Official GIGABYTE Forum
Questions about GIGABYTE products => Motherboards with Intel processors => Topic started by: briligg on October 07, 2010, 03:59:01 pm
-
I just installed this motherboard yesterday, it replaced an Asus Rampage II Gene. I had the same ram issue with the Rampage and never resolved it. (Then the house got struck by lightening and the powerbar didn´t quite protect it.)
I haven´t run memtest on the ram recently (post-lightening) but the bios sees all three sticks. I can´t spend much time in bios because it makes the cpu heat up - it´s hit 89c twice after only a couple of minutes. Even during the boot process, the time spent in bios seems to heat the cpu, then it calms down once win7 is loaded. The LEDs for the cpu usually show green (over 60c) during early booting and often briefly hit red (over 80c). Also the LEDs for the north bridge show it running at full load during early boot. Once win7 is loaded, everest shows the cpu at 45 - 50c. And win7 says there is 6gb ram installed, 4 gb usable.
I updated the bios to FB - i don´t know what version, i downloaded it from the gigabyte site. That hasn´t helped. And at first the ethernet connection wasn´t working. I´m not sure what fixed that, at first i tried to run the ´smart lan´ tool in bios to see what it said, and it caused the bios to hang, twice. Then i tried enabling the ´onboard lan boot rom´. The connection still didn´t work when i rebooted, but then i went out, and when i came back and booted, it worked.
In the M.I.T. section of the bios, i went to a section that i think said ´current status´. It showed the ram setup. It showed slots 1, 3, and 5 occupied (which they are), but below that, it showed only slots 1 and 4 active. That´s how i remember it, since i can hardly spend any time in bios, i may not be remembering exactly the terms used.
I am about to go see if there is a more complete manual available for download. What do you think?
-
oh - system specs:
core i7 920
6gb ocz 1600mHz ddr3 (running at 1066mHz, that´s what the system chose)(OCZ3P1600LV6GK - pretty sure it´s on the list)
pny geforce gt 240 with 512mb gddr5
2 80gb sata drives, wd caviar - they´re old, but they do the trick
win7 x64
-
Hi and welcome to the Gigabyte Forum
Your manual is available below along with other useful items:
http://www.giga-byte.co.uk/products/product-page.aspx?pid=3449&dl=1#dl
You don't say what cooler you are running? If it is the stock one make sure that all four fixings are thoroughly home.
All the lights should illuminate on POSTing and then go out as booting proceeds.
6gb ocz 1600mHz ddr3 (running at 1066mHz, that´s what the system chose)(OCZ3P1600LV6GK - pretty sure it´s on the list)
You need to be more than pretty sure. ;)
-
Yes, it´s the stock cooler. I´ve taken it off and replaced it several times, during trials and tribulations with the Rampage, so i´m used to checking that the little black ´male´parts are sticking all the way out of the little white ´female´ parts. (tee hee.)
And i checked the list - the ram is on it.
-
Have you replaced the thermal compound each time you replaced the heatsink?
-
i got the manual, it´s the same as the printed one. i´d hoped it would be more complete.
The lights all go on at post as you say, but then throughout booting the cpu lights go off, and then start to come on again. When i say early boot, i mean before windows starts loading. Once windows is loading, the lights quickly go off.
Yes, it´s new compound each time. This time it was cheap stuff, i ran out of the coolermaster. But since the heating stops once windows has loaded, i have run the computer for a couple of hours in windows. Everest shows normal temps.
-
It's either that the temperatures are not being reported correctly or that there is something wrong with the type or the way you have applied the thermal compound. Did you clean the old off first? Some compounds take time to cure. Beyond that I cannot think of a way that it would start off hot and then cool down. It is contrary to normal usage.
-
yes, i cleaned off the old stuff each time, from both surfaces. What i used this time is white stuff, silicon grease i could buy here in town, the good stuff i have to order. rummaging around, i found a small tube of coolermaster, the kind that comes with one of their heatsinks. It should still have enough to replace the compound that´s there now.
The temperature LEDs haven´t come on while i was in windows, so it isn´t just everest. The side is off the case, so i always have my eye on them. Contrary to normal usage - yeah, tell me about it. My only thought was that it gets into some kind of loop and doesn´t get out of it again until windows has taken over. After all, at first it was hanging when i tried to run the ´smart lan´ tool.
A friend suggested removing the ram from slot 5 to see what happens. The manual says that isn´t an acceptable config for ram, but perhaps it would be useful to see if leaving just a stick in slot 1 makes a difference.
Then i want to return to the bios section that showed the ram slots occupied and the ram slots active. Is there a place to change that?
-
One thing to check is does the CPU fan start up and keep running as soon as you press the button? Also is it running at full speed?
-
Yep, it spins up right away. It also noticeably quietens once windows is running. Actually, i'm on it right now. I decided to pull the 2 dimm sticks and see what happened. I spent a little time in bios, not a lot. I went to the current status section and watched it for a bit. The temps were still climbing steadily, they'd reached 74c for core 0 by the time i left. I looked around a bit to see if you can adjust the active slots yourself, but didn't find anything. I was only in BIOS for maybe 3 min total. Probably less.
Since the machine was started from cold, i don't think there has been any improvement from yesterday, when BIOS got the cpu up to 89c. I downloaded an updated driver for the realtek sound chip and the lan chip. Perhaps installing those could help - it's a wild guess, but it can't hurt. There were a couple of other more recent drivers for raid arrays, but i'm not using one, so i suppose those wouldn't make any difference.
Right now everest says the cpu is 51c, and the fan is spinning at 1600 rpm. I've been in windows for maybe 15 min.
-
In that case it would seem to be that the fault lies with the BIOS not reporting the temperature correctly. The only way to be 100% sure would be to use an external meter to check the temps. I wouldn't say the sensor is faulty because it seems to work fine in Everest.
-
So, would that mean i should flash the BIOS again?
I powered down and replaced all the ram, in a different order this time - the sticks are numbered, because i went through this grief with the Rampage too, and it made memtesting them easier. Bios again shows the right three slots occupied, but only 2 slots active, slots 1 and 3. I guess that´s an improvement, it doesn´t show an empty slot as active any more.
But when i exited BIOS, the machine never managed to get into windows, after a few seconds it shut down and started rebooting instead. So i did a forced shutdown, figuring it was probably because of an overheating alarm.
Lacking an external meter, my hand will have to do. I should have checked the cpu heatsink right away, but i didn´t think of it. I´m letting it cool off for a bit right now, in case it really is overheating. When i get back into windows on it, i´ll update those 2 realtek drivers. From there, i don´t know what else to try. Except maybe flashing the BIOS again.
As this board is new, i can return it for a replacement within 7 days. After that, i can only get a repair. I have opened a tech support ticket, but of course the forums are a huge help, while i wait for feedback from tech support. I am in Mexico, a warranty repair would take a long time, if i need to use it. I put up with 4gb of ram on the Rampage instead of rma-ing it because that board was bought out-of-country, and an rma would have cost more than the board itself.
-
Right there are two reasons for the wrong reporting of the memory.
The first is the BIOS not reporting it correctly which can be cure by a BIOS update if one is available.
The second is if any of the pins are bent in the CPU socket. If you want to check that you must be very careful and will need a magnifying glass and good lighting. If you take a couple of Macro shots(close ups) of the socket we can check the pins for you.
I would suggest that you try all the available options and then if you are running out of time take the board back for replacement.
-
Well, i flashed again, and loaded optimized defaults, and installed the new realtek drivers.
Then the machine started crashing on startup when it reached a line saying ´recovering lost dram size´. It crashed, and then start booting up again, so i did a cold shutdown. I flashed bios one last time, and reloaded optimized defaults, but it did the same thing at the same place.
The store recommended i return the board, so i went ahead and took everything apart and packed it up. Just wanted to leave a complete record of what happened, in case it might be of use to someone else.
Thanks for your help, Dark Mantis.
-
Thank you for keeping us informed of the progress. Please let us know if everything goes ok with the new component. ;)
-
If you do not have any bent CPU socket pins then this is probably incorrect QPI/Vtt voltage. Are you 100% sure no pins are bent?
Are you setting your own QPI/Vtt and ram voltages, if so what voltages are you using? I see you mentioned 1066 memory speed, so you are probably leaving these voltages on auto as well right? That could be the issue, you should set everything yourself manually before going further. Set your ram voltage to spec, normally 1.66V for Triple channel, and then QPI/Vtt will need to be 1.25-1.35+ depending on your CPU and what memory speed and timings you set. Then set your Memory multiplier to whatever gives you stock spec ram speed, then set uncore multiplier manually to 2 x the ram multiplier.
It could also be the memory, have you tested each stick one by one with Memtest86+ for 5-10 full passes?
You should use this version, updated for X58A
http://www.memtest.org/download/beta/415b2/mt415.rar
-
Thanks for the memtest tip. I note that it´s a beta - do you find it stable?
The mobo went back to the store, so at the moment i am just waiting for a replacement. I had similar problems with undetected ram on my Rampage, though, so i have a couple of thoughts:
- at first the Rampage would randomly bsod, and it turned out it had to do with the memory settings, but not what you might think. I had to adjust back to back cas delay from auto to 8, and then the bsods stopped. i had my 6gb of ram running nicely then, but only for a few months, then it dropped to 4gb and nothing i could do changed that.
- but the 4gb i had worked fine. So i think that if you have poor settings on your memory, maybe that would cause instability, but i don´t see how it would cause the system to not see one of the installed sticks. How could 2 sticks be working fine and one not, if they all have the same settings?
- i have to admit i´m suspicious of this whole bent-pin thing. Our tech guy can´t believe that crossed pins would cause such a specific and rather minor problem as undetected ram. And if it did, why can´t anybody tell us the specific pin involved? Surely you can´t just cross any 2 pins and the result would always be undetected ram. It would have to be one very specific pin that causes that - if the idea makes any sense at all.
Of course, i can´t explain the various posts i´ve seen where people swear that straightening a pin fixed their memory problem.
But i do want to make an earnest plea, if someone finds a bent pin like that, please make a note of WHICH PIN IT WAS.
-
There is TONS of pins that affect memory, since the memory controller is now all in the CPU. Vtt, DDR3, and other ground pins, ect all are in the CPU Socket now. So if any one or more of them is bent you can loose channels of memory.
These are all noted in a diagram in the Intel PDF's, if you would like me to link you to those and tell you what pages you need to look at I can, let me know.
But to give you a short answer on that, yes it's possible and happens ALL THE TIME. I help in the Gigabyte USA Forum at Tweaktown, here, and several other overclock forums and I see this all the time.
This is not always the cause though, there can be many other causes, but this one does happen a lot more than you would think since the memory controller is now in the CPU bent pins are far worse now compared to things that happened in the older sockets. Here is a list of other possible causes by Eva2000
http://i4memory.com/ramdetect/
The main thing other than bent pins is incorrect QPI/Vtt and dram voltage, I can cause missing channels anytime on the fly with incorrect voltages.
Adjusting B2B CAS Delay slows the memory down, higher the value the slower the memory bandwidth is. So that could be why it worked for a bit for you, if you didn't have bent pins then maybe you had some voltages off and slowing down the memory some helped it work better with the incorrect voltages you were using.
-
Ah, fascinating info. I have asked before, and never gotten a response regarding the objections of our tech guy (who had me convinced). I was aware that the memory bus is on the cpu in i7, but i didn´t realize that translates into many of the pins being part of that. I guess for someone more tech-y that´s a ´duh´kind of thing, but to me ´bus´is just a word. I never thought it through.
In my quest to fix the Rampage, i did look for bent pins - posted macro photos and everything. I couldn´t see anything bent myself, and i didn´t get a response in the forum where i posted them, which i figured meant probably none were bent. And then i lost my faith in the idea anyhow, after talking to that tech guy.
The Rampage set my memory automatically to 1066mHz and undervolted it to 1.5v, and at first i tried to correct the bsods by changing the memory settings to the ram´s specs, both with the Rampage´s automatic utilities, and by hand. But it didn´t help. Then i tried slowing down the timings, but that didn´t help either. Finally i read somewhere that the Rampage II Gene´s auto setting for back to back cas delay causes problems in some cases, and changing it fixed the bsod problem completely. All the other ram timings were set to spec. When i started having problems again, it was a problem of undetected ram, but before that the system always detected the full 6gb.
I do believe i saw that Eva2000 article during my efforts to fix the Rampage, and going through it´s recommendations didn´t restore the 6gb in my case. I´m not an overclocker - i ended up with ram outside intel´s recommended spec because i bought a package deal. So perhaps i didn´t do enough tweaking, as i was really learning as i went. But i was conservative, i never risked damage to the hardware. I got a better cpu heatsink, even, just in case the fiddling with the ram stuff might raise temps there. The north bridge always ran hot and required its own fan too. The tech guy couldn´t fix it either.
If i continue to have problems, i may take you up on your offer to point me to the relevant pages of intel´s docs. Narrowing down the pin area to search has to help, right? I managed to determine that the A1 slot was the one offline on the Rampage, so if i had had that data available, i could have focused on a much smaller area of the pins when searching for the problem.
Thanks for the info. :)
-
To be honest if you can post a couple of GOOD quality close up photos of the socket we can check them for you anyway. ;)
-
thanks for that. Hopefully, the board i get back will just hum like a porsche, right out of the box, like i always dreamed when i bought myself an i7 system. :P
-
I´m thinking about the lightening strike now...
There was a crack and a flash like nothing i´ve ever experienced before. It freaked out the dog. A while later i went to turn on the computer, which had been turned off on account of the storm, as had it´s Belkin powerbar. Lights on the board showed it was receiving power, but nothing happened. When i took it to the shop, tech guy figured it was probably the psu, but it turned out it wasn´t. He replaced two fried electrical components on the board, but it still didn´t work. So i told him to forget it. Based on his testimony, i assumed the other parts of my system were ok. Now i wonder.
My husband´s computer, which was plugged into the same powerbar, was unaffected, as were all the other electrical appliances in the office, and the rest of the house. However the lamp-posts on our street didn´t work for over a week.
The wiring in this apartment isn´t grounded, and that isn´t unusual here. We are now thinking of installing a grounded outlet especially for the office. The powerbar still has its green LED lighted, saying it is protected, and also its red LED, saying it isn´t grounded. Over my years here, there have been several close lightening strikes and our computers were never affected before. It felt like that lightening bolt struck our very house, but i guess it didn´t.
Could there be some damage to other parts of my system? I didn´t get far enough with this last Gigabyte board to get into memtesting, but i know it needs doing. What about the PSU and the CPU? Are we right to specially install a grounded outlet?
-
The thing is with a lightning bolt or any other massive discharge of energy a EMP (Electro Magnetic Pulse) is emitted the same as when a nuclear explosion happens. Whether you have an earth or not wont help you in those circumstances and anything electronic will be fried. Well it would seem as if only your computer really caught it along with the streetlights. Often in these circumstances the pulse travels down the phone lines or whatever connection you have for the internet. Was there anything different about your computer as opposed to the rest of the electronics nearby? It does help(only help) if they are encased in metal.
The special ground wire is a good idea in other ways though anyway. It will need to go at least 1 metre into the ground though depending on the conductivity of the soil. If it is dry and sandy it would need to be much deeper. You will need local advice on the specs.
-
An EMP? gee, that´s sort of cool...
My computer was the closest thing to the router connected to the cable lines. My husband´s computer was also connected to the router, but mine was less than a foot away, while his was across the office. The tower case had one side off, because the heat sink was a little too large to close it. The router was on the other side, though.
The soil here is mostly clay and quite dense. In the dry season it does get very dry.
We have an antenna on the roof that is about 6 m tall, for a close-circuit radio. i wondered if it had been struck. It is the tallest thing in the immediate vicinity, but there is a taller antenna about 80m away. The radio hasn´t been used in some time, wasn´t plugged in, and hasn´t been checked since because the office has been reconfigured and its plug doesn´t reach any outlet. But at any right, i suppose if it had really been struck, there would be clear evidence.
-
Chances are if it had been struck directly you would be able to see the stars through your roof now :o If it was earthed and was sufficiently strong it would act like a lightning conductor and protected the house and a nearby area as it has a sort of umbrella effect. It's possible that the other higher antenna actually had this effect and saved your house. It is quite a complicated subject but interesting nonetheless.
Your computer probably took the worst of it because it was open and so not in a "Faraday Cage".
-
Surge protectors work by clamping the voltage spike to ground. So if the outlet does not have a ground the surge protector does nothing at all. It seems strange to me that yours shows a ground fault but still shows as protected. If possible get a grounded outlet. Surge protectors will not protect against EMP. A close lightning strike can create EMP so strong it can destroy electronics that are not even plugged in. A friend of mine just lost 3 PC's plus related equipment over the summer from EMP do to a close lightning strike. Lightning hit a tree in his front yard about 20 feet away from his house. All his equipment was protected by APC brand UPS. APC cut him a check for $2700.00 USD.
BTW in general an UPS gives much better protection against line spikes than a surge protector. This is because besides the spike being clamped to ground the UPS switches over to battery power. On a quality UPS this all happens in less than 10 ms. Nothing can protect electronics against EMP but EMP damage is rare while spike damage is much more common.
Bill
-
Surge protectors work by clamping the voltage spike to ground. So if the outlet does not have a ground the surge protector does nothing at all.
Good point Bill, I missed that one being so used to having an earthed situation.
-
I thought that if the powerbar is turned off, the circuit is broken and surges can´t reach our electronics.
-
It would keep your equiptment safe from normal mains surges and spikes but unless it was totally isolated (unplugged) it wouldn't stop a lighning strike or even a near miss as it would just bridge the gap between the contacts. If it is EMP it doesn't matter anyway as it doesn't need to be connected it instigates a charge in the device by induction.
-
Well, the store received the board on the 11th. After a couple of days i sent them an email asking how it was going. After a couple days more i sent another. They didn´t respond, so finally yesterday i called them. They said they sent the board to Gigabyte. I said fine, but your guarantee says that i am entitled to a replacement, because i returned the board in less than 7 working days. Then they said that there was a small bang on one corner of the board, so they would have to wait and see what Gigabyte said. Call Thursday.
That board was in perfect condition when i sent it off, there is no way there was a bang. I treated it with kid gloves, it was my baby. Well, my 2nd baby, after the Rampage died. It was well-packed, and they signed off on the delivery. I find it very suspicious that the man i spoke to seemed to know nothing about that board until i mentioned the guarantee, and then suddenly he knew it had a bang on it.
I went back to the tech support ticket i have, and asked if the board could be tracked through its serial number. The guy on my ticket gave me a couple of email addresses where i could make that request. I´m waiting on if there is anything they can tell me. I let them know my concern about the store. I imagine there is nothing they can do about that, but if they give me what i need to establish that there was no physical damage to the board, perhaps i can still get my replacement board. My tech support ticket is through the Mexico portal, so there might be some extra steps to the process, maybe.
So, now that i´ve vented, do you know if there is any way i can use my guarantee directly with Gigabyte? The rma department should have the board, and i can prove my ownership. If the store continues to avoid their guarantee, i´d like to have the option of pursuing that.
-
This sort of thing is always a problem when there is a third party involved. I should think that as you can prove ownership you would have a good case for contacting Gigabyte directly and see what they say.
-
Well, tech support and sales are looking into it, so that´s hopeful.
I know this forum isn´t for this kind of thing. Thanks for your 2 cents. I guess it is a good sign that inquiries like this don´t crop up here.
-
No, Gigabyte, like any large company have their faults but on the whole make good products and treat their customers well. ;)
-
Well, now i have my new motherboard, and it is working like a dream. All the ram shows up, cpu is 59c after some time working in photoshop testing it out a little, fan at 1500 rpm. Until i did that, the cpu was at 47c and its fan was spinning at 1200rpm. Life is good.
My theory about the whole 'banged corner' spiel the store rep got into was that he didn't want to go into the whole thing with me about how they don't diagnose things themselves, they just send them on, and because of the border Gigabyte wasn't going to get the rma for 2 weeks, and they weren't going to send a replacement until Gigabyte gave the ok. Rather than just saying that, the rep decided to pull out an all-purpose excuse. I sure wish he hadn't. Anyhow, as soon as Gigabyte had the board, the store sent out a replacement.
One last thing, i have to say i've received much better tech support in this forum than i did in the Asus forums. Quicker responses, better information, just generally better. ;D
-
Yes well I brought the experiences from the Asus forums with me when I came here from them and tried to only use the good parts ;) I am still trying to improve on a few things but we are getting there slowly and there is a good team here. Glad you enjoyed the visit and got everything sorted. 8)