Test #8 screwing w/ me

[Seabee]

Alright, I have 2 GB of el cheapo Centon memory...I know, that is my first mistake.

But, here is the odd occurance. I tested each stick individually, all tests ran good for 2 hours. Paired them in every way possible, still good for 2 hours. Also, I paired them together, AND in both sets of DIMM slots. Once I stick all of them in, I error in the 1st pass of test #8.

I loosened the timings all the way up to 3-6-6-14(spec @ 3-4-4-8 @200hz), still no good. Pumped the vdimm up to 2.8v(spec @2.5v) and nada.

The rig is #1 in my sig. Again, each tests good by itself and when paired. I only get this problem when using all 4 sticks. Anyone else ever run into this??

[meionm]

change command rate to 2T, 2gb with 4stick may not run well in command rate 1T

[Seabee]

I have always had it running at 2T. But, I did notice that in SisSoft Sandra, it is reporting as running at 1T. I have tried both the "Auto" and "2T" setting in BIOS, but SisSoft still says it is at 1T.

[RedDragonXXX]

DL the A64 Tweaker and enable the ECC Mode and try again but be sure to save the settings at the startup or otherwise it won't take effect.

This is nothing unusuall as running with 4 stick of unbuffered RAM will take a hit on your memory controller and cause you to error out.

Also what PSU do you have, cause you will need something decent with more then 17A on a 12V rail.

[Seabee]

I have an Antec 400w PSU.

Turn ECC on? Since my memory isnt ECC, wouldnt that cause problems with it?

[RedDragonXXX]

I have an Antec 400w PSU.

Turn ECC on? Since my memory isnt ECC, wouldnt that cause problems with it?

No I ran with ECC mode before with my 2GB sticks and gave me better stability.

[Seabee]

Ok. I DLed v.50 I think and will try that later.

As far as real world operations, what does a failure in test 8 mean?

[RedDragonXXX]

Ok. I DLed v.50 I think and will try that later.

As far as real world operations, what does a failure in test 8 mean?

When you get a chance try reading through my Memtest guide (it's the 4th sticky down from the top), I know it's quite long :p but should answer pretty much all of your questions.

[Seabee]

I did, and didnt find anything more than 'Test 8 is more intensive than test 5'.

Since many benchmarks and testers use theoretical stuff, I was just curious as to what real world issues failing test 8 might bring about.

[RedDragonXXX]

I did, and didnt find anything more than 'Test 8 is more intensive than test 5'.

Since many benchmarks and testers use theoretical stuff, I was just curious as to what real world issues failing test 8 might bring about.

Test 8 [Modulo 20, ones&zeros]

Using the Modulo-X algorithm should uncover errors that are not detected by moving inversions due to cache and buffering interference with the the algorithm. As with test one only ones and zeros are used for data patterns.

[Seabee]

Test 8 [Modulo 20, ones&zeros]

Using the Modulo-X algorithm should uncover errors that are not detected by moving inversions due to cache and buffering interference with the the algorithm. As with test one only ones and zeros are used for data patterns.

Ok. That doesnt say what a failure would mean. I was looking for real world corelation to normal operations.

[RedDragonXXX]

Well if you want to know that you need to be able understand hex/bin. As for myself I'm in a process of learning.

[Seabee]

Used A64 Tweaker and enabled ECC. Instant lock-up, but somehow I expected that.

[RedDragonXXX]

Used A64 Tweaker and enabled ECC. Instant lock-up, but somehow I expected that.

Guess not all sticks respond to it same. I have been using it for some time now and gives me much better stability without any bandwith loss.

[Seabee]

I dont doubt that, but you are using much, much higher quality RAM than I currently am.

I have seemed to gotten it to run 2T now.

[Tiesto]

is your CPU overclocked?

i remember reading somewhere that test#5 is more for RAM testing, while test#8 puts the CPU's memory controller thru the wringer.

[Seabee]

Nope. For memtest I put the CPU at stock for that exact reason.