Error in memtest test #7

[teko]

I've just got some Mushkin lvl2 BH5, they run stable @ 240 2-2-2-5-1T @3,0V, but when I up the voltage to 3,2 and above they start making errors in test #7 :( Have they been abused or des anyone now what the test's mean? I tried to search the forums, but the results only regarded test #5 and #8..

[TL1000S]

Test #7 is the first/best test you use to check if your RAM-chip(s) works at a given speed/timing.
As I've seen from several reviews/forums it's "normal" that a chip fails when you apply more voltage(well over "stock" that is).
They respond up to a given increase in Voltage. Over that it fails.
And as any other components (ie. CPU) "your result may vary".. unfortunately...

[teko]

But this is BH5, the old type, it should be able to do with 3,5V+, I have 1*120mm Sunon at 12,5V cooling them :/ But it seems like it's only with both sticks, single they run 250+.. Dunno why, anyone has an idea why?

[teko]

Could it be the mobo that isn't able suply clear voltage to them @ 3,3V+ (when I'm using less I'm still using the 5V jumper and they work @ 3,2V)

[Monge]

Try Dynamic Counter = disabled

Swap memory around:
Fastest stick Slot 1 or 2
Slowest stick Slot 3 or 4 (near the CPU)

[babyelf]

swap the sticks around?

edit : lol beaten

[Monge]

babyelf: I edit to much lol...

teko: also try:
tRC= 7
tRFC = 12 or 13
Tref = 4708
Max Async Lat = 9ns
Read Preamble Time = 7ns
Idle cycle Limit = 16clk

Also play with Drive Strength and Data Drive Strength.

[teko]

I have tried to swap them in alle possible ways, no change :( Now I'm trying to run without the 5V jumper, and see if that helps.. So far it's running 245@3,2V.. But that's no fun, they should at least be getting 3,3V.. I'm just going to let it run throug one memtest, and then i'll try to disable TDC..

[teko]

It's not due to the use of 5V jumper, it made the same errors.. Tried to use the settings you adviced, now it doesn't boot :/

[funkflix]

First flash to 704-2BTA bios and try this Timings:

http://img478.imageshack.us/img478/4845/page15yw.jpg

http://img478.imageshack.us/img478/4629/page20jl.jpg

Good luck! :)

[teko]

@ GravediggA, it didn't help.. now it's just crashing before it ever get's to memtest :/ The problem is mainly that they get errors at voltages higher than 3V, and that it's only in test #7.. If run test #7 @200 MhZ 2-2-5-2-1T @3,2V it gets something like 16-25 errors per pas..

[teko]

I don't believe that they haven't been burned in, but they have been out of the computer for 4 days, could it be that they need to be burned in again or something like that? A sort of cold boot bug?

[funkflix]

Hmm, i have the same sticks, i will look, if i get errors in test #7, never run it before.. :D

Edit: Never burned in my sticks and they run more then great. ;)

[funkflix]

Hmm, i think ur sticks are not intact, try to test it seperatly, to see which stick make the errors!

255MHz 3,3V tightest timings --> no errors here.

http://img499.imageshack.us/img499/8964/memtest77ob.th.jpg (http://img499.imageshack.us/my.php?image=memtest77ob.jpg)

[teko]

They didn't :mad: one of them is making the errors! I tested them earlier, at that time they both worked fine, I must have made some kind mistake :( Well, now all there is to do is get a hold of the guy who I got them from..

[Onepagebook]

Try Trrd as either 2 or 0
Tref as 100, 1.95us or 133, 1.95us

Dram drive strength/ Data strength---->7/3 , if 7/3 doesn't do it, the try 7/2...7/1..7/auto..5/3..5/2...etc.

I would assume that's the problem from strength/ Data strength;)

**you might wanna look at my setting here, a bit crazy, but may help a bit for a64tweaker settings.
http://www.xtremesystems.org/forums/showthread.php?p=1109502#post1109502

[Monge]

teko: I played with 2*512MB Twinmos UTT/CH-5, DFI NF4 and Venice 3000+ DPMW/bios 510-2 xcel2 yesterday and I could provoke memtest 86+ #7 error at Vdimm 3.5v and HTT 265MHz. But I got errors at #2 and #3 too. So I tried to lower tREF value to 0648 and I got rid #2 and #3 but the #7 errors were still there. Also I did try different tREF, Drive Strength and Data drive Strength and still errors at #7. I set vdimm lower and no errors at #7. In my case maybe it's the IRC of the Venice 3000+. I'm really in doubt, what the #7 errors indicate.
My Twinmos UTT/CH-5 really should be able to handle high voltages.
256 MHz@3.2v and 258MHz@3.3v no problems with memtest86+.

