Solved Problems in several areas....looking for more TS tips to find solution

[RangermanLV]

This all started about 2 months ago. After a driver update to my Nvidia video card. I started getting system hangs which started triggering Win7's TDR and reseting my video drivers. At first it was only in one game and I suspected it was a game update, but I recently got The Sims Medieval and it started happening in that game almost right away. After that I susspected it had something to do with the .NET 4 framework that The Sims installed as part of it's install but then realized that dident make any sense. I also found that my OS HDD had a huge ammount of reallocated sectors but they all came at one time. I was watching it for awhile and the count never seemed to get higher so I assumed it was some kind of one time event. Then I started thinking, ok what can cause errors in several different parts of a system seemingly at random? My RAM. Ok so I went online and looked for something that would check my RAM. Windows Memory diagnostic is fairly useless, all it told me is that there are hardware errors. Wow, gee, THANKS Windows for telling me that, too bad you couldent tell me A BIT MORE. In any case I started running a free program I found called Memtest86+. It seems to be a VERY exhaustive test. After running for 1 1/2 passes it had found 17 total errors spread across 3 different DIMM's from what I can tell. So here is where I need a little help. I know the traditional method of checking ram is to run one stick at a time, one slot at a time, to see if any individual stick is consistently having errors and/or there is a problem with my DIMM sockets. I am wondering however if anyone could recommend a more informative memory test program that could tell me which DIMM the error came from and/or if it's a specific address on the DIMM? I am a very old school tech and my brain is seeming to remember that there was some way to either tell a memory stick that a specific address is bad and have BIOS and/or Windows write it out of use, but I may be thinking of ECC memory and mine is not ECC memory. In any case I need a program that is a little more specific about where the errors are coming from other than just giving me the memory address of the error at the time of the test. Also as I did not let the test completely finish it's standard 2 passes perhaps it would have given me more info after the end of the test? Anyone familiar with Memtest86+ in this regard? Also on a partial side note, I have noticed in Speedfan it gives me very specific information about my system voltage levels which I like. However I seem to be getting some strange numbers and I am wondering if this is something to worry about or it's just a symptom of Speedfan perhaps not being 100% in tune with my Motherboard or does someone feel there really might be something wrong with my PSU and perhaps this is contributing to my memory issues. This is what I see on the bottom of Speedfan in regards to my system voltages. Vcore1: 1.23V -12V: -16.89V Vcore2: 3.30V -5V: -8.58V +3.3V : 0.00V +5V: 4.87V +5V: 4.84V Vbat: 3.14V +12V: 12.16V Now aside from the fact that i've worked for Dell tech support and was a store tech for Gateway for several years, I am not THAT much of a tech to know EXACTLY what my Vcore voltages should be other than to say i'm not overclocking anything at all in this system. And I know the +3.3V is pretty much still not used at all yet so I undertand that being 0.00V However the -12V and -5V values worry me. My mind seems to remember that memory uses 5v but I am unsure if it also uses -5v for timing or anything like that. and the -12V value being at -16.89V simply puzzles me. However as the +12V value seems to be in line I am not sure if the -12V value being so out of whack is affecting anything either. Any help in this area would be greatly appreciated also. My apologies for this post rambling on however being a tech and having had to deal with people asking for help before I understand the value of knowing what has already been done and getting ALL the symptoms of the problem to help determine some kind of solution. I am in dire need of trying to keep this computer running without having to need to replace parts that are going to cost me money as I am currently unemployed and can't afford any kind of hardware replacements. The memory and the PSU are fairly recent in the system as I got the PSU to support the GTS 450 video card I just got at Christmas for a present. So they may all possibly be still under warranty if they do need to be replaced. My next step until some kind soul can give me a bit of guidance here is to run the Memtest86+ program to it's full 2 pass completion overnight and see if it gives me some more detailed info after it's completed, but the fact that it seemed to be saying I had errors on 3 different memory sticks is what is also making me wonder if this is some kind of strange voltage issue also. Thank you for any help. PS also my apoligies for this looking like one giant run on sentance. I tried to edit the post and format it a little better but for some reason it's not showing the changes in the post.

