System instability when running memtest

[Macgruber]

Standard 100 times x 5 passed
high 20 times x 2 passed
very high 20 times x 2 passed
maximum (14gb) 5 times passed.. 3 times failed

should i keep increasing vcore?

i7 2600k @ 4.4ghz v1.315 in bios on a msi g45 mobo with 16gb 1866mhz corsair vengeance pc15000


all other settings are auto, vdroop, pll, should i keep increasing vcore untill i can pass the maximum more than 50 times?

to be honest i havent ran it much since it takes 10 minute for 1 result, so ive kept trying while i watch a movie(which is probably bad to do eh? ^^)

 

[ZipSpeed]

I had an issue with my X6 1055T where it would pass IBT except at maximum. Thought the processor needed more juice so I continually pumped more voltage. By the time I got to 1.47V, I started to wonder if the culprit was even the CPU. I ran memtest86 and it passed. While searching through the forums, I stumbled upon a thread where IDC mentioned HCI memtest due to a similar problem he was experiencing. HCI memtest detected errors that memtest86 did not. Turned out it was RAM problems I was experiencing. Threw in some new RAM and the CPU has been humming along at 3.5 GHz/1.40V with no issues.

 

[Macgruber]

THANKS For the insight dude

yeah i got that 1866mhz ram that seems it might be causing me various diff issues, im not willing to overclock the ram or mess with it and i also got cold boot issues

im running memtest.exe http://hcidesign.com/memtest/purchase.html made by those people
unfortunately i got the free version and i dont know if running 8 x 2GB tests will be the same as a full 16gb test on it

hopefully its not my ram seeing i just bought 8gb for 70$ at the store a few days ago while i bought the original 8gb for 60$ months ago

edit- i can run the computer for dayssssss with everything loaded and running, and its just a few random tests i fail, while i want to be "100% stable" in stress test terms ,i dont know whether its worth the added voltage im putting..

 

[Idontcare]

im running memtest.exe http://hcidesign.com/memtest/purchase.html made by those people
unfortunately i got the free version and i dont know if running 8 x 2GB tests will be the same as a full 16gb test on it

It is the same.

 

[Vectronic]

im not willing to overclock the ram or mess with it

Mess with it... I'm not going to pretend like our systems are the same (Z77 Ext4, 3570K, Patriot 1600 @ 1.5v), but my combination despises 1866.

I didn't bother saving the settings, but it basically goes like this (i gotz sum 15 second photoshop skillz up in dis biotch)

it hates all the 133/66 multipliers, but especially 1866

Edit: that is... *after* it passes memory tests, it might be worth looking into 1800 or 2000, shouldn't need to change the voltage, possibly timings (+1/+1/+1/+2) for 2000.

 

[tweakboy]

Fix your vdrop and raise your PLL to max if you can and youll be set. Thats what happens when you buy a cheap motherboard and cheap ram. Patriot ?: Tom Brady ?

 

[Macgruber]

79% on 8x2gb memtest.exe tests 0 errors.

it sounds like what youre saying might be true then eh ;0

does this mean i wont be able to set to 2000 with XMP and avoid actually messing with the voltage, and timings myself?

 

[Phynaz]

What are you doing? You're having boot issues, right? Then why are you screwing around with this?

 

[Macgruber]

What are you doing? You're having boot issues, right? Then why are you screwing around with this?

im testing my memory to see if its causing IBT to fail at the random times it fails

Because then i can take off about a good 0.2v off my CPU vCore and still run without BSOD's

but yeah im getting COLD BOOT sometimes more than twice rebooting

 

[Phynaz]

First rule of stability testing - and you have multiple stability issues - run everything at stock.

 

[Vectronic]

does this mean i wont be able to set to 2000 with XMP and avoid actually messing with the voltage, and timings myself?

It depends, some RAM comes with multiple XMP profiles, some don't... some mobo's don't allow access to profiles, etc

Otherwise, it's not very complicated, you should be able to "use" the XMP profile, and just change the main timings manually.

For my RAM, this isn't very good... it does the typical/stock timings correctly, but really underclocks the more technical/specific timings, so I generally use "Auto" for a given multiplier, set voltage, set timings... if it works, then move onto the complicated timings.

Your RAM is (assuming) running at 1.5v... that's not hard to remember... you're not really trying to overclock, just get it stable... so leave it at 1.5v.

Have you double-checked that things are running at stock/designated/suggested values?... the voltage and timing might be corrent, command rate might not be... it might be 1T instead of 2T (some RAM is 3T, kinda rare for "performance" DDR3).