[babyelf]

teko which bios are u on?

[teko]

I'm on 704-2BTA, but it's only one of the sticks that are making the errors :/

[Monge]

Damm... I have a similar problem with memtest86+ #7 and my 2*512MB Twinmos UTT/CH-5 now. At 3.3v they give me errors at 508.1MB. Up to 3.2v ok, but at 3.3v gives me errors at #7 at 508.1MB.

[teko]

Dunno if it's maybe a bios problem? What bios are you running with?

[Monge]

Dunno if it's maybe a bios problem? What bios are you running with?
Ran bios 510-2 XCEL2 and 612-2BTA.
I did a quick test with each stick in yellow slot 1 and one was good for 260MHz@3.3v a memtest86+ 1 pass. The other stick could not run 250MHz @3.2v without errors at 273.3MB in test #2 and 5 and 7. So I believe the second stick is bad. UTT dead huuh... But the strange thing is, that the good stick ran in Orange slot 2 and the bad in Orange slot 4. I wish I could find a matching stick, but the Twinmos UTT/CH-5 is EOL.

EDIT:Running the bad module 250MHz@3.1v Bios 613-3 from 3.3v line now without errors in memtest86+. But if I increase voltage past 3.1v I get memtest86+ errors in #2,5,7.

[Monge]

Update: Ran the "bad" Twinmos module with bios 613-3 250MHz@3.1v and tREF=3120. Got some errors at 257.3MB and 273.3MB .Then changed the Drive Strength = AUTO and Data Drive Strength = 2 and got rid of them at 250MHz even if I increase to 3.2V at 250MHz. But 255MHz is not possible for me.

Monge

[teko]

You are still running from the 3,3V line? 255 should be possible with a bit more power, but then it might make the same errors in #7 again.. I've been wondering if it might be a riple error..

[Monge]

You are still running from the 3,3V line? 255 should be possible with a bit more power, but then it might make the same errors in #7 again.. I've been wondering if it might be a riple error..
Yes, still running from 3.3v line. If it's a ripple error I don't know. It's only on one stick I get errors at 257.3MB and 273.3MB.

Btw I changed
Trrd = 0
tWTR = 2
tRWT = 3
tREF= 3120
Drive Strength = Auto
Data Drive Strength = 2

No more tight timings with this stick I suppose.
I'll try another memory test program later today.

[Monge]

Finnally I did a test with Goldmemory 6.68 at 200MHz (Load Optimized Default) and Goldmemory gave me an error in first run. So the stick is bad, no?

Update: Set 200@2.7v, Drive Strength = Auto and Data Drive Strength = 2, the rest as Optimized Defaults and passed 1 test of Goldmemory.
Set 245MHz@3.1v and Goldmemory throws lots of errors.
So I give up.
The stick gets worse for overclocking I think.
Only good for standard speed and timimgs lol...

[teko]

I don't know goldmemory, so don't know ;) But I feel pretty sure about what memtest tells me, so I would say that one of the sticks is a bit crappy :/

[RVWinkle]

I ended up with similar probs as the op. In my notes I've remarked:
Test 7 fails when volts are too high
Test 5 fails when volts are too low or bus is too high


I haven't messed with oc'ing my cpu and mem for a couple months now but I'll be watching this thread since I finally got my water loop installed and plan on oc'ing some more soon.

[Monge]

I don't know goldmemory, so don't know ;) But I feel pretty sure about what memtest tells me, so I would say that one of the sticks is a bit crappy :/
Memtest86+ is good, but I also think Goldmemory is pretty good. I never used it before now. Did you find a solution for your Mushkins?

[Monge]

Orange slot 2 and the bad stick:
250@3.1v + .03v
Load Optimized Defaults +Timings 2-2-2-5
and

tRC = 8
tRFC = Auto
Drive Strength = Auto
Data Drive Strength = 2

3 rounds memtest86+ No errors
1 round Goldmemory no errors

But if I increase to 255MHz@3.3v I get error in #7 in second pass of memtest86+.
But changing the tRC to 8 instead of 7 gave at least 1 memtest86+ pass at 255MHz.