48-52 is a bit high, but very much possible if you're using stock cooler with CPU thermal management turned off. Also, 45c from motherboard sensor is quite usual -- modern motherboards generate insane amounts of heat.
Personally, i don't see anything overheating, all components show quite average temperature levels. So, either something is damaged (memory or MB most likely, use Memtest86+ to check), or your PSU is too weak. What kind of PSU is installed in your system?

 

[uhjghvt]

uhh 110 is definitely not normal actually my 1900XT starts beeping at 100

and 70 is too high for no load it should be in the 50s even if the fan is really low

 

[dagorkan]

My PSU is 700 Watt or something overkill like that

The real problem I think is the graphics card shutting down all the time. I did some tweaking with ATI Tray Tools and reduced both GPU core speed and memory speed by 10% and all fan speeds for particular temperatures by 5%. I go into that software's benchmark/test which shows the outline of some kind of sci-fi space ship spinning around and it locks up a few seconds later (screen goes black, fans in overdrive) then my computer switches off automatically.

The ATI Catalyst Control Center did the same thing, it's as if the GPU just can't stand even the slightest demand. The original core and memory speed were exactly default before my underclocking so it doesn't seem that they're damaged or anything.

My GPU temp's down to 65-67 now with the changes.

 

[MetalCraze]

Was it a cheap PSU, what is the manufacturer?
Cuz it seems there are only 3 possibilities:
1. shitty PSU (does not depend on wattage)
2. shitty drivers (well ATI drivers)
3. videocard overheat (less likely - but possible. GPU temperature maybe ok, but memory chips may help you cook some bbq)

Also welcome to the magic world of ATI videocards, crap ATI drivers and shitty support.

 

[mondblut]

I had similar problems a while ago. After several days of constant shutdowns I found out the GPU cooler melted and doesn't spin Replacing it fixed the thing.

 

[szoreny]

Was it a cheap PSU, what is the manufacturer?
Cuz it seems there are only 3 possibilities:
1. shitty PSU (does not depend on wattage)

I second that , a crap PSU can cause you a world of problems, heat included. Check the rep of your brand and model and get something quality if its lacking. Efficiency and stability is more important than advertised wattage - wattages are often exaggerated on lower quality units anyway - 520w off of a quality unit would be more than enough for your set up.

 

[dagorkan]

Was it a cheap PSU, what is the manufacturer?
Cuz it seems there are only 3 possibilities:
1. shitty PSU (does not depend on wattage)
2. shitty drivers (well ATI drivers)
3. videocard overheat (less likely - but possible. GPU temperature maybe ok, but memory chips may help you cook some bbq)

Also welcome to the magic world of ATI videocards, crap ATI drivers and shitty support.

It's an X-POWER 600 W (not 700), Intel ATX 12V ver 2.2 compliant. It cost more than the others, I replaced the basic one which came with my box

 

[RadioGnome Invisible]

Can you borrow another videocard for a test somewhere? Still looks like some hardware failure to me - video memory overheat causes graphical artefacts and lockups, not shutdowns.

*edit*
PSU is good enough - i've used X1950 card with even weaker without any problem.

Check if videocard's additional power cable is connected correctly, and you're using the right 6-pin cord.

 

[Licaon_Kter]

i've build my systems using referenced hardware tested on SPCR, try to read a bit there about air flow, cases, cooling and stuff in the articles and on the forum.

BTW they're using a X1950 to test video card coolers since it's known as a "warm" card :D

 

[LCJr.]

I'm surprised no one has suggested reformatting.

 

[dagorkan]

Reformatting what? I'm defragging my hard drive and it really needed it but I doubt it's the cause of my problems. However I don't know what the temp of my hard drive is, I don't think they put a sensor for it just the motherboard generally.

 

[Carrie Underwood]

Don't listen to the hate. ATI's drivers are fine unless you doing something wierd or running a wierd configuration (or have an integrated card), then they might cause trouble. Same thing with NVIDIA. You don't want to go search out those "unofficial" drivers. They suck. Hard.

You said you cleaned it out, so I'm assuming you bought a can of air and sprayed the hell out of it.

The PSU isn't your problem here, unless the fan on it is busted.

How are your fans organized and how many do you have? The second computer I built had a side air duct on the case. I put a fan on it and didn't pay attention to which side I mounted. Needless to say, after a while, the thing started overheating like hell. The CPU fan and the side fan were blowing straight at each other, and both were encased in the duct. I'm lucky the thing didn't fry.

Do you have a pci exhaust fan? There are exhaust fans for only a few bucks that fit into a pci slot. They usually don't do anything except in the most dire of situations, which this seems to be.

