You should NEVER use Prime to test for memory instability. Prime is for testing CPU stability alone. Memory cannot calculate square numbers, period!
Fast fourier transform (fft) is used by the CPU to calculate the square of very large numbers and hence find Prime numbers. Small fft's are automatically created to to a ffT size range which will fit into the L2 cache of your CPU. As a result, the small FFT test is the one which accesses your main memory the least but it still puts some memory to use so instable memory will also show up on small fft's albeit only the section of memory being accessed. This is THE BEST stability test for CPU if thats all you want to test(the CPU). Large FFT's are too big for the L2Cache and therefore have to be held in the system memory which on all but the AMD 64bit CPU's would mean the on board memory -> northbridge->CPU link but as we know a64's/optys have on board memory controllers so the northbridge is cut out of the equation and the memory controller in its place. The 'In place' Large FFt's only test a specific section of memory whre as Blend tests a lot more memory over a wider area.
There is no 'best way' to test for stability nor is there one test for all that will stress your CPU more than another, hesky is right in that each individual test tests a certain part of the PC. Prime is for CPU (which includes cache and memory controller in our case). Saying that the Large FFt stress your CPU more is untrue, it stresses both the CPU an on die memory controller which will raise temps higher than small fft's which just tests the L2cache and processor. So to sum up to test for CPU stability, small fft is the test you want to run, if it passes that and fails large fft's then your memory controller is unable to cope either through heat or frequency NOT the cpu.
Hope this helps.
Fast fourier transform (fft) is used by the CPU to calculate the square of very large numbers and hence find Prime numbers. Small fft's are automatically created to to a ffT size range which will fit into the L2 cache of your CPU. As a result, the small FFT test is the one which accesses your main memory the least but it still puts some memory to use so instable memory will also show up on small fft's albeit only the section of memory being accessed. This is THE BEST stability test for CPU if thats all you want to test(the CPU). Large FFT's are too big for the L2Cache and therefore have to be held in the system memory which on all but the AMD 64bit CPU's would mean the on board memory -> northbridge->CPU link but as we know a64's/optys have on board memory controllers so the northbridge is cut out of the equation and the memory controller in its place. The 'In place' Large FFt's only test a specific section of memory whre as Blend tests a lot more memory over a wider area.
There is no 'best way' to test for stability nor is there one test for all that will stress your CPU more than another, hesky is right in that each individual test tests a certain part of the PC. Prime is for CPU (which includes cache and memory controller in our case). Saying that the Large FFt stress your CPU more is untrue, it stresses both the CPU an on die memory controller which will raise temps higher than small fft's which just tests the L2cache and processor. So to sum up to test for CPU stability, small fft is the test you want to run, if it passes that and fails large fft's then your memory controller is unable to cope either through heat or frequency NOT the cpu.
Hope this helps.
)
