PCX
Well-Known Member
- Reaction score
- 134
Disclaimer: Unless you are on-site, there is no reason why you should not run a full hardware diagnostics before every repair. Not doing so is just stupid. Software issues can easily be caused by hardware failure, so why would you immediately start troubleshooting software issues before ruling out hardware failure? Even if you are on-site, you should be running a series of short tests before troubleshooting software issues.
First, you need to be able to test several hard drives at once while running other tests on each individual computer. If you are interested in how to set something up like this you can PM me. I will give you a hint, you can do it with Linux and gsmartcontrol.
1. Unless the computer is an AIO desktop or a laptop that requires complete disassembly to get to the HDD, then immediately take out the hard drive and slave it to your HDD diagnostics box.
2. Start memtest86+ on each computer. Tests need to be run for at least 8 hours or 6 passes. If the memtest fails, then you test each stick of memory separately to find the bad stick. If each stick of memory passes, then one of the slots could be bad. Take the good stick of memory and test that slot. If that slot passes, then more than likely either one of the sticks of memory is no longer able to support dual channel or the motherboard or CPU is no longer able to support dual channel. The quickest way to determine that is to get another set of good tested dual channel sticks of memory and test them in the original computer. If they pass, its the memory, if they fail, its the MOBO or CPU.
3. After each memtest, run a MOBO, CPU, GPU test. I use PC Doctor.
4. PSU: I use a PSU tester as well as a good PSU to swap out for a second opinion. PSU testers are great for quick results, but it wont always catch intermittent issues.
Thats pretty much it. Once you rule out these things, you can start troubleshooting software issue.
If On-Site:
If you are on-site, here is my recommendation.
1. Slave the hard drive to a laptop that is either running Linux from a USB or to a laptop with an additional SATA bay. If 3.5, then it will need an addition power supply.
Note: SMART tests typically cannot be run via USB. USB controllers that support it are still not very accurate.
Do a short test and check for bad sectors. If it passes those first tests, then run an extended test for about 10 minutes. What does not pop in a short test will often times pop in the first few minutes of an extended test. Those will typically be read errors.
2. While running the HDD test, run memtest86+ for one pass. More than likely, if its a memory failure that you are dealing with, you will see it in the first pass. With that being said, in shop environment, do not neglect this test and only do one pass; more subtle errors will be found passed the 5th pass. One error is enough to cause a lot of problems.
3. The alternative to running memtest86+ for one pass is to run PC Doctor test 4 and 5. This will test the MOBO, CPU, GPU, and memory.
4. Linux live cd: This will help you to narrow down and issue to software or hardware. This is not an end all diagnostics, so do not rely completely on this. You have to keep a few things in mind.
- Linux does not run, nor does it crash or react to hardware failure in the same manner as Windows. Typically, Linux can hold up to hardware failures better than Windows.
- Things that are good to test in a Linux environment: network, sound, video, ports, DVD drive, etc.
Final Note: These are not diagnostics, if you say that you use these things in place of a diagnostics, I WILL call you out on it.
chrystaldiskinfo: This is a HDD monitoring tool meant to be used over a period of time. It only displays SMART data and past failures, it does not run diagnostics tests.
chkdsk: This is also not a HDD diagnostics. I am not even sure why anyone would even think that it could be considered a diagnostics.
Windows Memtest: While yes technically speaking this is a diagnostics too, its also technically junk. Use a real memory diagnostics like memtest86+
Load Test: While these are great to run to test stability or check for overheating issues, it is not a diagnostics. It will not in most cases pinpoint failing hardware.
Live Linux CD: While a Linux Live CD is a great way to narrow down certain issues, Linux is not a diagnostics tool in it self, nor does it crash or react to hardware failure in the same manner as Windows. Typically, Linux can hold up to hardware failures better than Windows. This means, that things like memory, CPU and MOBO issues may not show up the same or at all on in a live Linux environment. Also, while booted into a live Linux environment, you are not using the hard drive, thus you will not see any issues unless you actually test it or unless it is severely failing and the live Linux CD has a HDD monitoring system like Ubuntu.
