Test RAM/CPU/GPU etc

[RandBlade]

Sometimes for no apparent reason my Laptop is grinding to a really slow halt, with no apparent reason why. Restarting the machine fixes it, but sometimes it happens on startup.

There is some physical damage to the machine (I dropped it a year ago and needed to replace the HDD) - the main issue being that the screen "hangs" a bit out at the back corner where the case has fallen apart. Put the screen at the wrong angle/let it fall then it can become unplugged internally it appears/colours get messed up ... Windows also makes the noise it makes when you plug in/unplug a USB device.

However I'm wondering if the screen isn't the only issue, the internals should be a really decent spec so there's no reason for it to go so slowly, which it doesn't most of the time. Is there something to download/in Windows to test the CPU/GPU/RAM looking for damage. The BIOS check of RAM on bootup never detects a problem.

[Ominous Gamer]

http://www.memtest.org/

sounds like you may want to take a look at what processes are running as well. Hardware faults (have you run a disk check?) result in hard freezes more often than simple slowness.

[Khendraja'aro]

You should take a look into the event monitor first. If something is wrong you will find it there.

I'd also say that it's not a RAM issue. If it were you'd look at crashing programs and Windows locking up out of the blue.

You can reach the even monitor through the Control Panel, System and Security, "View Event Logs" under Administrative Tools (or simply type 'View Event Logs' into the search box).

Then, on the left, choose Windows Logs and choose the "System" inside this folder. You're looking for everything with a read exclamation mark and the Error Level (obviously). If you can't decipher the messages there, post them here.

[Steely Glint]

You may also wish to consider overheating as a possible cause. As Ominous says, actual physical faults with your hardware will usually result in the computer just outright breaking, not slowing down. But if you're experiencing periodic slowdowns with no evident cause, it may be that either the CPU or GPU is getting too hot, at which point they'll throttle their speed to stop themselves from melting inside the computer (a Bad Thing). The question is, does the computer go back to normal on it's own after a while, or does it always take a reboot to get it back to normal? If you have to reboot, I'd lean more towards a software issue, if it goes back to normal if you simply leave it alone I'd be more inclined to say heat.

[EmperorNorton]

Make sure to regularly dust out the vent for the CPU fan as well.

[RandBlade]

Overheating may be the issue. The Laptop does have a tendency to get VERY hot when I play games and sometimes when I'm not but close the lid. I can be playing a game and suddenly the machine turns itself off without warning and when I reboot the BIOS says it shut down automatically to prevent overheating. In order to prevent this I've bought a large fan which the Laptop sits on top of, this prevents it from shutting down no matter how long I use it for.

I'm amazed frankly that nowadays its still a BIOS issue to prevent overheating that involves unprompted/unwarned shut-downs. You would think Windows should be able to pop up a warning or something before that happens!

I've downloaded some freeware software Coretemp to keep an eye on the issue. Currently the temperature (having been on for hours on the fan, not gaming/active) is mid 50'sC. What temperatures should be warning signs?

Its also only showing temperature for the CPU not the GPU, any way to keep an eye on that?

[Ominous Gamer]

Its also only showing temperature for the CPU not the GPU, any way to keep an eye on that?

This is why Windows can't do it. Each board is different, each expansion add-on is different. It would be impossible for Microsoft to build in something that track something at this level. This is why the BIOS exists, to handle the specifics of the board, including fan speed to temps. There are programs out there that attempt to be the all-in-one solution, but they are always behind the times and only really work after they get the "blueprint" of whatever boards they have time to support. Its only recently that BIOS stuff like OCing has entered the area where it can be controlled via a program inside Windows. On desktop machines motherboards come with a CD loaded with motherboard specific programs that interact only with that motherboard model, and the same is true for graphics cards.

50C should not be a problem, but this is coming from a free program, you should run the computer under load and see how badly the temp climbs.

[Khendraja'aro]

Also, what if you install, say, Linux and haven't yet configured/installed the proper drivers for temperature control? Your laptop CPU would melt down.

Not to mention that this is the fastest version - the BIOS deems temperatures unsafe? Immediate power off. Regardless of the state of the OS that's running (if it's actually running - just imagine a situation where the laptop overheats before the OS is actually loaded!)

[Wraith]

If you need software to torture test your laptop, grab a copy of Prime95. Run that, and watch your CPU temp. Consult your owner's manual for the processor's heat tolerances. It varies, but 90 is usually a pretty dangerous number.

50C seems high for idling. It sounds like heat is your problem.

[Steely Glint]

Currently the temperature (having been on for hours on the fan, not gaming/active) is mid 50'sC. What temperatures should be warning signs?

