Old PC, dying or dead?

gupott

gupott

gupott

Associate
Joined
26 May 2008
Posts
252
Hi all,

I turned my main PC on yesterday, or I tried to. Its an older home build PC circa 2013 with a 3570k Intel CPU, Gigabytee Z77-D3H mobo, 8gb patriot ram and a Pallet 970 GPU, running windows 10. The only reason it has not been upgraded has been adding 500gb Samsung pro SSD and it still felt fairly nippy.

Anyway, on turning it on last night I kick-started a sequence of events which I have not been able to sort out yet. Initially the PC was on but nothing was on screen. After a few restarts and a delve into the bios I realised I could connect the HDMI cable to the main PC output and get the image back, so GPU then? I then made a mistake and disabled the mobo screen output in the bios and tried the GPU again. I could not get the GPU to work and couldn't get back into the bios. Mistake 2, to fix the first I pulled the cmos battery, which meant I could get back into the bios and fix mistake 1. However ever since then I cannot find a set of stable settings to run the PC, and I cannot find a way to put the bios back to default. No matter whether I try last know good configuration, a close approximation of the default settings or me playing around under the hood, the same thing happens. The PC will boot, post but crashes on the windows start screen. I am now wondering whether the Corsair PSU or mobo are at fault. It just turns off very abrubtly as windows starts to load. I can however boot into safe mode. Could this be a software issue, or is it likely hardware?

Anyway, I had hoped to last until the summer before buying again, does anyone have any thoughts or tips to try?

Luckily I think all my stuff is backed up on Onedrive and my NAS, although recreating the regime on any new PC is already making me nervous! lol.

Cheers
 
Last edited:
BIOS/CMOS battery could be near empty and is at least cheap to change.
Graphics card could be dying if it originally worked with CPU's integrated.
Of course motherboard or memory could be failing. (try with one DIMM at time)
And you didn't specify precise PSU model so that could be some cheaper not designed to last model.
 
EsaT said:
BIOS/CMOS battery could be near empty and is at least cheap to change.
Graphics card could be dying if it originally worked with CPU's integrated.
Of course motherboard or memory could be failing. (try with one DIMM at time)
And you didn't specify precise PSU model so that could be some cheaper not designed to last model.
Click to expand...

Thanks for replying.

1) CMOS battery - I can probably change, but not sure what difference it will make.
2) Graphics card - agreed, but i have completely removed now and problem still persists
3) Mobo or memory failing - agreed, I will trying pulling one DIMM at a time.
4) PSU - it wasnt a generic part, it is a corsair tx750w
 
CMOS battery problem can cause some crazy symptoms.
In my first ever PC it caused floppy drive to disapppear at first, but found it at next boot, while refusing to find HDD.
(booting only from floppy)
 
1) reset cmos +/- change cmos battery
2) ensure only the basic core components are connected (ie cpu, mobo, 1 stick of ram only)
3) plug in monitor and keyboard
4) boot up, run memtest

if passes into bios, then you can be sure it's something else.

then, rebooting after each step:
5) second stick of ram, run memtest with both sticks
6) gpu
7) one hdd (preferably a spare one)

if all okay:
8) install windows
 
I tried both sticks initially but PC still crashes on windows splash screen. I have just been going through memtest. I did actually have two errors, right at the end of memtest on the 4th pass. so 47/48 tests passed (97%). Bits in error was 2. Lowest error address was 4992MB and highest 7791mb. I assume that means one stick is faulty?!? Back to try one again.

I am not sure I can install windows, I only have a win 7 disk which has been upgraded to 8 and then 10. hmmm.
 
Last edited:
rosscouk said:
I tried both sticks initially but PC still crashes on windows splash screen. I have just been going through memtest. Currently on 4th pass with both sticks in with no erros (just ram may be vulnerable to high frequency row hammer bit flips).

I am not sure I can install windows, I only have a win 7 disk which has been upgraded to 8 and then 10. hmmm.
Click to expand...
If you've done 1-5, and still crashing with a current install of Windows...
Then can consider a fresh reinstall of windows. You can download win 10 from Microsoft website and install via usb stick
 
rosscouk said:
I did actually have two errors, right at the end of memtest on the 4th pass. so 47/48 tests passed (97%). Bits in error was 2. Lowest error address was 4992MB and highest 7791mb. I assume that means one stick is faulty?!? Back to try one again.
Click to expand...
Are you running the ram at stock speeds? Also what voltage?
 
Thanks tamzzy. I think the ram was at stock (9-9-9-24 @ 1.5v) but I did overclock years ago. Its at stock now but it could have been o/c, and looking at my dog-eared notes in the mobo manual the speeds may have been 1866 up from 1600 but still at 1.5v.

I didnt know you could d/l win 10 from microsoft, I am off to try that now.
 
rosscouk said:
I think the ram was at stock (9-9-9-24 @ 1.5v) but I did overclock years ago. Its at stock now but it could have been o/c, and looking at my dog-eared notes in the mobo manual the speeds may have been 1866 up from 1600 but still at 1.5v.
Click to expand...
run the ram at stock speeds 1600mhz cl9-9-9-24 and bump the voltage up to 1.55 or 1.60v just to make sure it's not the ram causing the issues. then run memtest again and see if you're getting any errors
(1.65v is the max - but you don't need that kind of voltage to run @ 1600mhz)
 
Not just a windows update playing up? Mine has just updated and for the first time for 6 months it had to try several times before it completely installed
 
tamzzy said:
run the ram at stock speeds 1600mhz cl9-9-9-24 and bump the voltage up to 1.55 or 1.60v just to make sure it's not the ram causing the issues. then run memtest again and see if you're getting any errors
(1.65v is the max - but you don't need that kind of voltage to run @ 1600mhz)
Click to expand...

I checked and it was already at 1.58, I bumped it to 1.6v.
 
Well, I was trying to create a windows 10 usb on my ancient laptop using the microsoft media creation tool, but that crashes, at around 90%. I have tried it twice, not a great start.

Anyway, I hadn't actually run a memory test again so I have just started that again with the higher ram voltage. Prior to that though, I did try and did a system restore which, interestingly I could do on the second try. I could even boot into windows. I quickly went across and stopped windows updating and then I just let it sit there for a few minutes. Then it crashed! Not sure why I could get further this time.

So, memtest first and try a different memory stick to create a win 10 usb.
 
So, I have just run memtest as a pair and I get errors but not when the test is run on a single stick. The errors always seem to come in test 13 with the lowest error address 4992mb and the highest 7791mb.

So given the fact that the sticks test okay when tested individually but not as a pair, what does this signify? Any I am not sure what steps to take next. Any ideas?
 
One thing to check would be if it's one memory channel which is giving errors.
I assume you tested DIMMs individually in same slot, so try them one at a time in another slot.
 
rosscouk said:
So given the fact that the sticks test okay when tested individually but not as a pair, what does this signify? Any I am not sure what steps to take next. Any ideas?
Click to expand...
test it as a pair, in the correct slots (as per your mobo manual). normally its A1B1 - but check your mobo manual to confirm
if it fails again, then as esat says, it might be one of the memory channels - then try the other channel (ie, the other slots)
if those fail, then it's either: (1) mobo issue, ie physical damage to the ram slots/circuitry/cpu pins, or (2) cpu IMC issue
 
Bit of an update. So, I have only ever used two modules in this build, 2 x 4gb. They have been used in slots 1 and 2 as per the manual. I have test both dimms in both slots 1 and 2 and individually they return no errors. But if you try them together, no dice, you get a few errors as I have described above. I can't try slots 3 and 4, as 4 is obscured by the CPU cooler.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom