Stability issue I'm having difficulty with

[mikeymikec]

The computer is an Ivy Bridge generation build, ASUS P8Z77-M board, Samsung 850 PRO SSD, Core i5-3450, 8GB DDR3 RAM, integrated graphics.

I built it in 2012. It was upgraded from Win8 to Win10 somewhere along the way. The HDD it originally had died about 6 months ago, so I put the SSD in and clean installed Win10.

The issue began as intermittent: Freezing during workload, then one day it wouldn't boot. Before I had a chance to visit and take a look at it, it righted itself. I advised that the user's data should be backed up straight away just in case.

It started giving problems again a few weeks ago. I found three problems:

1 - Norton had been crashing a lot (could be a symptom of a bigger problem, but it was cited as the cause of a lot of BSODs, so I ran the Norton Removal Tool on it).
2 - In the BIOS an overclock had been set (CPU and RAM). The person I built the computer for said they knew nothing about that and didn't know what overclocking was.
3 - Some crapware was on the machine. I ran MalwareBytes as well as autoruns to check the machine over in this respect.

After finding the overclock (which was the last issue I found) and correcting it back to stock settings, the machine gave no problems.

While I was away, he rang me again, stability issues again. I've got the machine here now and it was saying that it couldn't find the boot device, and even if I specified the SSD in the BIOS as a temporary boot device, it still wouldn't boot.

The RAM was overclocked again to the same setting as before. This time I just loaded optimised defaults but it hasn't changed the main boot problem.

Booting from a Win10 DVD, the startup repair got stuck on 'diagnosing this pc', as there was no HDD LED activity I pressed the reset button. I then used the DVD to get a command prompt which showed files on C drive, but the first chkdsk /f /v /r failed with "an unspecified error occurred". I immediately attempted to run it again but it hasn't even shown the first line of chkdsk output in the time taken to write this post.

My first thought is that the SSD has failed aside from the weird overclock issue and I'll probably pursue that to some kind of conclusion first. Any suggestions would be much appreciated.

 

[Ketchup]

Put a good drive in there and see what happens. This is why I am "iffy" on putting together a machine for someone, as many motherboards offer overclocking options, and of course they are going to say "I didn't do that." I hope they aren't trying to take your time with any kind of "warranty service" on this one.

 

[VirtualLarry]

Btw, once you get it working, it may be time to password-protect the BIOS setup.

 

[C1]

Just for giggles, do a command line check to verify that Trim is enabled (or at least that Win 10 thinks it is).

If that's okay, I would image off his SDD, then wipe it clear and image his stuff back & see what happens.

If there are still issues then you could also now swap the external drive used to hold the image copy for test purposes and give that a try or do all kinds of other possibilities like run a Win OS or Linux from a USB stick etc. to try to reproduce the issue in the shop.

Before you release the machine try the following procedure as a check on the client. Reset the BIOS and set a BIOS password for it before you give the machine back. (Bury the BIOS password in a text file along with directions on how to remove it, somewhere in Windows stuff where it wont be noticed, probably somewhere in documents & settings would be good enough.) If you get a phone call from the owner bitch'n, you know that he is mess'n around - he should not be having to. Otherwise, call back in a month to tell him about the password & where the file is so that he can zilch the password if or when he wants (or set auto updates to pop the file open in a month to alert the client). This verifies user mess'n around.

 

[DAPUNISHER]

My first thought is that the SSD has failed aside from the weird overclock issue and I'll probably pursue that to some kind of conclusion first. Any suggestions would be much appreciated.

Good place to start. Add it to one of your systems and see if you can access it from there. Run error checking, disinfect, etc. If everything goes OK, try a different sata port, power, and data cable when you put it back, just because. And password protect the bios this time. Yeah they can reset it, but they cannot put back the one you used. Then the odds they are lying about monkeying around in there go up exponentially.

If the drive died, I would start asking questions to see if it is bad luck or something they are doing is contributing to killing them.

 

[DAPUNISHER]

Just for giggles, do a command line check to verify that Trim is enabled (or at least that Win 10 thinks it is).

If that's okay, I would image off his SDD, then wipe it clear and image his stuff back & see what happens.

If there are still issues then you could also now swap the external drive used to hold the image copy for test purposes and give that a try or do all kinds of other possibilities like run a Win OS or Linux from a USB stick etc. to try to reproduce the issue in the shop.