 

[zigzag3143]

Regarding memtest. It is the gold standard and will give you the info you seek as to errors and addresses but does need to be run one stick and one slkot for the highest degree of accuracy.

It does take a long time. One other thing memtest tests the ram and mobo outside the windows environment so if it is software related it wont tell you anything

Run the test and if it reurnss any errors you can be fairly sure there is a problem.If the ram and mobo return clean

Run driver verifier from within windows to find out which driver is causing the grief

I'd suggest that you first backup your stuff and then make sure you've got access to another computer so you can contact us if problems arise. Then make a System Restore point (so you can restore the system using the Vista/Win7 Startup Repair feature).

In Windows 7 you can make a Startup Repair disk by going to Start....All Programs...Maintenance...Create a System Repair Disc - with Windows Vista you'll have to use your installation disk or the "Repair your computer" option at the top of the Safe Mode menu .

Then, here's the procedure:
- Go to Start and type in "verifier" (without the quotes) and press Enter
- Select "Create custom settings (for code developers)" and click "Next"
- Select "Select individual settings from a full list" and click "Next"
- Select everything EXCEPT FOR "Low Resource Simulation" and click "Next"
- Select "Select driver names from a list" and click "Next"
Then select all drivers NOT provided by Microsoft and click "Next"
- Select "Finish" on the next page.

Reboot the system and wait for it to crash to the Blue Screen. Continue to use your system normally, and if you know what causes the crash, do that repeatedly. The objective here is to get the system to crash because Driver Verifier is stressing the drivers out. If it doesn't crash for you, then let it run for at least 36 hours of continuous operation (an estimate on my part).

If you can't get into Windows because it crashes too soon, try it in Safe Mode.
If you can't get into Safe Mode, try using System Restore from your installation DVD to set the system back to the previous restore point that you created.

 

[RangermanLV]

Ok well here is something strange, at least to me since I havent had any TS experience in a work enviroment since before this came about. My MB supports dual channel memory arcitecture, I read up about it, sounds nice but dosent sound like all that big of a deal unless i'm running something that manipulates huge chunks of memory at a time. I already had 2 sticks of Kingston 1GB PC 800 matching pair when bought, recently when I upgraded to Win7 I bought a matched pair of 1GB Patriot PC 800 sticks and installed those also. I installed them per the color coding on the MB and it activated the dual channel mode on my MB. Just for kicks last night before I ran the full Memtest I took the sticks out and reseated them in unmatched pairs to force the system into single channel mode. This time the test came back with no errors at all on the memory after running the full test. So this now tells me what? The Dual channel mode on my MB is perhaps not totaly stable? Or perhaps simply removeing and reseating the memory has resolved the errors I was getting when running the test before? I am going to move the memory back into matching pairs and run the test again with the MB running in dual channel mode to see what happens. If again I have no errors then I suppose that solves whatever part of this problem was related to memory errors. In any case if I do get errors in dual channel I can apparently move the memory back to unmatched pairs, run the MB in single channel mode, and be safe that way, at least from an "outside the OS" enviroment mode. Regardless of the result I will run the program you suggested also to see if it comes up with anything also. I was wondering and since my last message was kind of a mess you might have missed it. Do you have any opinion on what look like strange voltage values I am getting from Speedfan on my -12v and -5v info? Thanks for the help and suggestions. Eric

 

[RangermanLV]

A small update....

I put the memory back in the configuration that let the motherboard enable dual channel mode and again ran Memtest86 and again got several errors. Aside from the fact that it is very difficult to seat memory in my motherboard, <I was always taught that if you don't seat it so the hold downs basically snap audibly into place it's probably not truly seated correctly> it appears to me that either for my MB the memory has to be 4 identically matched sticks or something is simply wrong with the way my motherboard handles Dual channel memory architecture as I continue to get no memory errors when the ram is in unmatched pairs in the MB which disabled the dual channel mode. I am going to run with the memory in this configuration for a while and see if I am still having problems with Windows deciding my video driver needs to be reset, and if so then I will run the Driver Verifier program that was suggested to see if the problem isn't solely related to my memory architecture and perhaps there is another problem. I will post again in a few days in this thread what the outcome is. Eric

 

[RangermanLV]

Regarding memtest. It is the gold standard and will give you the info you seek as to errors and addresses but does need to be run one stick and one slkot for the highest degree of accuracy.

