|
When you teach a bunch of programmers to make puns do they become funny?
Nope! 😛
If you can't laugh at yourself - ask me and I will do it for you.
|
|
|
|
|
My entertainment laptop which used to be my mainstream machine, an Acer Aspire 19" running Win 10 Pro, had a problem of an inadequately cooled GPU. In the summer it would frequently just lock-up - frozen screen, nothing working, so a power switch long press needed. It always happened when the GPU was working hard, but the CPU temperature never went above 78°C - well within bounds for a Core i7 - and usually ran at around 65°C.
Now, it is happening at modest room temperatures when the GPU should be doing nothing - static screen, no apps running, just either Edge or Chrome streaming music over the headphone socket, with HDMI disabled.
I use Malwarebytes Pro, free courtesy of my bank, and when I installed their Browser Guard it found a couple of dodgy extensions in both browsers, which it deleted, and everything was fine for that evening, but the next day the problem was back.
It rather looks as if I've been hacked. Unfortunately, being an older machine, Task Manager does not show GPU activity like it does on my newer laptop.
Does the CP hivemind know if there's anyway of finding what is accessing the GPU - and, more importantly - how to stop it?
Pretty please - it's driving me crazy! 
|
|
|
|
|
GPU heating ... doesn't that usually mean cryptojacking?
You could try booting to safe mode and see if it goes away - if it does, then something is using the GPU and crypto is a good bet. Reinstalling the OS is the best way I know to get rid of nasties like that ...
Or it could be that the heat paste on the GPU (do they use that on lappies?) is breaking down and it just doesn't have enough cooling any more!
"I have no idea what I did, but I'm taking full credit for it." - ThisOldTony
"Common sense is so rare these days, it should be classified as a super power" - Random T-shirt
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
OriginalGriff wrote: GPU heating ... doesn't that usually mean cryptojacking?
That's what I'm afraid of. earlier in the year I checked out a trial version of a graphic equalizer app. The trial version said it would use some of the processor/network capacity for VPN support elsewhere, when the machine was idle. I paid for the pro version, which meant they didn't do that. I am now deeply suspicious of what was left behind when I removed that app a couple of months ago.
OriginalGriff wrote: Reinstalling the OS is the best way I know to get rid of nasties like that ...
A bit of a problem, there. The laptop originally had Win 7 Home. I upgraded to Pro on a MSDN licence I had at that time. I then used the free upgrade to Win 10 Pro when it was announced. Incidentaly, that was when the GPU over-heating started, but only when the machine was really stressed. A couple of months ago I went to the Windows Upgrade site, and managed to replicate the process Win 7 Home to Win 10 Pro in a single swoop. Whether I can repeat this process remains to be seen.
OriginalGriff wrote: Or it could be that the heat paste on the GPU (do they use that on lappies?) is breaking down and it just doesn't have enough cooling any more!
Not usually. The usual way is to have a thick copper bar as a heat sink which runs over the CPU and GPU and terminates in a finned jobbie cooled by a small fan. The fan is definitely working right, and the machine is sitting on the best multi-fan cooling pad I could find.
|
|
|
|
|
Sounds like a silly question, but do you sleep, hibernate, shutdown, or restart your lappie?
I know that shutdown and restart don't do what you might expect: shutdown hibernates the kernel to make startup faster (which preserves dodgy drivers in situ) where restart rebuilds the whole kernel from a "blank template".
"I have no idea what I did, but I'm taking full credit for it." - ThisOldTony
"Common sense is so rare these days, it should be classified as a super power" - Random T-shirt
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
Ah ha! I always just close the lid to hibernate, but it does get a regular cold boot after it has crashed, been forcibly powered off, and left to cool, so I presume that counts as a full restart. Of course, I don't know the kernel state after the lockup, as stuff is still running - the machine doesn't seem to cool down after the lockdown.
|
|
|
|
|
The plot thickens! The machine can still crash sitting idle, doing nothing, with a blank screen saver showing.
I guess I bin hacked!
My latest experiment, I found a tvnserver.exe in the startup folder, but I uninstalled Tight VNC a several months ago when I went to Win 10 Pro and could use MS RDP (sheer idleness so I could change the music with my Windows tablet from my armchair). Anyway, it is now disabled, and am currently awaiting the crash...
|
|
|
|
|
yikes, re-installing the OS is a major pain in the butt and usually not necessary to fix PC problems.
One effective way of discouraging hackers is to setup am account to access windows and have it password protected. Then change the password every day. It's a little job, but it really messes with the hackers 
Another way is to not have your computer online all the time.
If you are set on scanning, try the spybot search&destroy engine. It's pretty good, and it does effectively remove a lot of hacker hooks, so running it every day can really cut down on unwanted code running on your pc.
|
|
|
|
|
Not really: for me it's about 1/2 hour, plus updates to apply - I create a clean image when first installed with all needed apps, and that's how long it takes to do a test restore.
It helps if you keep all data on a different drive as well ...
Virus, malware, hardware failure, ransomware: stuff the lot and reload clean!