I dunno, once again... my "scale" goes like this (for minimum stable):
1600: 8/8/8/20/1 (but I usually use 8/8/8/22/2)
1800: 9/9/9/22/1 (usually use 9/9/9/24/2) - same for 1866
2000: 10/10/10/24/1 (usually use 10/10/10/26/2)
2133: 11/11/11/28/2
2200: 11/11/11/30/2

More or less one level down from stock (1600, 9/9/9/24/2)... I don't usually bother trying for stuff like 7/8/7/20... no real effect, and mathematically I don't like it...lol

 

[Macgruber]

First rule of stability testing - and you have multiple stability issues - run everything at stock.

see this is the problem, everyone points out the obvious that theres a problem in which i previously state, except noone can help me properly diagnose

so to not be rude.

yes i have a problem..

running at stock speeds- ican leave pc on for 4 days running with everything

with my current speed/settings - ican leave pc on for 4 days running with everything

at stock speeds- it COLD BOOTS still.

at my current speed/settings - it cold boots still.

so obviously i have a issue, but do you know where it is? because if you do id be more appreciative than you can imagine, but telling me i have a problem is just rubbing it in my face. remember im asking you for help, im not here telling you you dont know what youre talking about.

so yeah. my setup randomly fails IBT sometimes, is my pc stable? yes, do i ever get BSOD? no, is there data corruption? no...


the last thing i want to do is go pay geek squad 300$ to have someone who could be even less knowledgeable than you, touch my computer

 

[Macgruber]

It depends, some RAM comes with multiple XMP profiles, some don't... some mobo's don't allow access to profiles, etc

Otherwise, it's not very complicated, you should be able to "use" the XMP profile, and just change the main timings manually.

For my RAM, this isn't very good... it does the typical/stock timings correctly, but really underclocks the more technical/specific timings, so I generally use "Auto" for a given multiplier, set voltage, set timings... if it works, then move onto the complicated timings.

Your RAM is (assuming) running at 1.5v... that's not hard to remember... you're not really trying to overclock, just get it stable... so leave it at 1.5v.

Have you double-checked that things are running at stock/designated/suggested values?... the voltage and timing might be corrent, command rate might not be... it might be 1T instead of 2T (some RAM is 3T, kinda rare for "performance" DDR3).

I dunno, once again... my "scale" goes like this (for minimum stable):
1600: 8/8/8/20/1 (but I usually use 8/8/8/22/2)
1800: 9/9/9/22/1 (usually use 9/9/9/24/2)
2000: 10/10/10/26/2
2133: 11/11/11/28/2
2200: 11/11/11/30/2

More or less one level down from stock (1600, 9/9/9/24/2)... I don't usually bother trying for stuff like 7/8/7/20... no real effect, and mathematically I don't like it...lol

Hey Vec, thanks for posting man, i need to check this

i hit 110% on all memtest.exe's and 0 errors, so i know its not memory causing my IBT fail at Maximum test... so this leads me to beleive its just cpu voltage and oc settings?

the memory still might be the problem for my cold boot i beleive. it would always set to 1600mhz when i do autosetup :/

 

[Vectronic]

I dunno honestly... if it was mine, and everything else was stable except that test, I would probably look into PLL, maybe VTT... I don't really have any suggestions there since mine seems to work (less is more) the opposite way of "normal" ones (more is more).

I have to drop VTT, PCH, and PLL by about 0.03v a piece from "stock" to get > 4.5GHz stable... no idea why because it makes no difference on temperatures.

Edit: something else to consider... maybe IBT is just crap... LinX** seems to work the best for the LinPack tests. IBT neither gets very hot, nor does it seem to stress RAM that well... but, I'm also on Ivy, not Sandy.

** I think that's the latest.

 

[Macgruber]

i gotta see how i can properly set the VTT and others man

no really good guides and ive had alot on my mind to put a clear configuration in order

last thing i want to do is run bad voltages and kill my cpu this week guess i got alot to look into

i wish someone had the same mobo + cpu and would let me cheat their settings to test my system ^^

Trying linx now and going to see if google has some more info for me

 

[Phynaz]

so to not be rude.

You are being rude. I am more than happy to help you with your issues, as evidenced by my replying to your PM about BIOS settings.

Many here will help you, but you are scattered all over the place with multiple threads and multiple issues.

So again, put everything back to stock. Then let us know which issue you want help with.

 

[Macgruber]

You are being rude. I am more than happy to help you with your issues, as evidenced by my replying to your PM about BIOS settings.

Many here will help you, but you are scattered all over the place with multiple threads and multiple issues.

So again, put everything back to stock. Then let us know which issue you want help with.

sorry if you feel that way, like i said not to be rude, but you can interpret it how you want, not trying to offend you but if you feel that way its your own conscience. yeah you pmed me and when i said i dont have that option and can show you a screenshot you quickly dissapeared, but low and behold youre in my thread.

yes many have helped, and many are all over the place with there response. remember i did ask for help, i didnt come here and say

"im only accepting help from COLLEGE EDUCATED civilians or military personall with DEGREE IN COMPUTERS"

the computer is basically at stock except for vcore.... so thats why im asking what is the point? i still experienced cold boot with stock settings lol i truly think its this combo of mobo+1866ram

 

[Kenmitch]

the computer is basically at stock except for vcore.... so thats why im asking what is the point? i still experienced cold boot with stock settings lol i truly think its this combo of mobo+1866ram

It's possible. Running with all slots full puts more stress on the MB and the memory controller. Sometimes it's just using the XMP profile that causes the issue.