It does take a long time. One other thing memtest tests the ram and mobo outside the windows environment so if it is software related it wont tell you anything

Run the test and if it reurnss any errors you can be fairly sure there is a problem.If the ram and mobo return clean

Run driver verifier from within windows to find out which driver is causing the grief

I'd suggest that you first backup your stuff and then make sure you've got access to another computer so you can contact us if problems arise. Then make a System Restore point (so you can restore the system using the Vista/Win7 Startup Repair feature).

In Windows 7 you can make a Startup Repair disk by going to Start....All Programs...Maintenance...Create a System Repair Disc - with Windows Vista you'll have to use your installation disk or the "Repair your computer" option at the top of the Safe Mode menu .

Then, here's the procedure:
- Go to Start and type in "verifier" (without the quotes) and press Enter
- Select "Create custom settings (for code developers)" and click "Next"
- Select "Select individual settings from a full list" and click "Next"
- Select everything EXCEPT FOR "Low Resource Simulation" and click "Next"
- Select "Select driver names from a list" and click "Next"
Then select all drivers NOT provided by Microsoft and click "Next"
- Select "Finish" on the next page.

Reboot the system and wait for it to crash to the Blue Screen. Continue to use your system normally, and if you know what causes the crash, do that repeatedly. The objective here is to get the system to crash because Driver Verifier is stressing the drivers out. If it doesn't crash for you, then let it run for at least 36 hours of continuous operation (an estimate on my part).

If you can't get into Windows because it crashes too soon, try it in Safe Mode.
If you can't get into Safe Mode, try using System Restore from your installation DVD to set the system back to the previous restore point that you created.

One question I have now about what's going to happen after I start running the system with Driver Verifier on? The problem I am having is NOT causing a blue screen crash, it's causing windows TDR. So When I run this what exactly are we looking for to happen? It it going to generate an event log somewhere if the TDR triggers in the programs I am running while Driver Verifier is running that will give us more info about what is causing TDR to be triggered?

 

[Dave76]

Driver Verifier stress the drivers and will indicate a driver that may be causing problems.

Any errors, even one, when running memtest86+ is too many.
These errors can be due to 4GB of RAM, 4 RAM cards installed or 2GB or larger RAM cards.
You will likely need a memory controller voltage bump, be careful too much voltage can cause problems.

If you can post some information, we can try to adjust some settings.


When you run memtest86+ you are looking for bad RAM cards, bad motherboard slots, so you need to follow the testing procedure in the tutorial.
Sometimes errors from these tests can be corrected by adjusting the settings.
RAM - Test with Memtest86+

Can you download CPUZ and post a snip of the CPU, Mainboard, Memory, and SPD tabs.

For posting the information , enter 'Snipping tool' in the Start button > Search box, hit enter. Select the area you want to post and save to a convenient place.
How to Post a Screenshot in Seven Forums


Have you been checking your temperatures?

UseReal Temp , run and check the CPU temps when at idle, under load and/or testing.
Let us know the minimum and maximum temps you get.
Go to your BIOS and tell us what the CPU, RAM and North Bridge (NB) voltages are.

 

[RangermanLV]

Ok here is some of the info you requested.

Ok first off as far as the RAM and errors. I could see with all RAM installed I was getting some kind of errors so I went ahead and tested everything. All 4 ram sticks passed with no errors in slot 1 on the MB. I checked one of each type of ram I have in all 4 slots on the MB and again this passed with no errors. However when I reinsert the ram in a Dual Channel configuration this is what I got.

Now as to the CPUZ info. I am posting screenshots of the tabs you requested. Actually there is a nice little function in CPUZ that if you press F5 it auto saves a very small .bmp file of whichever tab you are on at the time. Unfortunately it will not save multiple shots from the SPD tab if you switch slot #'s so I just had to get a little creative to get a SPD shot for both of the memory pairs I have in the system. Also someone needs to update the saved image in the link you sent me about "How to post a screenshot in Seven Forums" as the image shows a .bmp file and apparently those are not allowed now as they do not show up in the bottom of the attachment tool, so I just converted them all to .jpg's Also so that noone gets too worried. Yes the font size is different in the OCZ spd tab. I was trying to make things a little bigger and more readable but realized that while CPUZ is very nice to let you change those font sizes, there is no option to increase the programs box size itself.

As to the current voltages for CPU, RAM, NB. There is nothing in BIOS that directly tells me "this is the current CPU voltage, RAM voltage, NB voltage" I did have AI Overclock turned on in BIOS and had it set for 10% overclock and this seemed to be working fine since I got the system setup the way it is now aprox 5 months ago. Since this problem started I have moved the AI Overclock setting back to AUTO which I am assuming lets the MB and BIOS control the voltages based on what it knows about the installed components. There is another section in BIOS that gives me these values for voltage.

Vcore 1.21
3.3V: 3.29
5.0V: 4.86
12V : 12.16

But it does not correlate these directly to any specific component in the system in that area. Additionally Speedfan 4.43 gives me a Vcore2 voltage of 3.30V if that is helpful at all.

As far as temps goes. Idle temp on both cores is 47c. While running The Sims Medieval and in super speed mode, which I understand stresses the CPU fairly well, the highest temp recorded was 60c on one core and 63c on another.

For now I have taken the RAM sticks out of matched pairs to force the MB NOT to try to use Dual Channel memory mode. I am going to run Memtest86 again overnight and see if I am still getting errors with the memory not in a Dual Channel configuration. I am a bit confused about what the memory is doing though. Looking at the CPUZ info for the individual sticks the only value I see different in the JEDEC level they should be running at is the tRAS value of 15 for the OCZ memory and 18 for the Kingston. Is this enough of a difference to be causing me all these memory errors when the sticks are installed together? Could it also be a possibility that since the Memtest86 program does not understand and cannot bring up any kind of information about what my memory or motherboard is while it's running that it cant perform a truly accurate test with all the memory running together and/or in Dual Channel mode?

In any case than you all for the continuing help in this. As I said I am a support tech from way back but I never had to get this deep into a system and exactly how things are running to resolve issues. I still need to run Memtest in the current memory configuration and also run Driver Verifier probably while I am in the game that is still giving me the bad problems with video driver resets while i'm playing in certain areas. I did see something new yesterday, but it might be a symptom of me playing around with my memory so much, Sims Medieval crashed to desktop while I was playing and then when I went to perform a restart I got a "Bad Pool Header" BSOD. As the file size limit for .ZIP or .RAR files is 8MB according to the attachment tool and the file is 100MB even zipped i'm not sure how I would be able to send it to you. Also I noticed that 2 days ago while I was playing DDO which is the game I am having the most problems with, In action center if I look in the Reliability Monitor it shows an event for the 20+ video driver restart errors I had over a 2 hour period in the game. All of them have a BCCode of 117 and they do have 2 .xml files and a .dmp file associated with each event so I will upload a copy of those also in case that provides some extra help here.

 

[Dave76]

Good info in your reply, will answer your questions along with information and some of my opinions.

First, Important thing is the voltages for CPU, RAM, NB.
CPU voltage which is Vcore, your BIOS reading of 1.21v and the CPUZ CPU tab reading of 1.248v are in the right range and will change if you have EIST or CPU stepping enabled.
If SpeedFan was correct with that Vcore2 voltage of 3.30V, there would be smoke coming from the case with a simultaneous blank screen. It must be the 3.3v rail feed.

For the DRAM and NB voltage go to BIOS > Advanced tab > JumperFree configuration > Voltage Control.
From here 'Vcore Voltage' is the CPU voltage, 'Memory Voltage' is the RAM voltage, 'NB Core Voltage' is the memory controller voltage.
The Vcore is fine for now.
Memory Voltage, set manually to 1.80v, if this option isn't available leave it on 'Auto'.
NB Core Voltage, stock setting should be 1.20v, set it to 1.30v if that option is available, if not available set it to the next higher setting, which may be 1.393v.

The North Bridge Voltage, NB Core Voltage, is for the memory controller.
When you have 4GB RAM or 4 RAM cards you will usually need to bump the NB voltage to get the system stable.
This should stabilize your RAM.

It is highly recommended to use the same RAM in all four slots, from the same manufacturer, same batch, purchased at the same time and place.
If it is not the same RAM, as you are experiencing, it may or may not run together.
Sometimes, it can be coaxed into working.

The CPUZ memory tab will list what the timings are currently set at, the voltages are not the actual voltages but the recommended voltages.
The CPUZ SPD tab is just a list of the manufacturers recommended settings and voltages, btw not guaranteed to work on your system.

The CPUZ memory tab is showing that the RAM is set to the 'looser' of the two timings, the OCZ set with 5-5-5-15 timings are more apt to work at 5-5-5-18 than the Kingston set with 5-5-5-18 possibility of working at tighter timings of 5-5-5-15.
So, that should be good.
If the NB voltage bump doesn't fix the issues then we can manually set the timings at 5-5-5-18 or looser, sometimes 'Auto' can cause problems. But, that is for later.

Let's see if the NB Voltage bump will let all four RAM cards run in Dual channel mode, run memtest86+ again, if you get errors stop the test.
Then run memtest86+ again with all four RAM cards in Single channel mode, if you get erors stop the test.

Let me know the results.

 

[RangermanLV]

Good Deal...

Yea when I used to work on computers for the public, the idea of being able to change any of these settings in BIOS was like "yer kidding right? Why would anyone want to change that??" Course then also no matter what you wanted to have however many ram slots you had installed with identical ram sticks and having a Gig of ram was like WOW!!!

Ok i'll try changing that stuff in BIOS as you said. I know if I take manual control of the voltage settings it will let me change things in any or all of those voltages so that shouldent be a problem. I already moved the ram back to the unpaired config with the Kingston pair in A1 and A2 and the OCZ in B1 and B2. This on my motherboard is the unpaired config. Having paired sets in A1/B1 and A2/B2 triggers the Dual Channel mode. And yea i'm really wishing at the time i'd gotten the ram I had just gotten 4 1gig sticks all the same but that was when I was still running XP a year or so ago and when I upgraded to Win7 is when I learned/was advised/heard thru the grapevine that 4Gig is kind of a minimum for Win7 to run decently.

Oddly enough with the ram as it is now I ran Memtest last night overnight and it made it thru 5 passes before I woke up to check on it and start using the computer for the day. I posted a screenshot below of the errors I am getting. Apparently I have ghost ram somewhere past the 4Gigs I have installed LOL In any case you can look at that and tell me what you think is causing it. Seems pretty weird to me.

I will report back after I change the voltages and run memtest again.

Eric

 

[Dave76]

Where did you see the ghost RAM?

The NB voltage bump should help, if not you will have to test each RAM card separately, which will take a while with four cards.

Let us know the results of the test with the NB voltage bump.

 

[RangermanLV]

Well we're haveing fun now....

Ok well the lowest manual voltage option for memory was 1.92, so I am leaving that on AUTO. The closest NB voltage I saw was the one you mentioned might be the only one avalible at the 1.393v so I set it for that. I ran Memtest with the memory in dual channel mode and immidiately got many more errors than before. So leaving the NB at the 1.393V I put the memory back in single channel config and booted to windows. Something about changing the NB voltage windows didn't like as it wouldn't boot properly. I ran the boot repair tool and it did a few things and it seems to be working now in windows HOWEVER, I notice on CPUZ on the memory tab it's saying my tRAS is running at 23 clocks. LOL I am going to go in and manualy set that for 18 and leave the memory in single channel for now and see how that runs tonight while I am playing on the computer. After I am done I will make sure that's still manualy set for 18 in BIOS and put the memory back in Dual Channel slotting and run Memtest again and see what happens, if I get errors there again I will move it back to Single channel and make sure the tRAS is still manualy set for 18 and run Memtest again and see what happens.

