Error Verifying Data in RAM once again.

[jacek@pcp.ie]

I get that error on several machines although I have tested (memscope memory test, ms memory diagnostic... etc.) and replaced memory chips several times (different sizes, different brands) and other parts as well (cpus, hdds, opticals). Machines/OS (XP Pro OEM SP2 with all updates) seems to work stable, machines don't "freeze" (even being very "hard tested"), basically they don't give any hassle. Motherboards are ASUS: P5GZ-MX and P5VD2-MX (latest BIOS). Should I concern?

[Cosmo]

I have seen this as well in our production floor. While the error may be misleading it should be used only as a reference.

Since you have tried the obvious, try the no so clear options.

Keep in mind how BIT works. The RAM is tested with a large file, then dumped to a file on the hard drive, so multiple things can go wrong.

I have seen this error in the past, and have had a bad IDE (Hard Drive, IDE cable, Mainboard controller), and also a CPU, since it has to compute the differences.

Try the quick options, CPU mainly, then if that does not help, try Hard Drive/Cable (IDE or SATA).

[David (PassMark)]

It would also be interesting to know how many of these errors you had, and how often you get them. Plus a dump of the log showing the details of the errors.

It is not really correct to say that the RAM is tested with a large file. The RAM is tested by allocating a large block of free RAM from the operating system, then writing and reading different test patterns. There is no file invovled.

The only file that might become invovled is the paging file. Depending on what else is happening in the machine, Windows might move some of the physical RAM out to disk, into the paging file.

So it is true to say that an error in the paging file on disk can look like a fault in physical RAM.

It is also true that a bad bus, EMI or bad CPU can cause a problem that initially looks like a RAM issue. As Cosmo pointed out, the CPU is responsible for checking what is read from RAM is the same as what was written. But if the CPU is malfunctioning, then CPU errors might look like RAM faults.

[Cosmo]

I am very sorry, you are correct. I just took it down to a simple use. Actually there are multiple threads for reading, writing and verifications, so it is not one BIG file, but there are multiple files used as a comparison.

The only reason I mention this is I faced about 5 days of trying to locate a fault of this nature from one of our production units. And in the end we changed the CPU as the last measure and it fixed the disk error.

While the BIT software is great, sometimes the hardware just doesn’t meet the needs. In our case there were no CPU errors, just disk errors.

Sorry about the misunderstanding.

[David (PassMark)]

Just to clarify.
If you are using the standard memory test, then there is just a single thread for RAM testing. (but there might be other threads running for the other tests).

There are not multiple files created for the RAM test. Not even a single file is created. Disk access is not required to test RAM.