Uh, I think varies by CPU - they all have their own set temperature at which they will simply shut themselves off to be prevent damage. Also it varies by what temperature the cooling systems inside your laptop can actually cope with. Higher temperatures may result in CPU throttling itself to cool down, which may be the cause of your slowdown problems. But generally 50C is fine, and if it starts getting close to 100c that's probably an indication that you have problems. You will find the CPU temp will very over time depending on CPU usage. You need to have that thing running when you start having speed problems and seeing what kind of results you're getting.

[RandBlade]

This is why Windows can't do it. Each board is different, each expansion add-on is different. It would be impossible for Microsoft to build in something that track something at this level. This is why the BIOS exists, to handle the specifics of the board, including fan speed to temps. There are programs out there that attempt to be the all-in-one solution, but they are always behind the times and only really work after they get the "blueprint" of whatever boards they have time to support. Its only recently that BIOS stuff like OCing has entered the area where it can be controlled via a program inside Windows. On desktop machines motherboards come with a CD loaded with motherboard specific programs that interact only with that motherboard model, and the same is true for graphics cards.

50C should not be a problem, but this is coming from a free program, you should run the computer under load and see how badly the temp climbs.

My issue isn't why doesn't the OS control it, it's why doesn't the OS give any warning? Like with a flat battery a warning is given at 20% then 10%. If say the BIOS turns itself off at say 100C, couldn't Windows give a warning at about 80C or 90C BEFORE (not at) the point it reaches the critical temperature.

[Khendraja'aro]

My issue isn't why doesn't the OS control it, it's why doesn't the OS give any warning? Like with a flat battery a warning is given at 20% then 10%. If say the BIOS turns itself off at say 100C, couldn't Windows give a warning at about 80C or 90C BEFORE (not at) the point it reaches the critical temperature.

That's not up to Windows. Windows cannot know which particular motherboard/CPU combination withstands which temperature.

For that you either need a tool from the manufacturer (who knows the correct temperatures) or a 3rd party program which displays the temperatures (but you have to find out the critical temps for yourself).

Since this is not standardized stuff, it doesn't really fall under the purview of "task the OS has to deliver". If you really want to know why the OS can't do that without your or your manufacturer's help, you might want to look at the procedures you have to run through under Linux. It's actually pretty complicated.

[RandBlade]

That's not up to Windows. Windows cannot know which particular motherboard/CPU combination withstands which temperature.

For that you either need a tool from the manufacturer (who knows the correct temperatures) or a 3rd party program which displays the temperatures (but you have to find out the critical temps for yourself).

Since this is not standardized stuff, it doesn't really fall under the purview of "task the OS has to deliver". If you really want to know why the OS can't do that without your or your manufacturer's help, you might want to look at the procedures you have to run through under Linux. It's actually pretty complicated.

Again it doesn't need to be exact (the BIOS does that in emergencies) but it shouldn't be too difficult to at least attempt a warning. While the exact temperature that's your max varies everyone seems to settle on about 90C+ starting to be concerning so if its able to detect that its getting to 90+ then it should be able to put up a warning.

Downloading Prime95 thanks.

EDIT: Wow within seconds of starting Prime95 it jumped up, will be interested to see how it goes.

[Steely Glint]

Screenshots!

[RandBlade]

Really? Hardly exciting. After half an hour of this test running (with the fan underneath running) the temp quickly ramped up from 55C to 85C but has kind of plateaued there. Tempted to keep it running overnight to see how it goes. System running perfectly snappy too despite the 100% CPU usage, not a hint of slow down yet. The machine is moderately warm to the touch now, when it overheats and shuts down it gets a lot hotter to touch, so something is a lot warmer than that 85 when it does overheat.

This seems like a really good way to test the CPU thanks, but any similar way to stress test the GPU? Or is there some sort of setting in this to be able to stress test that?

[Wraith]

You should only need to run Prime95 for a couple minutes if you're only looking for temperature. Running it overnight is something you do if you want to check the damage that temperature does to your processor.

I don't think Prime95 does GPU. I had a good application that could do it, but can't recall the name right now. A quick search gave me this:

http://www.geeks3d.com/20110407/down...mbustor-2-0-2/

That's not what I used, but I don't see why it wouldn't work for you.

Figure out what your heat tolerances are before running it.

[Aimless]

The slow-downs may be a hard-drive issue coupled with confused software. At the very least you may be able to detect excessive and borked swapping and HD temps, other issues I dunno 'bout. Repair and/or replace HD again.

[RandBlade]

OK its been running non-stop for over an hour now and its not rising any further.



Yellow is temperature, grey is CPU usage (100% for all 4 cores). I do have my external fan on though, I'm tempted to unplug that and see what happens.

EDIT: I did that and within about a minute it went up to 96-97C and stayed there for the next 10 minutes, but didn't reach the point where the Laptop shut itself down (or even slowed at all). Plugged the fan back in and quit the program now and the temperatures dropping back down very quickly. Guessing something else is the problem, trying that GPU stress next.

PS I forgot to mention, there's nothing of interest in the Event Viewer.