Before you release the machine try the following procedure as a check on the client. Reset the BIOS and set a BIOS password for it before you give the machine back. (Bury the BIOS password in a text file along with directions on how to remove it, somewhere in Windows stuff where it wont be noticed, probably somewhere in documents & settings would be good enough.) If you get a phone call from the owner bitch'n, you know that he is mess'n around - he should not be having to. Otherwise, call back in a month to tell him about the password & where the file is so that he can zilch the password if or when he wants (or set auto updates to pop the file open in a month to alert the client). This verifies user mess'n around.

I was still composing and typing when your post went. Good advice.

 

[mikeymikec]

The best I managed with resurrecting the installation was that instead of 'please insert boot device', I got a Windows BSOD However, bootrec couldn't find a Windows installation, and until I used diskpart to assign a letter to the 100MB boot partition, it wasn't represented at the command prompt (I tried bootrec before and after this). Furthermore, the BIOS had forgotten its UEFI platform keys so I had to reinstate those.

I've gone for a BIOS password (though interestingly it is required to specify a temporary boot device, not that I see that as being a problem) and a clean Win10 install. Both went without incident, I ran a full chkdsk on the main volume then I zipped up the customer's data and transferred the archive file as well as copying all that data in its original form over to the SSD. I then unzipped it a few times and ran archive checks on it, all of which passed. SSD I/O performance was within expectations during those file transfers and integrity checks (consistent 380MB/sec read for example).

No errors in the event log that point to iffy I/O. I wish I had come up with something more conclusive, and while the RAM overclock could have potentially resulted in some file system trashing, it does seem a bit of a stretch.

One question - I can't find an Intel AHCI driver for Ivy Bridge on Win10 64. Would you just run with the standard driver?

 

[RLGL]

What is wrong with the drivers from Asus? https://www.asus.com/Motherboards/P8Z77M/HelpDesk_Download/
The latest bios for the board, Chipset driver and Intel MME driver All on that page.

 

[mikeymikec]

No AHCI driver. IMEI gets downloaded via WU automatically. The chipset driver does nothing for AHCI.

 

[UsandThem]

No AHCI driver. IMEI gets downloaded via WU automatically. The chipset driver does nothing for AHCI.

You could install Intel Rapid Storage or use one of the stand-alone AHCI drivers from here:

http://www.win-raid.com/t2f23-Intel-RST-RSTe-Drivers-newest-v-WHQL-v-Beta.html

http://www.win-raid.com/t11f23-Modded-Intel-AHCI-and-RAID-Drivers-digitally-signed.html

My SSD performance increased slightly going from RST to a stand-alone chipset driver.

 

[RLGL]

No AHCI driver. IMEI gets downloaded via WU automatically. The chipset driver does nothing for AHCI.

So what you are telling me is that Asus does NOT provide all the drivers to run the product! If that is so why when I am only using the provided drivers , my board works and I am NOT having issues.

 

[mikeymikec]

So what you are telling me is that Asus does NOT provide all the drivers to run the product! If that is so why when I am only using the provided drivers , my board works and I am NOT having issues.

I'm not sure why you're capitalising words HERE and THERE but if it's to convey emotion it's not really helping. Asus doesn't supply a Win10 Ivy Bridge AHCI driver because Intel (at least from what I've seen, I hoped to be mistaken) does not. Possibly because the Win10 AHCI driver is adequate for an older chipset than Win10?

The board is from 2012 and Win10 was released in 2015. While it would be nice if Asus kept their driver sets up to date when new operating systems came along, I don't necessarily expect it, as long as I can retrieve optimal drivers when the stock ones are suboptimal.

 

[VirtualLarry]

This is common, actually. Recent Windows releases that are releases AFTER a popular board, often contain the necessary drives OOTB for the mobo in question.

For a few examples:
1) An ECS A55 FM1 board, certified for Win8. Works perfectly fine in Windows 10, all drivers accounted for OOTB after going online for the video driver.
2) My ASRock Skylake boards with Intel LAN. With Win7, and Win10 1511, I had to download and install the ProSet LAN drivers from Intel. Win10 1607 has them built-in.

 

[mikeymikec]

I personally suspect that the reason that there isn't an Ivy Bridge AHCI driver from Intel is because Intel simply can't be arsed* to make one (combined with there being no show-stopping argument to make one, e.g. data corruption occurs because the default driver can't handle something odd about the chipset). I would be extremely surprised if the default AHCI driver is made optimal for all recent-ish chipsets, and think of the risk of making changes to the AHCI driver and rolling it out on Windows Update.

* - "cost-effective", yadda yadda yadda.

 

[mikeymikec]

Update - no problems and three weeks have gone by.