Memory not being recognized properly

[gabrielgpm]

Hello,

I am having a strange problem since I changed from 6GB OCZ Platinum to 6GB G.Skill Trident. Sometimes on the first boot of the day instead of the 6GB BIOS only detects 4GB. Strangely this problem is overcome by shutting down and restarting the PC once or twice.

I am using 2x3GB G.Skill Trident PC3-1600MHz CL6 DDR3 F3-12800CL6T-6GBTD memory, Asus P6T Deluxe V2 motherboard and latest BIOS version (0704). Settings are X.M.P. and the CPU is overclocked to 4.0 GHz but the same problem happens with CPU at default settings.

I have tried changing positions of DIMMs and also tried changing some BIOS settings from Auto to fixed values without success. Is there anything else I can do as I would like having all the memory working correctly all the time?

[Tradesman]

You'll want the sticks in slots 1-3-5, the black ones, and since you mentioned XMP would suggest loading base timing, et all manually, 6-7-6-18, command rate at 2N (or 2T depending on how it's shown) and the DRAM Voltage at 1.6. Also may want to check the sticks individually in slot 1 w/ Memtest to make sure you don't have a bad stick and since it's an ASUS P6T Deluxe you might possibly want to raise trhe DRAM Voltage to 1.63 or 1.65 (with XMP enabled the mobo doesn't always read voltage true)

[gabrielgpm]

Thanks for your reply. Actually I have the modules in the orange slots which I think are the correct ones for 3 DIMMs on Asus P6T Deluxe V2 (Sockets DIMM_A1 DIMM_B1 and DIMM_C1). As additional information, when I load X.M.P, Asus TurboV reports QPI/DRAM Voltage as 1.375 and DRAM Voltage 1.66 and I changed manually the Command Rate from 2N to 1N with no influence on memory detection. Anyway, I?ll try loading all the timings you suggest manually.

I have already tried different DRAM Voltages from 1.66 to 1.60 with no improvement on the frequency of detection of the 6 GB. Regarding your suggestion on a possible bad stick, I have already done several testes with memtest and prime95 with the 3 DIMMs inserted with no errors, but I?ll run memtest again with each stick individually in Socket A1 and report back.

[gabrielgpm]

Changing from X.M.P. to Manual didn't work. I chose Ai Overclocker Tuner [Manual] and typed the timings 6-7-6-18-2N, set DRAM voltage to 1.6 and QPI/DRAM Voltage to 1.375. Next morning the PC booted with 4GB again and I had to shutdown and restart the PC three times to get the full 6GB showing on BIOS.

One thing I noticed is that DRAM Round Trip Latency (at 2nd Information) on CHA A, CHA B and CHA C which are all set to [Auto], keeps changing on each boot, is this normal?

[Tradesman]

Sounds like it may not be reading the sticks correctly.....may try running memtest on each stick individually also may want to up the CPU-CPU/MC voltage a tad.

[gabrielgpm]

Sounds like it may not be reading the sticks correctly.....may try running memtest on each stick individually also may want to up the CPU-CPU/MC voltage a tad.

I switched Ai Overclocker Tuner back to [X.M.P] (from [Manual] - the timings I was using previously were 6-7-6-18, sometimes with Command Rate (CR) 2N and other times with CR 1N) and ran Memtest86+ V4.0 from a boot CD for each Module individually in socket A1 and it passed with no errors. The tests weren't done overnight, just a couple of minutes for each module (2 passes). I will run a longer memtest on each stick individually if you think it is necessary, but as I already have done an overnight run with all 3 sticks in sockets A1, B1 and C1 with no errors, they appear to be OK once they are detected.

One thing that I noticed this time which was also unexpected is that X.M.P is giving a Command Rate of 1N instead of 2N as it originally used to and as it is shown by CPUID SPD X.M.P. (2T). I checked BIOS a couple of times now and verified that CR is set to [Auto] and showing 1N all the time with X.M.P. Shouldn't it revert back to 2N?

The detection of 4GB instead of 6GB continues to occur at the first boot of the day. Once 6GB are detected I can shutdown or restart the PC many times and 6GB are always detected. Seems that modules need to "warm up" for the detection to occur correctly.

I didn't have this problem with my previous modules (6GB OCZ Platinium 1600 CL7) and if these G.Skill modules are "okay", is it possible I having compatibility issues with these Trident G.Skill modules and the motherboard, or even between the three sticks?

[DaKo]

Gabrielgpm, I have the exact problem as you on a Gigabyte X58A-UD7 with 6 GB G.SKILL F3-12800CL9T-6GBNQ (as written in this thread: http://gskill.us/forum/showthread.php?t=3198), but I haven't been able to solve it yet. A member of the German/European G.Skill team adviced me to set the settings manually (instead of using the XMP profile), but I haven't tested it yet. I hope this problem can be solved somehow, I built my new computer two weeks ago in order to not have any problems anymore .

[gabrielgpm]

Hi again,

I cleared Real Time Clock RAM in CMOS and set X.M.P. in BIOS - Command Rate is now back to 2N. I'll be trying new tests with everything at default except X.M.P. from now.

Suggestions are welcome. These are my current settings:

____________________________________

values after brackets are from CPU-Z on Vista 64 bits

Ai Overclock Tuner ................. [X.M.P]
CPU Ratio Setting .................. [Auto]
BCLK Frequency ...................... [Auto] 133.6
PCIE Frequency ..................... [Auto] 100
DRAM Frequency ..................... [Auto] 1603mhz
UCLK Frequency ..................... [Auto] 3207
QPI Link Data Rate ................. [Auto]

____________________________________

DRAM Timing Control

CAS# Latency : 6
RAS# to CAS# Delay : 7
RAS# Precharge : 6
RAS# ActivateTime : 18
tRFC : 88
Command Rate (CR) : 2T

The rest is [Auto]

____________________________________

values after brackets are from Asus TurboV on Vista 64 bits

CPU Voltage ........................ [Auto] 1.22500
CPU PLL Voltage .................... [Auto] 1.8
QPI/Dram Core Voltage .............. [Auto] 1.375
IOH Voltage ........................ [Auto] 1.2
IOH PCIE Voltage ................... [Auto] 1.50
ICH Voltage ........................[Auto] 1.1
ICH PCIE Voltage ................... [Auto] 1.5
Dram Bus Voltage ................... [Auto] 1.66

[Seaquest]

Hello gabrielgpm,

I've had the same issue with my Perfect Storm F3-16000CL8T 6GBPS on a EVGA Classified. I've tried every setting I could think of. When I put the sticks in another memoryslot, the "bad" stick didn't show in that slot. I checked it with CPU-Z which slot was not showing my RAM. Seemed that every time the same stick didn't show. Running memtest was no problem once the stick is recognized. I did send them for RMA.

[DaKo]

Seaquest, it's exactly the same problem that I have with the F3-12800CL9T-6GBNQ memory. I found out that it's always the same DIMM which isn't recognized at a cold boot. However, it is recognized in Memtest86+ (although it can't be used):



RAM isn't recognized properly at a cold boot (it doesn't matter in which slot the DIMM is):



I've sent it back to the shop I bought it from today. I hope they will send me a new kit instead of sending it to G.Skill in Taiwan for RMA .

[gabrielgpm]

Hi DaKo,

On your second pic it seams that the slot of the second channel is not picking up the timings correctly. So if you changed the position of the DIMM to another channel this problem would be transferred to new channel as well?

P.S.: Regarding your memtest, I also thought that it recognized all the time the 3 DIMMs as well, but I managed to prove this wrong by running memtest from a bootable CD at the first boot of the day when the detection problem is more frequent and on this memtest only 2 DIMMs (4BG) were detected.

[DaKo]

On your second pic it seams that the slot of the second channel is not picking up the timings correctly. So if you changed the position of the DIMM to another channel this problem would be transferred to new channel as well?

Exactly. I had the "buggy" DIMM in slot 1 at first. I first thought that this could be a mainboard issue because the settings were always not remembered in slot 1. Then I moved the DIMM to slot 2 and the next day (the PC "cold booted") the settings from this slot couldn't be "read". So I found out that it has to be this DIMM which can't be recognized properly when beeing cold.

P.S.: Regarding your memtest, I also thought that it recognized all the time the 3 DIMMs as well, but I managed to prove this wrong by running memtest from a bootable CD at the first boot of the day when the detection problem is more frequent and on this memtest only 2 DIMMs (4BG) were detected.

This picture was taken when the computer booted a couple of times this morning in order to let it recognize the full 6 GB (I didn't boot into Windows and only 4 GB were detected). Memtest seems to see the DIMM (like the BIOS as seen on the first picture), but it can't use it (like Windows; when I boot into it only 4 GB can be used, but it detects 6 GB).

Last week (when 6 GB were detected on a cold boot right at the start) I ran Memtest and this was the result:



However, if I start Memtest when the PC has been running for some time no errors can be found. This "cold boot bug" is really strange. I hope that the new kit resolves this problem.

[Seaquest]

Dako,

I've had the same. With the only difference dat the ram never showed on a cold boot. I hope I'll get my new kit soon. It's been already 3 weeks since my RMA. The store I bought them has nothing in stock. So I'm glad I have my laptop I can use for now.

[gabrielgpm]

Exactly. I had the "buggy" DIMM in slot 1 at first. I first thought that this could be a mainboard issue because the settings were always not remembered in slot 1. Then I moved the DIMM to slot 2 and the next day (the PC "cold booted") the settings from this slot couldn't be "read". So I found out that it has to be this DIMM which can't be recognized properly when beeing cold.




This picture was taken when the computer booted a couple of times this morning in order to let it recognize the full 6 GB (I didn't boot into Windows and only 4 GB were detected). Memtest seems to see the DIMM (like the BIOS as seen on the first picture), but it can't use it (like Windows; when I boot into it only 4 GB can be used, but it detects 6 GB).

Last week (when 6 GB were detected on a cold boot right at the start) I ran Memtest and this was the result:



However, if I start Memtest when the PC has been running for some time no errors can be found. This "cold boot bug" is really strange. I hope that the new kit resolves this problem.

I?m not sure yet if the cause of our problem is the same. In my memtest only 2 slots were recognized when booted with 4GB. Nevertheless, in Vista CPU-Z did detect 6144 Mbytes Triple Channel DDR3 while in Vista Welcome Center only 4,00 GB RAM was shown. I am not sure if the detection problem is independent of slot though since my BIOS doesn't have independent timings for each channel as your BIOS has. However, I can test this with CPU-Z and today I?ll change RAM positions and see in tomorrow?s first boot if the problem is coming from a different slot. Anyway, this problem only started when I got my G.Skill modules - before I had memory from a different brand on the same motherboard and 6GB were always detected.

[Seaquest]

In my case CPU-Z did not detect 6 gb of ram. just 4 gb. Also not in E-Leet, or Bios.