• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

6gb ram installed, 4gb working

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

briligg

Registered
Joined
Sep 15, 2009
i checked a thread by d4rkst4r about this same problem, but his solution hasn't worked for me, and the thread has sort of petered out, so i am posting a new one.

i had a problem a few weeks ago that apparently was due to a failing psu, according to our local techie. My machine wouldn't post. He figured the old psu wasn't providing sufficient voltage. i got a new psu and he got the system stable again, after a fair bit of struggle, he said.

when i first got this system i had a ton of problems with bsods, until i was advised to adjust the back to back cas# delay, and then everything worked fine with that setting at 8. i realized our techie hadn't adjusted that, and so i did, and then i noticed i only had 4gb of ram available.

the moderator in the asus forum for my mobo blamed it on the dimms not being on the official qualified vendor list. the store advised me to check the pins. i opened a trouble ticket with asus, and 16 days later they have still not replied.

i did check the pins, although i don't think our techie moved the cpu when he was fixing my last problem. a good hard look under a magnifying glass didn't reveal any bent pins. all dimms pass memtest86+.

any advice?:confused:
 
i did check the pins, although i don't think our techie moved the cpu when he was fixing my last problem. a good hard look under a magnifying glass didn't reveal any bent pins. all dimms pass memtest86+.

any advice?:confused:

Have you checked the pins on both the cpu and the motherboard? Because it could be either one that has an issue.. Also, have you made sure you have your ram installed in the right DIMM slots for the mobo, and have you tried different DIMM slot configurations to see if that makes a difference? And you are in a 64 bit OS correct?
 
Have you checked the pins on both the cpu and the motherboard? Because it could be either one that has an issue.. Also, have you made sure you have your ram installed in the right DIMM slots for the mobo, and have you tried different DIMM slot configurations to see if that makes a difference? And you are in a 64 bit OS correct?

I'm running win7 x64. I didn't seriously check the cpu's contacts, I just took a quick look at them, but i figured not much could go wrong with them, because they are flat contacts, not pins. I don't recall if you have the Rampage II Gene, or if it was the Extreme or the P6T, but i have my dimms in the 3 blue slots - a1, b1, and c1. The manual doesn't show any other possible configurations for 3 dimms. I pondered just trying some out, but that seemed a little random.

how evident are bent pins? i've thought of taking everything apart again and taking a high-res macro photo to get a better look. Also someone said on Tom's Hardware that the heatsink could be the problem, if it is pressing down more in one area - and that could well be the case with my heatsink, the coolermaster hyper t3, because it fits but wasn't made for the i7, and it clearly is a bit misaligned, not centred ove the cpu. i'm not quite ready to buy a new heatsink just to test that theory - i gave my stock heatsink away... wonder if i can get it back...

(ah yes, i see now you have the P6T)
 
how evident are bent pins? i've thought of taking everything apart again and taking a high-res macro photo to get a better look. )

I dunno, it was pretty evident for me, all the pins are supposed to be aligned in same direction, and misaligned pins usually pop out.. if you want to you could upload a higher res picture of the pins and have a second opinion on it.. I'm not too sure about whether or not your heatsink would be to blame due to incompatibility with i7, however this does seem unlikely to me. And yes I do have the P6T.
 
i'll try taking that photo tomorrow. also i guess i'll try swapping around the 2nd and 3rd dimms.

i should look for that tom's thread. someone there claimed their problem was solved by adjusting their heatsink. this same problem seems to be popping up here and there.
 
i'll try taking that photo tomorrow. also i guess i'll try swapping around the 2nd and 3rd dimms.

i should look for that tom's thread. someone there claimed their problem was solved by adjusting their heatsink. this same problem seems to be popping up here and there.

Sounds good.. And be sure to report back if you find a fix!
 
try running memtest86, it might help... does bios report 6 GB?

there are memory brands some mobos just dont like.
 
i ran memtest86+ from boot early on, all sticks passed with no problems. the moderator on the asus forum for the rampage II gene blamed the ram model, it isn't on the qvl. but the store insists other customers have had no problems (the ram was sold as a package deal with the mobo and cpu), and they advised me to check the pins (the store is 5000km away, or i would have taken it in).

this morning i tried moving around the dimm sticks a couple of times. moving c1 to a2 caused the system to stop booting. moving that one back to c1, and moving b1 to b2 booted fine, but now the system only sees 2gb of memory. so i moved back to the a1, b1, c1 config, and tried to recover the other 2gb, so far without success. i tried booting up with the memOK button, but it hasn't helped. bios sees only 2gb.

i actually don't remember if i checked the bios before to see if it saw 6gb or 4gb, but everest was seeing 4gb, and i relied on that.

now i'm about to take off the cpu and take a high-res photo of the pins, to get input about if there are bad pins i didn't see before.
 
okay, here´s the photo. it looks all good to me.
btw, i was fiddling with the fasteners on the heatsink before i took it off. it´s really hard to get them to clamp down properly, and as i went to take it off, i noticed that two of them weren´t as tight as they could be. so i tried adjusting them and booted up again, and i had 4gb again. i popped the c1 dimm out and put it back again in the process of taking out the mobo, so maybe that is what caused it to be recognized again - but it wasn´t loose before or anything.
anyhow...
here´s the photo
i cleaned off the dust on the left side. the cpu lives with a 80mm fan right next to it cooling the north bridge. it´s super dusty here, so it accumulates quickly. don´t know if that would be a factor though.
 
I see that your signature says you have the RAM underclocked to 1066, why? I also see that the DDR3 1600 is supported but is an O/C setting on that board. You should check your BIOS and make sure that the timmings are set to 7-7-7-24 the voltage is 1.65V and the frequency is set to 1600MHz using the correct divider. If youre having problems with that, do a hard reset of the CMOS and go back and set it all manually.

It is strange that you say all three sticks check out on memtest. Is memtest actually seeing all of the RAM and identifying it correctly?
 
i checked the sticks one at a time with memtest86+ - it runs from boot, outside windows.

i have had serial ram problems, which is why sometimes i run it at 1066. also my north bridge runs hot. it now has its own 80mm fan, and it still rises above 65c in warm weather or when working hard - and although i don't know why, there seems to be a connection between the ram running at 1600mhz and the nb being warmer than when ram runs at 1066mhz. typically, the timings are 9-9-9-24, back to back cas#delay is 8, voltage is on auto at about 1.5v. if i put the ram up to 1600mhz, it's the same, but i put the dram voltage up to 1.65 and the qpi voltage at 1.35. it may be conservative, but i've just had too many bsods, and i do work that could really get screwed up by a bsod at the wrong time.
 
talking to our techie today, he said he couldn't see how bent pins could cause such a specific problem, one that doesn't even affect system stability. he recommends i check that the dimm sockets are good, and beyond that he'd need to see it.
d4rkst4r, or anyone else, which pins are supposed to be involved with this problem?
 
i checked the sticks one at a time with memtest86+ - it runs from boot, outside windows.

i have had serial ram problems, which is why sometimes i run it at 1066. also my north bridge runs hot. it now has its own 80mm fan, and it still rises above 65c in warm weather or when working hard - and although i don't know why, there seems to be a connection between the ram running at 1600mhz and the nb being warmer than when ram runs at 1066mhz. typically, the timings are 9-9-9-24, back to back cas#delay is 8, voltage is on auto at about 1.5v. if i put the ram up to 1600mhz, it's the same, but i put the dram voltage up to 1.65 and the qpi voltage at 1.35. it may be conservative, but i've just had too many bsods, and i do work that could really get screwed up by a bsod at the wrong time.

Ok a couple things here...

If you are having problems with the Serial presence detect (SPD) you can set it manually in the BIOS like I suggested in my previous post. If your RAM isnt stable at the manufacture recommened settings, RMA it or test it in a different system.

Once you have the correct settings on set in your BIOS check the RAM individually with Memtest. If they all work, try each of the slots one at a time till you find the one thats giving you problems. If all of the slots work on one stick try putting all of the sticks in a1, b1, c1 or a2, b2, c2 and run memtest again. Be sure to note if memtest is reporting the correct RAM size, speed and timings or not in the different configurations.

Sounds like you might have an issue with the mainboard because there is no reason you should have to down clock the RAM that much to get the system stable. IMHO you should run RAM at its recomended voltage even if you are turning down the frequency and timmings (which I don't think you need to do).
 
sorry, i was trying to be clever by saying 'serial ram problem' - as in over and over, not as in spd. my knowledge of my system is only what i have had to learn in my quest to get it to run stably without running to the techie every week. and unfortunately, i live in mexico and bought the system in my home country of canada, because i7 systems weren't available here, and the parts for my system are still only available by special order at a huge markup. if i rma'ed anything the cost of shipping would be more than the component is worth. nor can i test parts in another system, as it is likely my system is the only i7 machine in a 40km radius.

but once i set the back to back cas delay to 8, actually my bsods stopped. after that, really i was running at 1066 for the sake of my north bridge. it reached 61c today even with the ram running at 1066 and not much load. god knows what the relationship is, but with ram at its rated settings it hits 65c regularly despite its 80mm fan, right on top of it. i had the system assembled outside the case yesterday for a while, waiting for feedback on my photo of the pins before i put everything back, in case i actually was going to have to straighten pins. i had the system running like that for maybe 45min before it shut down - the fan was off the nb and it overheated. that's how badly it needs that fan.

the ram sticks are fine, but i do need to test the slots. i'm a little caught up with other things, but i'll try to do that tomorrow.

btw, i do have the first 4 timings at 7-7-7-24 again. i set them at 9-9-9-24 as i hadn't adjusted that in a while and forgot the correct numbers. thanks for pointing that out.
 
