Burn in

[GoLLuM4444]

I'm currently burning in my new BH-5 with memtest86.

When I burnt in my TCCD I just left it cycling and it would get errors at the end of tests 5 and 8 and there would be fewer with each pass. I think this is usual (all error counts were 1s).

However with my BH-5 I get about 18 errors at the end of test 5 (in 1s) and 101 more in test 6: 96 spread evenly over the test and 5 right at the end (96,1,1,1,2).

Can I get rid of this unusual error pattern by burning in or is there some other cause?

 

[jellybeard999]

What volts, speed, timings (2-2-2 i presume tho)

 

[Gray Mole]

Defo need to know the timings and all that, speed volts etc.

However, if you just run 5 for now (picks up errors fast and don't have to wait it out the whole test) and see if the errors are the same, or if they increase or decrease.

If they increase, more fan powa requ'd.

How old? What make?

Ensure your base timings are loose for burn in (not the 2-2-2-5 but trfc etc) just to make sure you're not getting extra errors.

Gray

 

[GoLLuM4444]

Yeah sorry I didn't give many details. The RAM is the old Corsair 3200LL BH-5 stuff, timings were 2-2-2-6 but others might have been too tight, speed was 269Mhz, voltage was 3.6V. After leaving it for 6 passes the total number of errors was over 7000 so the errors were clearly getting more frequent. I may try again later with other timings loosened but I want to see if it's bench stable yet.

 

[GoLLuM4444]

Ok it wasn't bench stable so I tried setting all other timings to auto and just looping test 5 and the number of errors is still increasing drastically on each pass. Does this mean fan is not good enough? The RAM is warm to the touch but the mosfet is very hot. I'll try again when I can find a fan to fit on it.

 

[GoLLuM4444]

I'm not sure it's cooling related - it's even worse with a big fan by it. It still gets these random errors in test 6 at lower frequency. Could be voltage.

 

[gamesaregood]

Oh dear mate

Whats the highest you can do at those trimmings @ 3.6 error free?

Michael

 

[HARRY]

More Voltage

 

[GoLLuM4444]

Well I think the 3.6V is the problem - just need to cool RAM and MOSFET better. I can get about 268Mhz stable (2-2-2-6) - I think that was with 3.5V.

 

[HARRY]

how are u measuring the volts matey

 

[Gray Mole]

Each test is a little different, I'll have to have a look for the list of which is affected by which, but if your test 5 is clean, that's where you commonly see 'bench stable'

Too high or too low volts can do the same, and cooling the ram is good, and the mosfet that supplies it can be even more crucial to prevent 'noisy' voltage getting in there.

Harry is spot on, you need a dvm and a read point, then you can set your volts to exactly what you want, and know that it's right.

I needed 3.65 on my 256 bh5's to run that speed, but some need less, and some a little more, just keep finetuning and you should have 265 to 270 butter smooth as long as your timings aren't going out, or the default ones are too strange for memtest if you can't set 'em all (ie Neo2) and you can run your stability in windows with A64 tweaker to see what timings work for you.

Gray