BSOD when using two 8gb sticks of ram

[UchihaRyuk]

Hey there guys.

To begin, I have a cheap custom build of the following (main) parts:

- Biostar TZ75B Z75 Socket 1155 VGA DVI HDMI 8 Channel Audio ATX Motherboard
- boot drive = OCZ 120GB Vertex Plus SSD - 2.5" SATA-II - Read 250MB/s Write 160MB/s - 3 Year Warranty
- PowerColor HD 6870 PCS+ 1GB GDDR5 Dual DVI Dual Mini DisplayPort HDMI PCI-E Graphics Card
- Kingston HyperX 8GB 1333MHz DDR3 Non-ECC CL9 DIMM HyperX Blu
- Intel ® Core ™ i5- 3350P (Boxed , FC-LGA4 , Ivy Bridge )
- running = Windows 7 Pro 64bit

I recently bought:

-Crucial 8GB DDR3 1333 MT/s (PC3-10600) CL9 @1.5V Ballistix Sport UDIMM 240pin

Here is the chronological history of what happened:

- I installed the new Crucial ram in to slot 3. First start it recognised the ram (dxdiag detected it and photoshop spotted it as well).
- I didn't notice much improvement in PS even though I cranked up the ram usage to 12gb...
- I shut down and came back to the PC in the morning.
- Booted up to Windows log in screen, I logged in and as soon as the desktop started it crashed to bsod.
- After a few restarts and attempts to use windows built in tools to spot the problem I decided to try removing the ram.

Stage 1 - Removed the NEW ram. Left the old kingston ram in Slot 1. The machine booted up fine, and restarted two times fine.
Stage 2 - Switched the kingston ram for the crucial ram (Slot 1). Machine booted up fine using the crucial ram, restarted twice without issue.
Stage 3 - Put both ram sticks back in together. Crucial slot 1. Kingston slot 2. First time it booted up fine (just like last night), dxdiag and PS detect it again. I shut it down, restarted and hit BSOD again.
Stage 4 - Gave up, pulled out the crucial ram, moved kingston ram back to slot 1 and came to TomsHardware forum.

So, hopefully someone can help me with this. Thanks in advance!

Sincerely,

David Lawrence

 

[johnbl]

memory slot 1 and slot2
Note: the slot numbers start at slot zero these end up being the two middle slots on a 4 slot system
if you picked the first and second slot (got ya,one of the few combinations that will not work)

if you have 4 memory slots on your machine. Normally the slots are paired every other one
slot 0 and slot 2 are a pair
slot 1 and slot 3 are the second pair
(not the ones next to each other, done this way to let heat radiate better)
there are other acceptable combinations
(slot 0 and slot 3
slot 1 and slot2 Note: the slot numbers start at slot zero so

if your memory is in the correct slots then:

I would start with running on one stick and updating your BIOS, then your chipset drivers
from:
http://www.biostar.com.tw/app/en/mb/introduction.php?S_ID=583#dl

normally, when you pair RAM you have to match the memory or you end up searching for very detailed memory timing differences.
Hopefully, a BIOS upgrade will have fixed the issue.

The system can appear to boot up fine even when it is not ok. for a problem like this you would want to boot up on a CD or thumb drive with
a copy of memtest86.exe to confirm that your memory system is working as expected. (no memory failures)

- you could also have a bad slot
- some older motherboards actually require you to underclock your memory if you fill all the memory slots
(I have one of those board, The only indication was a * and a foot note in the manual that indicated this)

 

[shortstuff_mt]

Combining RAM from different manufacturers is never a good idea. After updating the BIOS and testing each stick individually with Memtest86+ I would try bumping up the voltage a bit to see if that helps. Don't go crazy with the RAM voltage. Maybe try 1.55v instead of 1.5v. You may never get the system stable with that combination of RAM.

 

[UchihaRyuk]

Ah thanks for the clarification @ johnbl. Yeah I had them 0 and 2 to start, then tried 0 and 1 the second time (this morning).

I have just updated bios and the chipset driver from the website above. I then installed the second stick again (kingston in slot 0, cruiser in slot 2) and it booted up first time ok.

You can see a picture of dxdiag and PS 'seeing' the memory in this photo - http://dclawrence.co.uk/images/full/a4d7252df854bc4a0fd995f8d18ee360.jpg

I chose to restart, and it booted up fine - checked dxdiag and PS again and all seemed well. I thought that it was too good to be true, so I shut down totally for 30 seconds and started it back up. Got to login screen and then BSOD.

I am now downloading the memtest86 memory checker, and will report back in due course.

 

[johnbl]

try the two memory slots in the center of the for slot group.
I think it will put them on different channels.

you should also download the user manual for your mother board to confirm which slots to use. each combination of slots will do different things. (dual channel, single channel)

run memtest86 after after you make a change, it will fail the fastest.

Ah thanks for the clarification @ johnbl. Yeah I had them 0 and 2 to start, then tried 0 and 1 the second time (this morning).

I have just updated bios and the chipset driver from the website above. I then installed the second stick again (kingston in slot 0, cruiser in slot 2) and it booted up first time ok.

You can see a picture of dxdiag and PS 'seeing' the memory in this photo - http://dclawrence.co.uk/images/full/a4d7252df854bc4a0fd995f8d18ee360.jpg

I chose to restart, and it booted up fine - checked dxdiag and PS again and all seemed well. I thought that it was too good to be true, so I shut down totally for 30 seconds and started it back up. Got to login screen and then BSOD.

I am now downloading the memtest86 memory checker, and will report back in due course.

 

[UchihaRyuk]

I decided to test slot 0 with both sticks first.

I let my old kingston stick do two full passes in a two hour period. 0 errors.

My new stick just went in to slot 0 and at 8% of its first pass it is already at 100 thousand errors and climbing.

Should I proceed with this pass? Or just try it in different slots? Or tweak settings somewhere in bios?

 

[johnbl]

the slot should be ok, the stick is bad if the timings are correct

I decided to test slot 0 with both sticks first.

I let my old kingston stick do two full passes in a two hour period. 0 errors.

My new stick just went in to slot 0 and at 8% of its first pass it is already at 100 thousand errors and climbing.

Should I proceed with this pass? Or just try it in different slots? Or tweak settings somewhere in bios?

 

[UchihaRyuk]

Ok so a little bit more info. I did as was recommended by shortstuff_mt concerning tweaking the dram voltage in manual voltage settings in the bios. I upped it to 1.53 and 1.56, saved and exited to memtest and both times within seconds there were hundreds of errors (as before)

Dead ram?

 

[UchihaRyuk]

I just took out the new stick, reverted to kingston so I could upload some relevant photos.

The kingston ram memtest:

Picture taken minutes in to the crucial ram memtest. I gave up after about ten mins:

my first adjustment of dram voltage:

similar (if not exactly the same) memtest result of crucial ram post-voltage-tweak:

I considered a picture of me face palming, but thought against it.

From what I could tell before purchase, (ignoring the different branding) these ram sticks numbers match up fine? So would this confirm a dud stick?

P.S. Many thanks for your help so far,

Dave

the slot should be ok, the stick is bad if the timings are correct

I decided to test slot 0 with both sticks first.

I let my old kingston stick do two full passes in a two hour period. 0 errors.

My new stick just went in to slot 0 and at 8% of its first pass it is already at 100 thousand errors and climbing.

Should I proceed with this pass? Or just try it in different slots? Or tweak settings somewhere in bios?

 

[shortstuff_mt]

Yes, you need to start an RMA of your new RAM. RAM that is working correctly should get 0 errors even during multiple hours of Memtest86+ testing.

 

[johnbl]

return the ram if you can.
Note, there are many more memory timings than the ones commonly listed.
some times you will see a 2n or 2T command cycle that means the memory can work at the rated speed put takes 2 command clock cycles to complete. I am thinking that the crucial it that type but I could not find the actual detailed chip specs so I would just return it as defective (it fails a memtest86)
-if you can get another stick of the same working RAM for you machine.

just fyi: I read a study and about 8% of memory will fail if actually tested. Most of the failures are single bit errors that can be corrected. I think about 1% had 2 bit errors.

Most of your failures looked like single bit errors
for example the test: wrote the value 0x7c63e8e7
1111100011000111110100011100111 <-- wrote this to memory
1111100011100111110100011100111 <--- but read back this
^ <--- single bit it wrong

I just took out the new stick, reverted to kingston so I could upload some relevant photos.

The kingston ram memtest:

Picture taken minutes in to the crucial ram memtest. I gave up after about ten mins:

my first adjustment of dram voltage:

similar (if not exactly the same) memtest result of crucial ram post-voltage-tweak:

I considered a picture of me face palming, but thought against it.

From what I could tell before purchase, (ignoring the different branding) these ram sticks numbers match up fine? So would this confirm a dud stick?

P.S. Many thanks for your help so far,

Dave

the slot should be ok, the stick is bad if the timings are correct

I decided to test slot 0 with both sticks first.

I let my old kingston stick do two full passes in a two hour period. 0 errors.

My new stick just went in to slot 0 and at 8% of its first pass it is already at 100 thousand errors and climbing.

Should I proceed with this pass? Or just try it in different slots? Or tweak settings somewhere in bios?

 

[UchihaRyuk]

Aaaah thanks very much for your assistance. Annoyingly have to wait til Monday to give ebuyer returns a call, but that doesn't stop me crediting you guys for your help!