Last edited:
Check each one of the slots to see if one of them is dead.

The RAM frequency is not what will really drive the temperature of the northbridge. Ussually its overclocking and voltage increases that make the northbridge heat up. Some northbridges run hotter than others depeneding on the specs but you should be able to run them at full load for 18hr+ under recomended settings without failure on something like prime95 blend. If P95 is failing and everything is at stock settings, RMA the board.
 
still haven't checked the slots, sorry, i absolutely will tomorrow - do appreciate the help, just get caught up in other stuff.
the temp of the nb has always confused me. nothing on my machine is overclocked, or ever has been, except that sometimes i run the memory at 1600mhz with 1.65v, qpi at 1.35. but, i can't rma. shipping would cost more than the board is worth.

yellowbeard, that thread is a big help. the steps 1 to 5 have all been mentioned, and i just have to check 4 and 5 still. getting confirmation of the heat sink being a potential cause is very helpful - now i'm definitely going to get my stock heatsink back and try putting it back. i gave it to the same techie who says he doubts bent pins or badly fitted heatsinks are a cause here. although obviously i have to check everything else, my coolermaster hyper t3 clearly doesn't quite align properly with the cpu, and i don't like its plastic clamps. and if i may reinforce something said elsewhere, your contribution to this forum speaks very well of your company, and i'll definitely look for corsair ram next time.
 
Good luck to you and glad to help. EVA2000 did the hard work, I just copied and pasted. And, I was an enthusiast LONG before I got my job and lots of people helped me along the way. I hope you get it sorted quickly.
 
i tested the slots, to the best of my ability. i took out the sticks and labelled the stick that was in a1 '1', bi '2', and c1 '3'.

with 1 in a1, 2 in b1 - 2gb memory working
with 1 in a1, 3 in b1 - 2gb
2 in a1, 1 in b1, 3 in c1 - system doesn't post
3 in a1, 1 in b1, 2 in c1 - 2gb
from this point on i booted using the memOK button on my mobo - which i understand checks the ram and loads safe settings for it. i don't really get what it does, but i figured it was the best way to boot while testing ram sockets.
3 in a1, 2 in b1, 1 in c1 - 2gb
then i cleared cmos
2 in a1, 3 in b1, 1 in c1 - 4gb
2 in a1, b1 empty, 1 in c1 - 4gb
2 in a1, 3 in c2, 1 in c1 - 2 gb
that was sort of a random idea, that arrangement. i don't think i can use anything other and a1, b1, c1 on this mobo unless i have more than 3 dimms. a couple days ago i tried putting the c1 stick in a2, and only 2gb showed up, and then i put b1 in b2, and the system wouldn't post.
2 in a1, 3 in b1, 1 in c1 - 4gb
2 in a1, 3 in b1, c1 empty - 2gb
2 in a1, 3 in b1, 1 in c1 - 2gb
hm. clear cmos. no change. clear it longer. no change. clear it half an hour, no change. adjust timings, no change.
try swapping around stick again -
1 in a1, 3 in b1, 2 in c1 - 4gb
i adjusted timings again - 7-7-7-24, b2b cas 8. and just on a whim, qpi 1.225v, dram bus 1.51v. i just wondered if putting the qpi a little closer to the dram bus voltage would help, but i still have 4gb showing up, and actually, the bios is so inclined to do things itself if you don't set everything manually, i see those voltages are still fluctuating anyways, as the asus bios adjusts them on the fly, which is what it usually does.

so what i'm taking from this is that there is a problem communicating with the b1 slot, and the board isn't happy when i put dimm sticks in weird orders. i don't understand why the 2 - 1 - 3 combo wouldn't boot.

checking IOH voltages and stuff is over my head, but i did check what they are, and all are at their minimum values. and considering my nb runs hot anyways, i don't want to increase them.

so, at this point i'm really hoping putting on the stock heatsink solves the problem. if it doesn't, i think i have to live with it as it is until i am in a position to buy new parts.
 
i got the stock heatsink back and put it in place, and i still have 4gb ram working. i saw the techie today, and he is convinced the heatsink couldn't have anything to do with the problem, but i was hopeful.
he did say that the guarantee on my mobo is probably more than a year, in which case the next time i visit canada, i can then rma it.

also i suppose i can work on adjusting those other voltages, the "IOH, IOH/ICH PCI-E, ICH voltages as well as DRAM DATA/CTRL REF voltage tweaks for each memory Channel A/B/C" mentioned in the post linked to above by yellowbeard. but i don't know how to do that, and would definitely need help.

forgive me for posting all the slot swapping joy of my socket testing, but i don't really know anything about that, so i wanted the information to be up so someone could say if i did it wrong or my conclusions are wrong. as it stands, it seems there is a problem with communication with the b1 ram slot, although i suppose that could still mean a variety of different things.
 
Last edited:
Back