You might have put the thermal paste on wrong, used a brand other than Arctic Silver, or kept the paste that came with the heatsink. Alternatively (or addition to that), your fan might be busted. Those stock fans have a high rate of failure (if you used one, otherwise you just got unlucky).

I think it's either the paste problem or the fan problem. Probably both. Unfortunately, the only way to rectify this is to take off the CPU heatsink and fan. Yes, it sucks, but keep in mind that the stock fan that comes with AMD/Intel CPUs is the hardest fucking thing in the world to get on. AFTER MARKET FANS ARE NOT LIKE THIS. Usually. If you get a good brand (Arctic Cooling, Thermaltake, etc.). I had the overheating problem after using my current computer (Athlon 64 X2). First AMD I built in a while, so when I did, I was horrified that my motherboard had to bend for me to get the heatsink on. I bought a Gigabyte fan (not a big cooling company of course, but a good company all around). Installation was smooth as butter.

Again, check that the case fans are blowing in the right direction and are moving enough air before you run out and buy a new fan and heatsink. If you kept the fans that came with your case then replace them with some Arctic Cooling fans. It sounds like you need as much cooling as you can get.

Don't underclock. There's no reason to underclock, unless you previously overclocked.


Oh yea, how's your motherboard looking? You've got a heatsink on both bridges, right?

 

[dagorkan]

I replace my expensive 600W super-special intel-compatible PSU with the 400W thing I got pre-installed with the box and whaddaya know, boots perfectly, GPU temp down to 59 celsius, CPU down to 43 celsius and motherboard is at 39 degrees. (just went down to 40 and 37 now)

My motherboard manual says minimum 550 Watts are needed but I have my hard-drive, graphics card, about six USB devices plugged and it makes less noise. The only problem is that there is apparently no sensor for that basic power unit so I'm getting warning signs saying it's failed (0 RPM).

Now I just hope my warranty is still valid.

 

[MetalCraze]

PSU as suspected. Buy 600W or so from Chieftec. They still release pretty hiQ PSUs.

 

[dagorkan]

You said you cleaned it out, so I'm assuming you bought a can of air and sprayed the hell out of it.

Nope, I just used a brush and a hoover lol

How are your fans organized and how many do you have? The second computer I built had a side air duct on the case. I put a fan on it and didn't pay attention to which side I mounted. Needless to say, after a while, the thing started overheating like hell. The CPU fan and the side fan were blowing straight at each other, and both were encased in the duct. I'm lucky the thing didn't fry.

I have two fans. One is integrated in the power unit and one is fitted over the CPU, that's it. I got two 'optional fans' with my motherboard which are meant to go over the heat dissipator things (look like small radiators), but the instructions say not to use them along with an 'active' CPU cooler because they conflict. Not sure if it would be better to get rid of the Intel base fan and use two fans on the side plus a 'passive cooler' on the CPU, I used the Intel included cover fan since the optional ones are smaller and looked less powerful.

Do you have a pci exhaust fan? There are exhaust fans for only a few bucks that fit into a pci slot. They usually don't do anything except in the most dire of situations, which this seems to be.

No, sounds like they could be useful

You might have put the thermal paste on wrong, used a brand other than Arctic Silver, or kept the paste that came with the heatsink. Alternatively (or addition to that), your fan might be busted. Those stock fans have a high rate of failure (if you used one, otherwise you just got unlucky).

I got the guy who fixed my PC when I first got it and sold me Windows to put some paste on, not sure which brand. Anyway my CPU is now down to 40-44 celsius which seems more normal.

I think it's either the paste problem or the fan problem. Probably both. Unfortunately, the only way to rectify this is to take off the CPU heatsink and fan. Yes, it sucks, but keep in mind that the stock fan that comes with AMD/Intel CPUs is the hardest fucking thing in the world to get on. AFTER MARKET FANS ARE NOT LIKE THIS. Usually. If you get a good brand (Arctic Cooling, Thermaltake, etc.). I had the overheating problem after using my current computer (Athlon 64 X2). First AMD I built in a while, so when I did, I was horrified that my motherboard had to bend for me to get the heatsink on. I bought a Gigabyte fan (not a big cooling company of course, but a good company all around). Installation was smooth as butter.

Are water coolers a good idea or not, if you don't overclock?

Oh yea, how's your motherboard looking? You've got a heatsink on both bridges, right?

My motherboard is says it has it's own unique 'cooling solution' I think it dissipates heat pretty much everywhere.

Here's the alternate fan setup they suggest: