All:I have recently built a new i7 975 system (Asus Rampage 2 Extreme, 6GB Corsair Dominator 1866, Nvidia 285GTX). Currently, both the BIOS and OS (XP64) only recognize 4GB of RAM.When I first built the system in late June, the BIOS would recognize all 6GB of the RAM about 25% of the time, all other boots would only show 4GB. In CPU-Z, however, it would always see all 6GB of RAM. I then tested all sticks individually with Memtest and they all tested good. I also tested a single stick in each slot...I found it would not POST with only one stick in A1 (B1 and C1 were fine). Asus tech support told me I had a bad slot and I RMA'd the motherboard to NewEgg.When I installed the new MOBO, all 6GB showed up fine. I checked both the BIOS and OS properties for at least the first 100 or so boots with the new MOBO, and every time they both saw all 6GB. I assumed it was indeed a bad slot...A few days ago, I decided to check back on the RAM. I was amazed to see the system was showing only 4GB again...I had made zero changes to the installation or the BIOS, nor have I overclocked yet since installing the new MOBO. For the life of me I cannot figure out what could have changed. In fact now, it is 4GB every time, and of course CPU-Z still sees all 6GB.From what I gather on both the ASUS and Corsair forums, this problem exists for a lot of people...but I have yet to see a soution. I am leaning towards a bad memory controller on the CPU, but the Intel tech support guy tells me that either it would be bad or not...it wouldn't have worked sometimes and not others.Have any of you other system builders seen this problem? And if you have, what was the fix? I am officially confounded, and I don't want to send the CPU back until I can better determine that is actually the problem...Thanks,Tim

All:I have recently built a new i7 975 system (Asus Rampage 2 Extreme, 6GB Corsair Dominator 1866, Nvidia 285GTX). Currently, both the BIOS and OS (XP64) only recognize 4GB of RAM.When I first built the system in late June, the BIOS would recognize all 6GB of the RAM about 25% of the time, all other boots would only show 4GB. In CPU-Z, however, it would always see all 6GB of RAM. I then tested all sticks individually with Memtest and they all tested good. I also tested a single stick in each slot...I found it would not POST with only one stick in A1 (B1 and C1 were fine). Asus tech support told me I had a bad slot and I RMA'd the motherboard to NewEgg.When I installed the new MOBO, all 6GB showed up fine. I checked both the BIOS and OS properties for at least the first 100 or so boots with the new MOBO, and every time they both saw all 6GB. I assumed it was indeed a bad slot...A few days ago, I decided to check back on the RAM. I was amazed to see the system was showing only 4GB again...I had made zero changes to the installation or the BIOS, nor have I overclocked yet since installing the new MOBO. For the life of me I cannot figure out what could have changed. In fact now, it is 4GB every time, and of course CPU-Z still sees all 6GB.From what I gather on both the ASUS and Corsair forums, this problem exists for a lot of people...but I have yet to see a soution. I am leaning towards a bad memory controller on the CPU, but the Intel tech support guy tells me that either it would be bad or not...it wouldn't have worked sometimes and not others.Have any of you other system builders seen this problem? And if you have, what was the fix? I am officially confounded, and I don't want to send the CPU back until I can better determine that is actually the problem...Thanks,Tim

I saw something similar when I was trying to put 12GB (6 x 2GB DIMMs) into my i7 build. Some utilities reported 12, others 6 GB. I found that slowing the memory clock down and/or relaxing the timings worked (in my case all the way down to 1333 MHz for my 1600 MHz rated RAM).You might try bumping up the RAM voltage by 0.05v to see if that makes a difference (while carefully observing the 0.5v max spread between QPI voltage and Vdimm). If it does help, then your RAM is probably running too close to the ragged edge at your current speed/timings. In that case you can either take a chance at running the RAM slightly overvolted, or back down the speed/timings to where it'll work. If you have the capability, checking the actual Vdimm with a meter or oscilloscope under load would be a good idea as well--three DIMMs at high speed may be causing enough voltage droop that it puts the chip on/past its threshold of stability. The voltage reported in the BIOS and/or maintained by the regulator are not always accurate.If the voltage change doesn't produce any improvement, you can still try backing down the memory clock and/or relaxing the timings one CAS level. If that fixes it, you can either run with the RAM derated slower, or possibly RMA the memory if that's an option.Also, it's important to keep in mind that nobody actually produces 1866MHz DRAM chips...an 1866 MHz DIMM module uses 1066 or 1333 MHz chips that have been binned and (one hopes) tested at the higher speeds. It's not unheard of for a RAM chip's performance to degrade slightly with use...they may indeed degrade below the DIMM's rated speed/timings, but probably not below the actual design speed/timings of the chips used on the DIMM. In that case, you have an issue with the DIMM manufacturer.Last, I have read that it's possible to degrade the memory controller on an i7 chip either by pushing it too hard/fast (there is no temp monitoring in that area of the CPU package), or by exceeding the QPI-Vdimm spread. I'm not sure how that degradation would manifest itself, but it's a possibility.RegardsBob ScottColonel, USAF (ret)ATP IMEL Gulfstream II-III-IV-VColorado Springs, CO

Also reported on the eVGA x58 forum by a number of people.Usually turns out to be a bad stick of ram, bad slot or one that is not seated well.I had the issue and a reseat solved it here.I dont remember hearing of this issue near as much with core duo builds...

