Gigabyte GA-870A-UD3 - 2 x 2gb Ram = OK. + 2 sticks = Blue screens

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
Hello, I'm still a bit of a PC noob so go easy on me please :)

So 2 days ago I put an extra 4 GB of ram in my machine. The same ram as the 4 GB thats currently in. The PC has ran fine all stock settings for 4 months.

Stuck the Ram in. After a while I start getting blue screens / games getting errors and crashing. The blue screens seems to be somewhat random but for a while atleased they were "An attempt to write read only memory..."

All the ram is tested and fine. Windows picks up that it has 8GB of Ram. 4GB is fine. (New or old). 6GB is fine. 8GB (all slots) is causing issues.

I have (as some suggested) tried upping the voltage to the NB and also the ram. It seemed to make the system say stable for longer but not solve the issue. I would get game crashes more than straight to bluescreen.

I can have the system stable at 1066 (all voltages AUTO) but the ram has a suggested 1333 and really, i'd like to be able to run it at this.

Recap - Bluescreen = "An attempt to write read only memory"
Prime 95 error = (FATAL ERROR: Rounding was 0.5, expected less than 0.4)

PC
1 x Gigabyte GA-870A-UD3 870 Socket AM3 8 Channel Audio ATX Motherboard
1 x Kingston 4GB (2x2GB) DDR3 1333MHz i5 Memory Kit CL9 1.5V
1 x AMD Phenom II X4 965 Black Edition 3.4GHz Socket AM3
1 x Gigabyte GTX 460 OC 768MB GDDR5 OC Edition
1 x Samsung HD103SJ Spinpoint F3 1TB
1 x Xenta 600W 12cm Fan PSU
Added Later
1 x Samsung SyncMaster P2450H TFT LCD 24" HDMI Monitor
Added Later
1 x Arctic Cooling Freezer 7 Pro rev 2
1 x NZXT Phantom Black Full Tower Case
Added Later
1 x NZXT Sentry 2 Touch Screen
1 x NZXT 200MM 11 Blade Rifle Bearing Fan
1 x Sony Optiarc AD-5260S 24x DVD±RW
1 x Kingston 4GB (2x2GB) DDR3 1333MHz i5 Memory Kit CL9 1.5V

No yes I will admit that I just went on the gigabyte website and found some ram that it said was compatable with the board and ordered it (heard Kingston was pretty good for ram)then later saw the "i5" bit... but its never been an issue (until now?)

If anyone can offer me some help I would be hugely greatful as I'm starting to pull my hair out with it now and so are my mates as I keep crashing from online games and making them lose! :(
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
Just another note I just thought of....

I have seen people going on about anti virus software sometimes causing strange issues like this. AVG Free did start acting very strangly and gave me "unprotected" errors etc, seperate parts of it were not loading correctly and I had to click them manually and use the "Fix" button.

Is it worth uninstalling AVG, disable fire wall and just run Prime95 again and see if anything changes or is this just garbage posted by people who don't know what they are on about?

Can't try anything for a few hours - at work, would be cool to get some advise so I can get to working on my PC when I get home :)
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
6GB is fine with... I had both new sticks and 1 old in as I started with the 2 new sticks (to see if I got an error) then added 1 old one, worked fine. all 4 just causes the issues. Tried swapping them around quite a bit and any combo of 3 seems to be fine
 

bilbat

Splendid
Something to try:

On the "MB Intelligent Tweaker(M.I.T.)" page of the BIOS, select the "DRAM Configuration" sub-menu;

(to change any settings here, you will need to toggle the second item, "DDR3 Timing Items" to "Manual")

start by looking at "1T/2T Command Timing", it needs to always be at "2T" for two DIMM per channel setups;

check your refresh timing (shown as "Trfc0 for DIMM1", "Trfc2 for DIMM2", "Trfc1 for DIMM3", & "Trfc3 for DIMM4") - this requires a bit of discussion:

on Intels, refresh timing is set like all the other RAM parameters - in 'counts' of memory cycles, and for two DIMMs per channel, it needs to be increased roughly 10-15% from the SPD 'stock' settings; on AMDs, it's handled in discrete 'chunks' of time expressed in nanoseconds. Also, with AMDs, it is possible (though unlikely) that the BIOS' memory 'training' functions can set individual DIMMs to different time periods - so here, we're looking for the highest value set - that's the one we're going to work with. Whatever the largest value shown is, set all of the refreshes to the next higher value...

Thus, if you start out with, say:

"Trfc0 for DIMM1" @ 90ns
"Trfc2 for DIMM2" @ 110ns
"Trfc1 for DIMM3" @ 90ns
"Trfc3 for DIMM4" @ 110ns

you'll want to set them all to 160ns...

In the much more likely case that you start out with all at 90ns, set 'em all to 110...

Exit back to the main "MB Intelligent Tweaker(M.I.T.)" page; set the "DRAM Voltage Control" about four hundredths higher; i.e., you RAM is listed as volt and a half, so likely will be at 1.500 - 'bump' it up to 1.540...

Do an <F10> to save the settings, exit from the BIOS, and reboot - see if she works! If you still have problems, try toggling the "DCTs Mode" opposite of what it's at - some setups are sensitive to this...

Best utility for memory testing is MemTest86+ - can find details about its use around half-way down, here...
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
start by looking at "1T/2T Command Timing", it needs to always be at "2T" for two DIMM per channel setups - it's 2T

"Trfc0 for DIMM1" @ 90ns
"Trfc2 for DIMM2" @ 110ns
"Trfc1 for DIMM3" @ 90ns
"Trfc3 for DIMM4" @ 110ns"

At the moment these are all set to 100 if I remember right. so 120?

