Custom Torture Test Query
 

Hi everyone,

I have a query about the Custom Torture Test Option in Prime95.

I was wondering that if one leaves all the values in the Custom Test section the same as they are in the Blend Test, and only adjusts the amount of memory used, for example to 256mb rather than all available memory, if Prime95 runs without any errors, is there a chance that if allocating all the system memory and not entering 256mb or any other random value in the memory allocation field would cause errors or failure of the test?

Same goes for if one runs two instances of Prime95 on a Hyperthreading supported Intel P4 CPU and keeps all the options the same in the Custom Test section but only changes the amount of RAM used for each test - if all is stable and there are no errors found, could letting the test(s) use all available RAM find errors as compared to if Prime95 is not allowed to use all the system RAM?

I guess the long and short of my question is that if a system is Prime95 stable on the Custom Torture Test with the values all the same as the Blend Test but anything less than all the system memory used for Prime95, is this is an accurate enough test and would allowing P95 to use all the system memory detect any errors that running with less memory wouldn't?

Cheers.

Running the blend test with less memory is still a very good stability test.

Even running the blend test with more memory will not test all of your RAM (we have no control as to how the OS maps prime95's address space to physical RAM). You might try memtst86 for a more comprehensive memory test.

Cheers for the reply.

That is what I was interested in knowing - I have used Memtest with no errors at all so was wondering whether a blend test using the custom option and allocating less memory and having a system pass without errors was a good enough stability test and whether any errors may occur if all memory was allocated.

Cheers for the explanation, I appreciate it.