Bob,I'm pretty sure it's not the clock or the timings...this occurred at the SPD of 1333, 9-9-9-24, DRAM V1.65 and QPI V1.21 (which also were the settings where it was working fine for that stretch). The problem still existed at Corsair's recommended 1866, 7-8-7-20, DRAM V1.65. I tried bumping up the QPI to 1.30 as well as reducing the DRAM to 1.50 (Intel tech support's recommendation) to no avail.Jackcnd,I have reseated the memory a few times, still no luck. I think I shall run Memtest on all of the sticks again just to be sure one didn't go bad on me. Perhaps there was a stick that was intermittently bad and it finally gave up.

Hi. I have a similar problem but it hasn't become that annoying yet. Let me explain:I have purcased the following components in order to build a core i7 system especially for OC:CPU Core i7 920 D0 stepMobo ASUS P6T Deluxe V2Mem Corsair DDR3 1600MHz 3x2GB tripple channel with timings 7-7-7-20When I first installed the parts the system wouldn't boot at all. There was power but no POST messages on the screen.I removed 1 DIMM and it booted with 2x2GB=4GB. I thought I had a bad slot/DIMM but first I tried to flash the BIOS with the latest version and then give it a 2nd try and fortunately it recognized all 6GB but at 1066MHz as I hadn't started experimenting with OC the 920 at that time so it was running at its stock freq. 2.67GHz. Recently I started experimenting with OC and I managed to get the 920 up to 4GHz with a 21 multiplier and 190BCLK: this pushed the memory close to its rated limit but not at 1600MHz: it went around 1533MHz and I also lowered the timings from 9-9-9-24 to 7-7-7-20, the rated max. values. Still no problem with the amount of avail. memory. 6GB both in POST and OS (64bit obviously). However when I tried 20 multiplier 200BCLK and 7-7-7-20 timings, only 4GB were available. So is this a memory problem, a dimm slot problem or CPU mem controler? This is exactly as Bob Scott explained: When I back down the speed (1600--->1533) all 6GB are recognizable. Personally I don't mind running my mem at 1533 because the 920 D0 stepping has a max multiplier of 21 so I can reach 4GHz with lower BCLK (190 instead of 200) :(. As long as I don't see any change in POST/OS in total amount of memory while running at 21x190BCLK, I will be happy. :(

After running MEMTEST again, I found one stick that would not POST when it was installed individually. After digging back into the Corsair forums, I found the announcement about the Elpida problem and my particular RAM was one of the models that was pulled. I am awaiting my RMA and hopefully the problem will be solved after replacing the RAM.As far as your 1600 RAM, if you have TW3X4G1600C6GTF, that is also one of the models affected by the Elpida issue. If not, I would say that your BIOS update cured the initial prob and that the current prob is just a result of clocking beyond its rated speed, as you stated.

Hi. I have a similar problem but it hasn't become that annoying yet. Let me explain:I have purcased the following components in order to build a core i7 system especially for OC:CPU Core i7 920 D0 stepMobo ASUS P6T Deluxe V2Mem Corsair DDR3 1600MHz 3x2GB tripple channel with timings 7-7-7-20When I first installed the parts the system wouldn't boot at all. There was power but no POST messages on the screen.I removed 1 DIMM and it booted with 2x2GB=4GB. I thought I had a bad slot/DIMM but first I tried to flash the BIOS with the latest version and then give it a 2nd try and fortunately it recognized all 6GB but at 1066MHz as I hadn't started experimenting with OC the 920 at that time so it was running at its stock freq. 2.67GHz. Recently I started experimenting with OC and I managed to get the 920 up to 4GHz with a 21 multiplier and 190BCLK: this pushed the memory close to its rated limit but not at 1600MHz: it went around 1533MHz and I also lowered the timings from 9-9-9-24 to 7-7-7-20, the rated max. values. Still no problem with the amount of avail. memory. 6GB both in POST and OS (64bit obviously). However when I tried 20 multiplier 200BCLK and 7-7-7-20 timings, only 4GB were available. So is this a memory problem, a dimm slot problem or CPU mem controler? This is exactly as Bob Scott explained: When I back down the speed (1600--->1533) all 6GB are recognizable. Personally I don't mind running my mem at 1533 because the 920 D0 stepping has a max multiplier of 21 so I can reach 4GHz with lower BCLK (190 instead of 200) :( . As long as I don't see any change in POST/OS in total amount of memory while running at 21x190BCLK, I will be happy. :(

I have almost exactly the *same* build. Same motherboard, same processor, but Mushkin memory. I'm having the same memory problems. I am having both the boot problems and the only 4 GB showing up problem. Both CPU-Z and SIW see all 6 GB but neither the BIOS nor the operating system do. For some odd reason putting a flash drive in the machine allows it to boot. I found out by accident when I went to flash the BIOS. However, when I take the flash drive out, the machine will not boot.Jim

Jim,From what I understand, Mushkin used some of the same Elpida hyper modules that are failing at a very high rate. Try booting your computer without the flash drive using only one stick of RAM...do this with all 3 sticks in A1 slot, then B1 slot and finally C1 slot. You will probably find that two of the sticks will let you boot and one will not...if so, you have a bad stick and it is probably one of the suspect Elpida models. I hear that Mushkin has a great return policy where they will ship you replacements immediately without having to wait for the RMA turnaround of your bad RAM (unlike Corsair...I am still waiting for my replacements).