"I have no idea what I did, but I'm taking full credit for it." - ThisOldTony
"Common sense is so rare these days, it should be classified as a super power" - Random T-shirt
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
The problem with that approach is that you then have a computer with an out-dated build online updating which makes it highly vulnerable to bots looking for computers that insecure to attack. The only way to keep that remotely secure is to have a windows cd you can fresh install from and all the updates on dvd or something so that you can install them all while the computer is kept offline until it's security build is fully up to date. That is a tricky measure to implement since you tend to only download what you need for your hardware build, so if that changes, all the updates that go with that new build are now slightly different than they were for the old build. In other words, the most secure ways of re-staging a PC can get pretty complicated over time.
You are very correct about having the data on a different drive. I have never regretted doing that.
|
|
|
|
|
I would start at sysinternals, process monitor.
Windows Sysinternals | Microsoft Docs[^]
I would also consider putting in another hard drive (SSD) and testing with a new install of Windows. I will avoid the obligatory suggestion of installing that other OS. 
>64
If you can keep your head while those about you are losing theirs, perhaps you don't understand the situation.
|
|
|
|
|
I am certainly considering upgrading to a SSD, as at present it is just a 7200 rpm hybrid, but for it's main purpose of music streaming, it hardly seems worth all those beer tokens.
Thanks for the Sysinternal prompt - I am just now going to give it a whirl.
|
|
|
|
|
Maybe it doesn't like the music you are streaming.
Seriously, how old is the laptop? I hate to say it but perhaps the cpu is just dying.
I have had that happen to me.
Hope you figure it out.
|
|
|
|
|
Have you checked that its airways are clear? Laptops have a habit of building a lovely dustbunny just downstream of the fan, against the fins it is trying to cool.
In most of them it's a matter of around 10 screws to remove to get access, but in some cases as high as 47! (Lenovo ThinkPad X300, I'm looking at you)
Maintenance manuals are generally available online.
Open it up, work through with a soft paintbrush and vacuum. I'm not a fan of canned air - it just tends to relocate the dust to somewhere worse.
Cheers,
Peter
Software rusts. Simon Stephenson, ca 1994. So does this signature. me, 2012
|
|
|
|
|
Peter_in_2780 wrote: I'm not a fan of canned air - it just tends to relocate the dust to somewhere worse.
I'd agree with that - normally into my eyes at some point.
"I have no idea what I did, but I'm taking full credit for it." - ThisOldTony
"Common sense is so rare these days, it should be classified as a super power" - Random T-shirt
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
All been done and there is good - but very very hot - airflow from the port. One minor curiosity - when the machine locks up, the fan continues to run, but I guess that's to be expected. Also the screen stays showing what was happening when it died.
|
|
|
|
|
Are you sure it's not something like a bad/clogged fan going to your GPU?
Real programmers use butterflies
|
|
|
|
|
The heat sink is a thick copper bar, as explained above, and it handles both CPU and GPU. If it were a blocked/failing fan then both would overheat.
|
|
|
|
|
Still could be the paste bonding the GPU to the sink went dry/bad.
That used to happen to me on old CPUs a lot. If the GPU routinely ran hotter than the CPU then I could see the paste decaying at different rates.
Anyway, just a thought. Most of the time I face catastrophic issues with my PCs it's either a hardware problem or a Windows update.
Real programmers use butterflies
|
|
|
|
|
|
Chris C-B wrote: switch long press needed. It always happened when the GPU was working hard, but the CPU temperature never went above 78°C - well within bounds for a Core i7 - and usually ran at around 65°C.
Just noting of course that the CPU and GPU are two different things. And in two different places.
Brief search also suggests that GPUs have a higher failure rate than CPUs. So if the GPU is on the edge of a failure then running hotter but not hot, might push it over the edge.
|
|
|
|
|
I am fully aware of that, but I had no way of measuring the GPU temperature. They are, however, on the same cooling bar, the thick copper bar heat sink that snakes from one to the other.
|
|
|
|
|
I am replying to my own message, having, I hope, resolved the issue. I stopped tvnserver.exe loading at startup, and all now appears to be well - it ran last night quite happily from late afternoon to 23:00. I am not suggesting that tvnserver.exe was the direct cause of the problem, but it may well have been letting something else in that was burning up the GPU.
|
|
|
|
|
OK, this place has a lot of good articles to read, and there was one that wanted to make a comment on, so I sign in, which then triggers an E-mail sent to me that has a link that logs me in. So I go back to the original article that I had wanted to comment on, but that page has me as not being logged in! Even if I copy & paste the URL from the original tab to the tab that got opened with the E-mail link, it still comes up as me not being logged in, and when I log in from there, to paraphrase Lou Costello, we're back again at getting an E-mail link (i.e., 1st base). 
|
|
|
|
|
What did you select as your cookie settings? If it's "everything off" then it can't store your login status and that may be something to do with it.
"I have no idea what I did, but I'm taking full credit for it." - ThisOldTony
"Common sense is so rare these days, it should be classified as a super power" - Random T-shirt
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