Il give it a shot! and yes the power supply was.... I had like £30 in my account and had forgot to order one so had to just grab whatever I could at the time as my old one is.... total crap. il have a look about on ebuyer and see what I can find.



 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
[Thu Mar 10 07:30:44 2011]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Thu Mar 10 19:31:13 2011]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

Anyone got any idea's?
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
Scrap that, that was at 1.5..... I did tell it 1.54 but yes. 1 of the tests has failed again at 1.54 "worker 2" rest seem fine. I'l try upping the voltage a little more and see what happens.
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
went from 1.54 to 1.60 - everything else on auto. Passed 7 Prime95 tests. no errors. no faults. seems it likes 1.60. Am I safe at this voltage? if not I will try bring it down until unstable then back up 1. Thanks for the help and I hope at leased this thread may help someone else too :)
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
Think i'm going to go for a http://www.ebuyer.com/product/159920 and buy a Frio cooler next and start OCing (my CPU) a little. This fix to my Ram also somehow bumped my CPU score on Windows Peformance text from 4.8 (always thought this was low for my cpu....) to 7.5! I'm leaving the dvd rom/200mm fan and fan touch screen controller sitting in the box next to me until my new PSU arrives tomorrow, just in case. Thanks for the advise!
 

bilbat

Splendid
Indeed, you are good at a tenth high - almost all high-speed RAM for the iCores is rated at 1.65; is made up of, in essence, the same chips; and often still takes a 'bump' from 1.65 to 1.67-1.68 to work successfully in two DIMM per channel setups...

Happy computing!
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
Quick update. after 3 hours of gaming all hell broke lose again. no blue screens though, just crashes and Prime95 giving errors. I tried a few things but the 2 that worked to stablise it again was.

overclocking CPU to 3.8.. this was a shot in the dark. work magic though, only issue is that until I get my frio. I cant do this without 70+ heat. (under prime95 torture test)

so back down it went.

Ram now seems happy back on 1.6 but instead of 9-9-9-24 its set to 10-10-10-26 I heard losing up the timings can sometimes work wonders. let me know if I'm being an idiot !

seems to have stablised it again though.

Going to play a few more games and see what happens


P.S > i also uninstalled AVG just in case. Going to go back to NOD32 now i can afford it.

Can confirm 20mins of torure testing all is fine. Still going to get on to Ebuyer though as I have a ticket open already and see if they can offer some adivise tomorrow.
 

bilbat

Splendid
Have three more suggestions:

Does your video driver have a card tuning section, where you can lower the VRAM bus speed, and the actual GPU core speed? Try tweaking it down a bit...

In the "MB Intelligent Tweaker(M.I.T.)" page of the BIOS, set the "PCIE Clock(MHz)" to "100", not "auto"; depending upon other settings, Award BIOS often set 'auto' incorrectly...

Pop over to TheSycon, download and run the DPC latency checker - see if you're 'in the green'...
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
I'll give it a shot when I get home! thats for the continued support! Also going to order the new power supply now just to rule that out too. I can alsways put my current one in my old machine anyway

Right, managed to find the antec PSU XDISPLAY - New but opened box for £63 and got it arriving from ebuyer tomorrow so I can atleased rule the PSU out if nothing else tomorrow
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
Right'o.

Here we go again!

I tried that but with no real luck, so i took the bull by the horns and started over.

Started back to 1066 (I knew it was safe here)
Passed

Tried
1.6v DRAM again at 1333 - instant fail

Upped NB voltage to 1.15
2 tests - fail on second

Upped NB voltage to 1.20
2 tests - fail on second

Upped NB voltage to 1.25
3 tests - fail on third

Upped NB voltage to 1.30
Now the magic begins.

27 tests in 36 minutes - 0 fails - stopped test as all looked good :)

Seems it just wanted even more bloody power!

New psu arrives tomorrow so this should help. I'll report back after some gaming and whatnot.
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
right so...... after all that i get bluescreens again. I did notice 4gb are 001 CHINA and other is 002 TAIWAN - although exactly the same ram. seems dif batches

I'm on to kingston to ask if its an issue. for now i've lost patience and pulled 2 sticks out
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
I seem to have come to a wall.

While waiting for the guys at kingston to get back I had my new psu and fitted all my new bits and bobs then got back to the Ram battle.

I'm at a point now where;

6GB stable - 6GB overclocked to 1600 stable.
8GB - 1333 - hopeless
8GB - 1066 - stable

I've pretty much ruled everythiong else that is phsically possible also over the weekend.

I guess I should be happy with 8GB 1066 - its not exactly bad...... but its not really what I paid for either.

My question is...... will more voltage fix this or will I have to settle for 1066?

I've gone it to 1.65v to Ram and 1.34 to NB - still unstable. I'm not really sure how high I can test :s

I did have it running for a few hours at 1333 8GB totally fine on auto 1.5v yesterday and it passed an hour of prime95 but then started acting up again
 

bluetoes

Distinguished
Mar 10, 2011
19
0
18,510
Well i've settled at 1066 now,

Have OC'd the graphics card slightly (around 10%) and bumped the CPU to 3.8 without increasing anything but the multiplier. 58degrees full load for an hour on prime.

Tried to be cheeky and push 3.9 but it bluescreened (i've heard 3.8 is kinda the wall on stock voltage)

I'm going to leave it at this until I start itching to by a Thermaltake Frio then maybe start the journey to 4.0+

Thanks for all the replies. Just seems that for whatever reason my system just couldn't handle 4 x 2gb @ 1333