GA-X58A-UD3R memory settings question

[Weps]

Hello,

I have a GA-X58A-UD3R with an I7 950 and a Corsair 6GB Triple Channel DDR3 Kit (TR3X6G1333C9) (link here for the kit: http://www.corsair.com/catalogsearch/result/?q=tr3x6g1333c9 ). This kit is supposed to work at 1333Mhz with 9-9-9-24 timings as per their site description.

Due to BSOD problems I am having at the moment, I have decided to force the BIOS RAM settings to those recommended values as a first troubleshooting step. The BIOS version is FB, and the screens are not the same as those described in the motherboard manual. I have already fiddled with the BCLK and memory multiplier to reach 1333Mhz but I have a question about the timings.

When I go into the Intelligent Tweaker > M.I.T. Current Status the following timings are displayed:

tCL tRCD tRP tRAS
9 9 9 24

However, when I switch to the Advanced Memory Settings section and look at the Channel A,B,C Timing Settings I see

CAS Latency Time 8
tRCD 8
tRP 8
tRAS 20

This happens both when I set the DRAM Timing Selectable (SPD) to [Auto] or [Quick] (so that the timing settings become adjustable) and save the settings.

So my question is, which is correct? The display in the Current Status, or the Settings display in the Advanced Memory Settings? Are those settings perhaps overriden depending on what frequencies I chose? All I need is to have the RAM work at the recommended 9-9-9-24 and I'm not sure if it is.

Thanks in advance,
Weps

 

[jaquith]

Welcome to Tom's Forums!

ref - http://www2.corsair.com/configurator/product_results.aspx?id=1362102

The kit you're describing: NOT XMP, 1333 MHz 9-9-9-24-2T @ 1.5v

BIOS:
Extreme Memory Profile (X.M.P.) -> Disabled
System Memory Multiplier -> 10 {assuming BCLK is default @ 133 MHz}
DRAM Timing Selectable (SPD) -> Quick
/9-9-9-24
/Command Rate(CMD) -> 2

* since the kit is low voltage I doubt the DRAM {DDR} Voltage -> 1.5~1.55v or QPI/DRAM {QPI} Voltage ->1.2~1.3v needs to be touched.

--
This is considered, in part, OC'ing your RAM above spec.
CAS Latency Time 8
tRCD 8
tRP 8
tRAS 20

 

[jaquith]

Footnote - If you plan to OC the CPU then I would set: DRAM {DDR} Voltage -> 1.5~1.55v or QPI/DRAM {QPI} Voltage ->1.2~1.3v

 

[Weps]

Thanks a lot for the answer, it seems I was a dope and misunderstood the column settings. There is actually a column with the values I mentioned and another column in which you can set the actual timings (default [Auto] on all timings). I assume the first column just displays what the BIOS reads from the SPD.

Anyway, I set the timings as instructed. I didn't know about the command rate so thanks from bringing that up too. Now I'll wait and see if remains stable.

 

[jaquith]

Good Luck! Let me/us know.

 

[Weps]

I received a couple of blue screens this morning and when looking into the M.I.T. again I noticed the DRAM Voltage indicator to 1.584V. So I went into Advanced Voltage Control and set the DRAM Voltage setting there to 1.500V. However, I didn't understand what QPI setting you were referring to.

I see two QPI settings, under CPU I have QPI/Vtt Voltage with a "normal" of 1.175V and under MCH/ICH I have QPI PLL with 1.100V, both on [Auto].

Again, I am confused about this "Normal" column which shows values in front of the actual column where you can set them in. Are those actually measured values, or are they recommendations calculated/read based on what components you have? Do they represent the system status or are just recommended values?

Which of the above settings should I set above 1.2V?

I also ran Memtest86+ twice after setting the DRAM Voltage, first time 3 passes, second time 2 passes, no errors.

 

[jaquith]

QPI/Vtt = QPI/DRAM ; it's the memory controller voltage, and unless needed to run stable then leave it as Auto.

Leave the MCH/ICH & QPI PLL as Auto.

 

[Weps]

Sorry to be a nuisance but my problems continue. Is it ok if I keep asking here or should I start a new topci elswhere? Anyway, I have also done some searching and found the problem documented here and there, with different RAM products amongst which also mine, I can post a topic from another forum is anyone is interested in additional reading.

I have not mentioned the whole story, here it is now:

After assembling the system under BIOS version FA I was having the 4GB out of 6GB available problem which I solved by upgrading to version FB. Then the system worked ok for a month and a half until I started receiving random BSODs.

Some of the people confronted with this problem report the BSODs occurring on a cold boot. I would say that 50% of mine occur on a cold boot and the rest during normal operation. I got at least one every week since this started, but most of the time the system boots fine. Eventually though one cold boot BSOD busts the OS beyond repair and I have to reinstall.

Since my last post I have done the following (moving on to the next step after receiving at least one BSOD):

1. Upgraded to BIOS FE - left default timings+voltage
2. Manually set up the above mentioned timings+voltage
3. After last round of BSODs I could not power down correctly, when looking in the M.I.T Current Status I received the "The system has experienced boot failures because of overclocking or changes of voltages. Last settings in this page may not coincide with current H/W states". I loaded optimized defaults, and changed just the ICH control mode to AHCI. Memory settings are for now back on [Auto] at 1066Mhz Memory frequency, reinstalled OS.

I have also ran a new Memtest86+ this time with just one dimm each in its own slot for 5 passes and got no errors.

From what I could gather from other topics on the Corsair and Gigabyte forums, the next steps would be to fiddle with the QPI/VTT upwards, or check for bent pins / improper contacts on the socket, especially since this is related to the 4GB/6GB issue (btw. some report the BSODs going away if they reflash to FA and then back again with newer versions - I haven't tested this though).

So what should I start with? The lazy me would prefer the voltage since I gets me out of going to buy new paste for the cooler. Should I just set it to a higher value one step a a time and wait to see if the problem occurs again?

If I get the cpu out, is there anything else I should mind? Since I consider the amplitude of my symptoms somewhat smaller as in other cases, and should I not find any bent pins, should I at least wipe the pads on the cpu with some alcohol to make sure those are clean?

Another thing is that most I read about reported the problem solved by changing RAM brands (or at least models).

Thanks for reading.

 

[jaquith]

Question - do you have an SSD, and how are your HDD configured?

I've found the QPI/VTT not to help anything, and values above 1.3v to make the BSOD worst.

 

[Weps]

No, I have a single SATA2 Seagate 1TB HDD. This, along with a DVDRW and two eSATA and HDD dock connections from the case are plugged into SATA2_0, SATA2_1, SATA2_4 and SATA2_5 respectively. ICH SATA is set to AHCI, and GSATA 6_7 and GSATSA 8_9 Controllers are Disabled.

PS: later today I got back in the room and found the PC shut down, although I had left it running (PC is set not to restart on BSOD). Upon restart I received an "C0000221 unknown hard error \SystemRoot\System32\ntdll.dll" error followed by BSOD during OS repair. I will try to reinstall when I get back from work and this time set the ICH Controller to IDE to see if this makes a difference.

 

[jaquith]

MS -> http://support.microsoft.com/kb/314474

Are you running XP? Also, if you restarted the PC with AHCI then it changed to IDE and visa versa then that can cause another corruption.

MS -> http://support.microsoft.com/kb/922976

If you are running XP then here's a Clean XP SP3 ISO from Microsoft -> http://www.microsoft.com/downloads/en/details.aspx?FamilyID=2fcde6ce-b5fb-4488-8c50-fe22559d164e&displaylang=en

 

[Weps]

Hello again, just wanted to follow-up on this. IDE made no change, and I am using Windows 7.

Two weeks ago however I took the CPU out, cleaned the pads with some alcohol a little (don't think it did much, but still) and since then I have not received any BSOD. I know I am not out of the woods yet, but before this I was getting at least one BSOD a week, so I consider this an improvement.

Did not notice any bent pins though, but when placing the CPU I had the feeling it did not "drop" perfectly on its final position and that it "rubbed" a little against the pins before settling down. Also, I noticed some marks on the CPU pads which were missing on some of them or off center on others. I don't know if those are supposed to be contact marks or not.

Anyway, I want to thank jaquith for his patience in answering each time even if the issue turned out not to be settings related, and I'll let you know if problems arise again.

 

[Weps]

I don't like to necro threads and apologize for the inconvenience. I just want to close this issue for the sake of other people searching for similar issues.

It seems it was the RAM. After running with just 4GB for a while for stability I decided to buy a new 6GB Kingston kit which was in the QVL. I've been using it for more than 3 months and did not have a single BSOD yet.

The vendor of the original RAM said it was nothing wrong with it after I took it in for analysis. I did manage to get a single memtest error out of a single dimm after running it 30hours+ (around 30 passes). Don't know if the original RAM is fiddly or if there is some incompatibility with this board but I tend to believe it's the latter.