Questions about memtest86

[ytoledano]

1. Is it possible for memtest to not find errors when runnning at a lower frequency than find some when running at stock frequency?

2. Does it happen often that no errors are found on the first passes but some are found on later passes?

Suicide is man's way of telling God "You can't fire me, I quit!"

 

[HammerBot]

1) No
2) No

<i><b>Engineering is the fine art of making what you want from things you can get</b></i>

 

[ytoledano]

About memtest not finding errors when running at lower speeds - I had my memory past the test at 200 MHZ FSB and freeze during it at 258 (it's PC4200 so it's rated for 267). I realize it's not an error, but it's worth taking note.

I've in the rain I've [-peep-] in the snow
I've [-peep-] in the air and [-peep-] in the sea
but I'll never be happy and I'll never be free
'till I'll [-peep-] the army like the army [-peep-] me

 

[HammerBot]

That memory should run equally well on 200 and 258, since both are below rating. If your machine freezes during memory test, I would suspect a memory error.

<i><b>Engineering is the fine art of making what you want from things you can get</b></i>

 

[Crashman]

1. Yes, it's possible for memtest to find errors at "stock" frequency, but not at a lower frequency. The likely cause is that your RAM isn't stable at stock frequency, this is a common problem with todays overrated RAM.

2. Yes, this can be caused by the memory heating up as it's used.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>

 

[HammerBot]

You make me look bad, since I gave opposite answers!

1) We agree on that. But you answered opposite. Read the question again. ytoledano asked if it was possible to find errors at lower than stock frequency, when none are found at stock freuquency.

2) Agreed. But, IMO, it doesn't happen often. The memory isn't very massive and I bet it reaches its steady state temperature before the first pass is over.

<i><b>Engineering is the fine art of making what you want from things you can get</b></i>

 

[pIII_Man]

i disagree...especially if copper heat spreaders are used...

Memory has a quite low thermal output so it does take a while for its temps to ramp up...but they do...


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[HammerBot]

Thermal <i>capacity</i> has nothing to do with thermal <i>resistance</i>. Adding heat sinks to memory will reduce the thermal resistance (by increasing what you call thermal output) and thus the temperature of the memory. But the thermal <i>capacity</i> is actually increased slightly and thus it takes a bit longer for the memory to reach a steady state temperature (to ramp up). This, however, is not very significant considering the many minuttes it takes for memtest to run each pass. Thus it does not happe <i>often</i> that memtest finds errors in subsequent passes that are not found in the first.

<i><b>Engineering is the fine art of making what you want from things you can get</b></i>

 

[Crashman]

1.) I reread the question, he asked it backwards. He asked if it was possible for it to find NO errors at the slower speed, then to find errors at the stock speed. Which is a backwards way of asking if it's possible to find errors at full speed, but no errors at a slower than stock speed.

2.) I've many times found memory that had sporadic problems test perfectly fine for a few passes and then start registering a few errors on subsequent passes. It's possible that heat buildup in the case contributes to increased RAM temps over longer periods of time.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>

 

[HammerBot]

Ok. Sorry Crashman, and ytoledano too. I misunderstood the question. It is asked backwards. Damn.

<i><b>Engineering is the fine art of making what you want from things you can get</b></i>

 

[pIII_Man]

you basically repeated exactly what i thought to be a given...i agree with the thermal capacity stuff...

You have to also remember that certain parts of memtest are not as stressfull as others i still say it takes a few passes to get max heat...


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[HammerBot]

Forgive me if I'm wrong, but when referring to memtest passes, I assume repeated runs of the same test suite. Thus they are equally stressfull. Each pass runs through numerous tests that stresses the memory more or less. Some of these tests may found errors that others don't. But that's usually because they use a different test pattern.

<i><b>Engineering is the fine art of making what you want from things you can get</b></i>

 

[pIII_Man]

ok loop test 5 and tell me if its hotter than test one...


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[ytoledano]

My memory (OCZ PC4200 2.8V) gets warm quickly but not very hot. When I point the door fan at it it cools down. It's also very heavy with it's copper heat spreaders.

I've in the rain I've [-peep-] in the snow
I've [-peep-] in the air and [-peep-] in the sea
but I'll never be happy and I'll never be free
'till I'll [-peep-] the army like the army [-peep-] me

 

[peterfagg]

here's a thought - you said that it would lock at 258mhz. How can you be sure that it is your memory that is failing? Does windows continue running in the background? I only ask because my processor won't go past 233 Mhz fsb (my ram is 4:5 so slower), so you'd be doing well to get yours up to that fsb stable. Mine will boot into windows at 250 FSB but won't do certain things (disk cleanup is a killer! also occasionally crashes on 3D mark 03)

Obviously if you've got some fancy watercooling or whatever then that frequency is far more possible...

Peter

 

[ytoledano]

Actually, I've found the unstable component and it's neither the processor nor the RAM, it's the Northbridge. I have a Thermalright SLK-947U and I'm pretty sure the processor would be stable at 267FSB, it's "only" 3.25 GHZ.

POST is way overrated, the black screen and beeps should be enough for most people.

 

[peterfagg]

Fair enough. What can you do in that situation? I suspect my northbridge is the weak link in my system too. It has a passive cooler but as I use a Zalman flower heatsink the (120mm) fan blows directly onto the northbridge. I don't really fancy fitting a powered cooler, so what should I do? I have a temperature probe partially under the heatsink, next to the thermal gunk (is that a good place to start?!?) and the temperature is only 39 deg C (oops, better make that "only"!)

Peter

 

[ytoledano]

What makes you think the NB is keeping you from overclocking? I have the stock Abit cooler when seems to be just a small fast fan with a very thin heatsink. the Zalman looks a lot better.

POST is way overrated, the black screen and beeps should be enough for most people.