From one of your other threads.

Edit- im using bios v2.6, newest is 2.9... a few people told me to NEVER update unless theres a specific problem in the updates that solves your problem so should i update or should i not ? :S - not sure if Intel RapidStart technology is the quickboot in bios but if it is i have set to disabled

Version 2.7

- Support Windows8 feature.
- Support GOP detection compatibility.
- Update ME firmware.
- Improved memory compatibility.
- Improved gaming keyboard compatibility.


Version 2.8

- Update CPU Micro Code.
- Update onboard Realtek LAN UEFI PXE driver.
- Improved PCI-E display card compatibility.
- Improved memory compatibility.
- Improved gaming keyboard compatibility.

Version 2.9

- Enhanced PCI-E display card performance.


The frequency of uEFI updates make me believe the MB has some bugs in the uEFI that MSI is trying to work out.

I know a lot of people don't like to update the uEFI/BIOS unless it's broken but when I get a MB and look at updates and see things stated like the above I'll update....I doubt I'm the only one that wouldn't even if it seems to be functioning fine.

 

[Macgruber]

It's possible. Running with all slots full puts more stress on the MB and the memory controller. Sometimes it's just using the XMP profile that causes the issue.

From one of your other threads.



Version 2.7

- Support Windows8 feature.
- Support GOP detection compatibility.
- Update ME firmware.
- Improved memory compatibility.
- Improved gaming keyboard compatibility.


Version 2.8

- Update CPU Micro Code.
- Update onboard Realtek LAN UEFI PXE driver.
- Improved PCI-E display card compatibility.
- Improved memory compatibility.
- Improved gaming keyboard compatibility.

Version 2.9

- Enhanced PCI-E display card performance.


The frequency of uEFI updates make me believe the MB has some bugs in the uEFI that MSI is trying to work out.

I know a lot of people don't like to update the uEFI/BIOS unless it's broken but when I get a MB and look at updates and see things stated like the above I'll update....I doubt I'm the only one that wouldn't even if it seems to be functioning fine.

would you personally update to the v2.8 or the 2.9? reason i ask is, as you mentioned, im scared to not "fix it" if its not completely broken

iguess im going to just save my bios, flash a new one, hope i dont mess up anywhere in the process^^ then im going to read more on lowering my VCore and making it droop less so i can decrease cpu vcore and have it stable with less of a droop

right not im at v1.3150 in bios and i droop to v1.280/288 when stressing

 

[Kenmitch]

would you personally update to the v2.8 or the 2.9? reason i ask is, as you mentioned, im scared to not "fix it" if its not completely broken iguess im going to just save my bios, flash a new one, hope i dont mess up anywhere in the process^^ then im going to read more on lowering my VCore and making it droop less so i can decrease cpu vcore and have it stable with less of a droop right not im at v1.3150 in bios and i droop to v1.280/288 when stressing

In your case I'm leaning towards v2.8 as I tend to keep one behind the curve just in case....Unless the most current one seems like the logical one to me. The description looks like it doesn't offer you much more.

You could try taking your memory off of XMP profile and setting it to spec in your uEFI first to see if it helps. You could also try pulling out two sticks to see if it helps. Another thing you could try is swapping the sticks around in pairs to see if it'll work.

If you use cpu-z and view the SPD tab switching from slot to slot do the profiles match?

If it was me I'd pull two sticks and set the uEFI to defaults before flashing the uEFI just in case tho.

 

[dbcooper1]

I'd set everything back to stock settings, plug into a UPS and flash to one of the two latest firmware levels, probably 2.9 but 2.8 is the latest that deals with anything you're having trouble with.

 

[Vectronic]

Personally I'd update... 2.5 was the best BIOS i've had for my board, but... I immediately went for 2.7 (2.6 was skipped by OEM) and then 2.8.

With those last two versions I've seen nothing but performance decreases, and no feature increases... not concerned, I can still hit 4.5GHz easily, and still get 4.8 if I want to... if I want to go higher I can always downgrade back to 2.5.

Yes there are horror stories about BIOS updating "bricked this, screwed that" etc... I've never had it happen (and I'm into the 20's for number of PC's owned), pay attention to what it's doing, don't do it through Windows, don't use any online/automated update from within the BIOS... download the BIOS toss it on a USB/Floppy/accessible HD whatever the method is... hit/type in the start command, go to the bathroom, come back... done.

Again, that's "personally"... IMHO, etc. You're probably worse off being paranoid about it, it's not open-heart surgery, you're just overwriting 8MB of data.

Updating the BIOS is something you probably should have done awhile ago in your testing, sort of assumed you had.