- Things that are good to test in a Linux environment: network, sound, video, ports, DVD drive, etc.
I think I got a great majority of the things I hear the most on the forum . . .
First, you need to be able to test several hard drives at once while running other tests on each individual computer. If you are interested in how to set something up like this you can PM me. I will give you a hint, you can do it with Linux and gsmartcontrol.
1. Unless the computer is an AIO desktop or a laptop that requires complete disassembly to get to the HDD, then immediately take out the hard drive and slave it to your HDD diagnostics box.
2. Start memtest86+ on each computer. Tests need to be run for at least 8 hours or 6 passes. If the memtest fails, then you test each stick of memory separately to find the bad stick. If each stick of memory passes, then one of the slots could be bad. Take the good stick of memory and test that slot. If that slot passes, then more than likely either one of the sticks of memory is no longer able to support dual channel or the motherboard or CPU is no longer able to support dual channel. The quickest way to determine that is to get another set of good tested dual channel sticks of memory and test them in the original computer. If they pass, its the memory, if they fail, its the MOBO or CPU.
3. After each memtest, run a MOBO, CPU, GPU test. I use PC Doctor.
4. PSU: I use a PSU tester as well as a good PSU to swap out for a second opinion. PSU testers are great for quick results, but it wont always catch intermittent issues.
Thats pretty much it. Once you rule out these things, you can start troubleshooting software issue.
If On-Site:
If you are on-site, here is my recommendation.
1. Slave the hard drive to a laptop that is either running Linux from a USB or to a laptop with an additional SATA bay. If 3.5, then it will need an addition power supply.
Note: SMART tests typically cannot be run via USB. USB controllers that support it are still not very accurate.
Do a short test and check for bad sectors. If it passes those first tests, then run an extended test for about 10 minutes. What does not pop in a short test will often times pop in the first few minutes of an extended test. Those will typically be read errors.
2. While running the HDD test, run memtest86+ for one pass. More than likely, if its a memory failure that you are dealing with, you will see it in the first pass. With that being said, in shop environment, do not neglect this test and only do one pass; more subtle errors will be found passed the 5th pass. One error is enough to cause a lot of problems.
3. The alternative to running memtest86+ for one pass is to run PC Doctor test 4 and 5. This will test the MOBO, CPU, GPU, and memory.
4. Linux live cd: This will help you to narrow down and issue to software or hardware. This is not an end all diagnostics, so do not rely completely on this. You have to keep a few things in mind.
- Linux does not run, nor does it crash or react to hardware failure in the same manner as Windows. Typically, Linux can hold up to hardware failures better than Windows.
- Things that are good to test in a Linux environment: network, sound, video, ports, DVD drive, etc.
Final Note: These are not diagnostics, if you say that you use these things in place of a diagnostics, I WILL call you out on it.
chrystaldiskinfo: This is a HDD monitoring tool meant to be used over a period of time. It only displays SMART data and past failures, it does not run diagnostics tests.
chkdsk: This is also not a HDD diagnostics. I am not even sure why anyone would even think that it could be considered a diagnostics.
Windows Memtest: While yes technically speaking this is a diagnostics too, its also technically junk. Use a real memory diagnostics like memtest86+
Load Test: While these are great to run to test stability or check for overheating issues, it is not a diagnostics. It will not in most cases pinpoint failing hardware.
Live Linux CD: While a Linux Live CD is a great way to narrow down certain issues, Linux is not a diagnostics tool in it self, nor does it crash or react to hardware failure in the same manner as Windows. Typically, Linux can hold up to hardware failures better than Windows. This means, that things like memory, CPU and MOBO issues may not show up the same or at all on in a live Linux environment. Also, while booted into a live Linux environment, you are not using the hard drive, thus you will not see any issues unless you actually test it or unless it is severely failing and the live Linux CD has a HDD monitoring system like Ubuntu.
- Things that are good to test in a Linux environment: network, sound, video, ports, DVD drive, etc.
I think I got a great majority of the things I hear the most on the forum . . .
Last edited: