Jump to content
Corsair Community

TW3X4G1333C9DHX - only stable when underclocked.


lalittle

Recommended Posts

I upgraded my Asus P5E3 Deluxe WiFi from 2GB to 4GB a few days ago. I went from TWIN3X2G1333C9DHX sticks to TW3X4G1333C9DHX sticks. I started having stability issues shortly after the upgrade where the system would spontaneously reboot during various tasks. It wasn't very consistent, but it would never go more than a few hours without this happening, and sometimes it would happen before the system had even finished the last reboot. It also seemed to get worse over time, but I was also using the system extensively for video editing, which may have just been stressing it more.

 

I ended up switching the BIOS settings back to "Auto" for both the frequency and the voltage (the two settings I needed to change to get it to work at the "intended" settings), at which point the ram switched back to it's "default" setting of 1066 rather than the official 1333 that these sticks are designed for. Since doing this, I have not had a crash for a full day now, including plenty more of the same video editing that was giving me so much trouble before.

 

Note that I am not overclocking anything on this system, and I did set the voltage to 1.6v, which was printed on the module labels, so I was using the correct voltage. Note also that the previous sticks (2GB versions of the same line of modules) worked at 1333 for months in the same MB without ever having this issue.

 

Given that it's been stable for a day now, along with the fact that the previous modules worked at their full 1333 speed without any issues, does this sound like a RAM issue? The processor is a Core 2 Duo 3Gig, and I'm using a 700w Silverstone OP700 power supply.

 

Thanks,

 

Larry

Link to comment
Share on other sites

Larry, I am having the same issue (but my motherboard is ASUS P5Q3 Deluxe WIFI) -

 

I had this ram before, and found out one of the DIMMs (stick) worked and the other one was faulty, so I rma'ed them.

 

I then got another kit, and tested each DIMM individually with memtest plus http://www.memtest.org/ and they passed 8 runs through memtest - so the individual DIMMS are ok this time. I then installed both DIMMS in the orange slots. I've been getting spontaneous lockups like you.

 

The moment I remove one of the DIMMS (have tried both of them, one at a time), the system becomes stable.

 

I haven't tried going to defaults (e.g. 1066 instead of 1333) , but I reckon it will work. But that's not really a solution tbh.

 

I've tried bumping up the Northbridge and Memory voltages, and I am now checking again using memtest with BOTH dimms.

Link to comment
Share on other sites

  • Corsair Employees

neo_x

Should not be a problem with your configuration and just make sure that you load setup defaults and then I would set the memory Voltage to 1.7 Volts and set the memory frequency at DDR1333 and set the timings to Cas 9-9-9-24 2T Command Rate and if it is still not stable move the modules to slots 2-4 and go through loading setup defaults and setting the settings.

 

lalittle

In order to run DDR1333 you need to be using a 1333 FSB CPU and if you have a E6000 series CPU it will be limited to DDR1066.

 

And for both of you if you had 2.0 Gig in the system when you installed it you need to make sure you have SP1 installed if it is Vista and I would try to uninstall the Video Driver and then re-start and re-install it.

Link to comment
Share on other sites

Sorry RAM GUY I won't hijack this thread and instead open up another one - will do the testing and get back to you in the other thread. Just posted on here as I thought we both had the same problem.

 

Thanks

Link to comment
Share on other sites

The X38/X48 is problematic with high density memory modules. Many people are having the same issues of 128M X 16 = 2GB modules.

 

The issue is the memory controller. This is shown by the fact that the memory passes memtest. The CPU passes LinX/Prime95 as well. The data stream of CPU <--> Memory Controller <--> Memory shows the issue isolated to the Memory Controller.

 

I have personally had this issue with an X38 (Maximus Formula) and X48 (Rampage Extreme). My result was to remove the 2GB modules and insert 4 X 1024 modules. The issue ceased. The other method was to drop the speed of the Maximus Formula 1066Mhz memory (2GB Modules) to 800Mhz and the Rampage Extreme (1600C7 Modules) to 1333Mhz. I refused to do that. The new i7 (X58 Chipset) do not have this issue, even when tested with the same memory.

Link to comment
Share on other sites

lalittle

In order to run DDR1333 you need to be using a 1333 FSB CPU and if you have a E6000 series CPU it will be limited to DDR1066.

 

It's an E8400 CPU, which has a 1333 FSB, so this isn't the issue. Note that it worked fine with the previous 1GB modules (2GB total) which were the 2GB version of the same series from Corsair (TWIN3X2G1333C9DHX.)

 

And for both of you if you had 2.0 Gig in the system when you installed it you need to make sure you have SP1 installed if it is Vista and I would try to uninstall the Video Driver and then re-start and re-install it.

 

I'm running XP Pro XP3.

 

Thanks,

 

Larry

Link to comment
Share on other sites

The X38/X48 is problematic with high density memory modules. Many people are having the same issues of 128M X 16 = 2GB modules.

 

The issue is the memory controller. This is shown by the fact that the memory passes memtest. The CPU passes LinX/Prime95 as well. The data stream of CPU <--> Memory Controller <--> Memory shows the issue isolated to the Memory Controller.

 

Could you clarify this? I'm not clear why this points to the controller. Also, isn't the controller used the same way with memtest that it is with normal use?

 

I have personally had this issue with an X38 (Maximus Formula) and X48 (Rampage Extreme). My result was to remove the 2GB modules and insert 4 X 1024 modules. The issue ceased.

 

That's odd because elsewhere on these forums the Corsair rep specifically warns against using four sticks since this can cause instability (i.e. the opposite of what you're saying.) The recommendation is to always use only 2 sticks.

 

Thanks for the feedback,

 

Larry

Link to comment
Share on other sites

I have some followup questions I realize that some of these responses were targeted at neo_x, but I thought they might apply to me as well.

 

just make sure that you load setup defaults and then...

 

1) Is there a difference between loading the defaults first and just starting with everything on "Auto"? Loading the defaults involves having to go through ALL the BIOS settings and resetting everything up again, which can be a bit of a pain with all the settings these days (SATA/AHCI/RAID settings, on-board devices, boot config, etc.)

 

If we start by saving all the settings on the RAM and CPU page as "Auto," can we just change the specific settings needed, or is there some other benefit to first loading defaults.

 

I would set the memory Voltage to 1.7 Volts and set the memory frequency at DDR1333 and set the timings to Cas 9-9-9-24 2T Command Rate

 

2) My voltage is listed as 1.6 volts on the sticks, and neo_x didn't specify what his modules say. Were you suggesting to neo_x to run at 1.7v even if the modules said 1.6v?

 

3) In another thread I was told that making only two changes in the BIOS was needed: Frequency from Auto to "DDR3-1333" and voltage from Auto to "1.6." The system sets 9-9-9-24 2T automatically when I do this. Is this still okay?

 

and if it is still not stable move the modules to slots 2-4 and go through loading setup defaults and setting the settings.

 

3) I was told in another thread that the orange slots were the better ones to use with 1333 RAM with the P5E3 Deluxe. Would there be some reason to use the brown slots, and if so, wouldn't this go against the recommendations for best performance?

 

Thanks again for you help,

 

Larry

Link to comment
Share on other sites

  • Corsair Employees

Larry

If the system is not stable at 1.6 Volts I would try it at 1.7 up to a max of 1.9 Volts. But you should not need to go more than about 1.8 with these modules.

And the memory frequency should be set manually to DDR1333 and test with memtest.org if its stable in memtest but you still have Windows issues try and uninstall the Video driver and use the default driver and see if you still have problems.

Link to comment
Share on other sites

Could you clarify this? I'm not clear why this points to the controller. Also, isn't the controller used the same way with memtest that it is with normal use?

 

If you pass both memtest and the CPU testing, and still experience blue screens, etc. the issue is not the CPU or memory and there is only one external left in the isolation, that being the memory controller.

 

That's odd because elsewhere on these forums the Corsair rep specifically warns against using four sticks since this can cause instability (i.e. the opposite of what you're saying.) The recommendation is to always use only 2 sticks.

 

And I agree with that recommendation except in this case where I have found that the 2GB density modules in X38/X48 are problematic and in alleviating this issue for myself and others has proven out that the 4 X 1024 is the best route.

 

I'm just giving you my personal knowledge in the system integration of these chipsets and high density memory.

Link to comment
Share on other sites

Does my particular symptom indicate anything -- i.e. the fact that I'm not getting blue screens or lock-ups, but rather that the system just spontaneously reboots?

 

Regarding the voltage, what is the theory behind this possibly helping, and is there any potential downside to using voltages higher than 1.6? I feel kind of weird using a voltage higher than what it says on the modules since this clearly indicates what I'm "supposed" to use.

 

Regarding the slots, is there any reason to think that the brown slots would work better than the orange with the stock 1333 speed?

 

I just created a memtest cd and am using it. Is the idea to just "let it run" for a while and see if you get any errors? How long (how much time and how many passes) should I let it run to get an indication of whether or not I have problems? Should I just let it go overnight, or...? I'm not clear if it's okay to just let it go for a while, or if it puts too much stress on things to let it go for hours and hours. I don't want to cause premature failure down the road by testing it "too much," but I'm not clear if this is even possible.

 

Thanks again for all the help here,

 

Larry

Link to comment
Share on other sites

  • Corsair Employees

We really need you to do bit more testing to get a better idea.

But typically random reboots with no error code logged in the Event Log might suggest a Voltage or thermal issue with the MB or chipset and testing the components in another system is the best approach at trouble shooting.

IE it could be cause by some issue with the CPU, or there may be a chipset or memory controller problem or like I said a Voltage problem on the MB or maybe heat related. What are the temps like? when it reboots go to BIOS and check them right away.

Link to comment
Share on other sites

Does my particular symptom indicate anything -- i.e. the fact that I'm not getting blue screens or lock-ups, but rather that the system just spontaneously reboots?

 

Regarding the voltage, what is the theory behind this possibly helping, and is there any potential downside to using voltages higher than 1.6? I feel kind of weird using a voltage higher than what it says on the modules since this clearly indicates what I'm "supposed" to use.

 

That's the tested value. Some boards need less, some more. You are warranted for up to and including 2.0v for that memory, although anything greater than 1.8v will not be necessary.

 

Regarding the slots, is there any reason to think that the brown slots would work better than the orange with the stock 1333 speed?

 

Not that I know of but hey, whatever works.

 

I just created a memtest cd and am using it. Is the idea to just "let it run" for a while and see if you get any errors? How long (how much time and how many passes) should I let it run to get an indication of whether or not I have problems? Should I just let it go overnight, or...? I'm not clear if it's okay to just let it go for a while, or if it puts too much stress on things to let it go for hours and hours. I don't want to cause premature failure down the road by testing it "too much," but I'm not clear if this is even possible.

 

I like two full passes. One pass for stability and another to be certain the first pass is not an anomaly.

Link to comment
Share on other sites

What are the temps like? when it reboots go to BIOS and check them right away.

 

I actually did do this, and I never saw any temps that seemed particularly problematic. MB typically in the low 30's ©, CPU mid 30's. When things get hot, both can get up around 40C.

 

Larry

Link to comment
Share on other sites

You can take a look at the hardware monitor portion of your BIOS and adjust the monitoring and alerts. This can also be done in ASUS Probe. You can set temporarily set the warning temperatures a bit lower than stock and that will let you know if you have any temps jumping up.
Link to comment
Share on other sites

I just got an error in memtest at 95%. Here are the results:

 

Test: 8

Pass: 0

Failing Address: 0008e505f7c - 2277.3MB

Good: 6fbdcefa

Bad: 7fbdcefa

Err-Bits: 10000000

Count: 1

 

Does this show anything significant? Does this indicate that this issue is directly related to the RAM and not the other possible issues mentioned above? Would the next step be to test at 1.7v, and is there ANY potential downside to this? I thought that using higher voltages put more stress on the MB, and that this was why the "1.7" setting gives the yellow "warning" color in the BIOS.

 

RAM GUY - One clarification about temperatures: Given that the previous modules (2GB TWIN3X2048-1333C9DHX) worked fine in the same situation at the same temps, does this essentially eliminate any temp issues involving any components OTHER than the RAM itself? It seems like this would be the case.

 

Thanks again for bearing with me in all this,

 

Larry

Link to comment
Share on other sites

When you ran Memtest, did you have Legacy USB Support disabled? This setting conflicts with Memtest and must be disabled.

 

It was on "Auto." I'll test again with it disabled.

 

One question: I was looking at the BIOS hardware monitor screen and it just went black, which I've never seen before. Can this be caused by RAM issues as well (i.e. bad modules, or too low voltage, or...)? I've had this system for several months and I've never seen any of these things before a few days ago when I put in the new RAM.

 

Thanks,

 

Larry

 

PS. The temps are mid 30s -- MB 36 to 37C, CPU 35 to 36C.

Link to comment
Share on other sites

Such responses, are indicative of future issues with some part of the system. When finding such an event, I isolate the video card as the issue, especially so if I can repeat this loss of screen.

 

I would test the modules singly, and the slots singly. I would test all four slots and two dual channels if you want to be certain of the memory controller's throughput. Keep in mind that memory is passive and will not damage components. It can be damaged by components, but can not damage components, although it can corrupt files.

 

I would:

 

Shut the system down. Boot to the Memtest CD with stick one in slot 1 (Closest to the CPU) and allow for two full passes. Shut the system down. Remove the first stick and insert it in slot two. Repeat memtest. Shut the system down. Remove the first stick and insert it into slot three and repeat memtest. Do the same for the last slot. You have now tested one stick and all four slots. Shut the system down and remove the first stick. Insert the second stick in slot one. Retest. Shut the system down and remove the second stick. You have now tested both sticks sticks singly.

 

This shows that the memory is good.

 

Now test for the Dual Channel Mode function of the memory controller. Insert a stick in slot one and three and memtest. Remove the set and insert it into slot two and slot four and repeat Memtest.

 

Results in the form as below please.

 

Single Mode

  • Stick 1 - Slot 1 = Pass/Fail
  • Stick 1 - Slot 2 = Pass/Fail
  • Stick 1 - Slot 3 = Pass/Fail
  • Stick 1 - Slot 4 = Pass/Fail
     
  • Stick 2 - Slot 1 = Pass/Fail

Dual Mode

 

  • Stick 1 & 2 - Slot 1 & 3 = Pass/Fail
  • Stick 1 & 2 - Slot 2 & 4 = Pass/Fail

 

 

 

 

It was on "Auto." I'll test again with it disabled.

 

One question: I was looking at the BIOS hardware monitor screen and it just went black, which I've never seen before. Can this be caused by RAM issues as well (i.e. bad modules, or too low voltage, or...)? I've had this system for several months and I've never seen any of these things before a few days ago when I put in the new RAM.

 

Thanks,

 

Larry

 

PS. The temps are mid 30s -- MB 36 to 37C, CPU 35 to 36C.

Link to comment
Share on other sites

Such responses, are indicative of future issues with some part of the system. When finding such an event, I isolate the video card as the issue, especially so if I can repeat this loss of screen.

 

Is it "possible" for a RAM issue to cause this, however, or is this something that a RAM issue would never do? I'm just trying to fully understand the possibilities here.

 

Also, if this was a RAM-related issue, would hotter temps be more likely to show failures even if those temps were still within acceptable limits? I just got through 4 passes without any errors, but along with turning off Legacy USB support I also I turned up the AC while I was away and I noticed that the temps were noticeably lower (30-32C for MB and CPU instead of the mid/high 30s like I had when I got the error.)

 

I guess the important question is this: If temperature effects this issue, could a higher DRAM voltage still fix it, or would a correlation between temperature and failures indicate that voltage was not the issue?

 

On this note, aren't my highest temps (around 40C for CPU and MB) still WELL within acceptable limits?

 

Thanks,

 

Larry

Link to comment
Share on other sites

One other question:

 

I've been searching for the specific symptoms of how "Legacy USB Support" interferes with Memtest, and so far the information I've found on this is that Memtest itself actually has problems "running" with Legacy USB Support enabled, but NOT that it can display errors that aren't really there. Can someone confirm that having Legacy USB Support enabled can actually cause Memtest to display RAM errors?

 

Thanks,

 

Larry

Link to comment
Share on other sites

  • Corsair Employees
Can someone confirm that having Legacy USB Support enabled can actually cause Memtest to display RAM errors?

 

Thanks,

 

Larry

Yes, I can. I have tested it on literally dozens of different models of ASUS motherboards stretching all the way back to the 845 chipset boards over the past 7-8 years.

Link to comment
Share on other sites

Yes, I can. I have tested it on literally dozens of different models of ASUS motherboards stretching all the way back to the 845 chipset boards over the past 7-8 years.

 

Thanks for the confirmation.

 

After having no reboots for 2 days at 1066, I went back to 1333 for the memtests and I left it there when I reloaded windows a few hours ago. I just had another spontaneous reboot, which would seem to confirm that 1066 vs 1333 is the critical issue.

 

The main thing I find confusing, however, is that these were installed on the 13th, and the problems didn't start happening until the 17th (4 days later.) Everything seemed fine during that time.

 

I switched the modules to the black slots to see if this makes a difference -- I had been using the black slots for the 1GB modules. I read a few reports of the black slots being more "stable" on this MB, one theory suggesting that the signal to noise ratio was higher on these slots, which made them better for overclocking, and could theoretically make them more stable in general.

 

I'll post back with any findings, but I'm still very interested to hear any further feedback on the situation so far.

 

Thanks,

 

Larry

Link to comment
Share on other sites

Update: Moving the modules to the black slots made no difference -- I got another spontaneous reboot after just a little while.

 

I need to use this system for my work, so I had to get it stable again. Instead of dropping down to 1066, I put the 1GB sticks back in and set the BIOS back to all "Auto" (since my TWIN3X2048-1333C9DHX sticks use the default 1.5v.)

 

Hopefully this will be stable, and it will tell me whether or not anything else "went bad" lately that might be causing this.

 

Larry

Link to comment
Share on other sites

It's not a common event so I am not certain, but the testing method I gave you will ensure that it is not the memory and then we can move to isolation of other parts.

 

That is if you wish to do the work.

 

Is it "possible" for a RAM issue to cause this, however, or is this something that a RAM issue would never do? I'm just trying to fully understand the possibilities here.
Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...