I think what you were saying about leaving things in AUTO can cause problems especialy when you start making manual changes to other things. Apparently bumping the NB voltage seemed to confuse the BIOS and make it want to set the tRAS to some whacky setting. At least I HOPE that's what it did.

In any case once I go back and manualy set the tRAS to 18 in BIOS I will post a little addendum here with how the system seems to be behaving with the NB voltage bumped up and the memory in single channel with the tRAS specificly set to 18.

 

[Dave76]

The tRAS changing is strange, looks like you are running the latest BIOS version.

This might be due to the two different RAM card brands and the 'Auto' RAM timings setting.

What is the boot repair tool?
From Asus?
I would be careful with a tool like this, no motherboard tools are very good at this kind of thing.
This could be the source of your problems.

Go to your BIOS and check if everything is on 'Auto', make a list of what is not.
Other option is to re-set BIOS to defaults, then check the RAM timings and voltages, including the NB voltage.


We may need to loosen the RAM timings, let's try testing with the current settings first.

 

[RangermanLV]

Well I found a config with no errors....

However it's in single channel mode. <shrug> From what i've read there isn't a HUGE difference between single channel and dual channel but I'm not sure how much it means from a gaming perspective. In any case this is what happened during testing:

In a dual channel config, tried various different locations for the ram swapping the pairs and swapping the manufacturers between slots, With the higher NB voltage I always end up getting numerous errors before even one pass has completed. Tried changing settings in BIOS to conform to what the SPD tabs say in CPUZ. <I.E. Manualy set memory timings to 5-5-5-18 and 23 in the BIOS. <There are many more timing settings in there but I left them all on AUTO> With the known settings applied in BIOS and the NB at the new voltage in dual channel configurations the memory either gives alot of errors before it even finishes one pass, or it locks up in the Memtest program on test #5 I believe. When I go back to the first Single channel config I was trying out before <the Kingston in A1 and B1 and the OCZ in A2 and B2> I am getting no errors at all thru one pass. I will run this overnight and hopefully it will get thru several passes without errors. So unless you can think of something else I can do in BIOS settings to make the 2 pairs happy with each other I guess i'm stuck in Single Channel mode.

Yes I am running latest BIOS for this MB. From back at the end of 2009 but it is the last update they made. <sigh>

Sorry when I said "boot repair tool" I guess I should have said WINDOWS Boot repair tool.

I will edit this message with the info about what is and is not Auto later before I start the overnight test on the memory in the current config. Right now I have the NB set to the 1.393V and the memory timings manually set to

CAS Latency 5
RAS to CAS delay 5
tRP 5
tRAS 18
tRC 23

As I said there are many more manual timings that can be changed in there, I will list them later.

 

[Dave76]

Manually setting the first four RAM timings is a good step as this will help to stabilize the RAM.
Just leave the rest of the RAM timing settings on 'Auto'.

Can you take a CPUZ snip of the Memory tab with all four RAM cards installed in the 'Single Channel' configuration.
This should actually be in Dual channel mode, with four RAM cards installed.
From your motherboard manual.
View attachment 150124 View attachment 150122

Your 'Single channel' mode installation is with the <the Kingston in A1 and B1 and the OCZ in A2 and B2> both Kingston and both OCZ cards installed in the same color slots, correct?

If this is right, it is the correct installation for your RAM, and should be in Dual channel mode.

Let's confirm if in your 'single channel' mode they are actually running in single channel mode.
With four RAM cards installed it should be in 'Dual channel' mode.

Not quite sure what 'WINDOWS Boot repair tool' is.
Is it Start up Repair?

 

[RangermanLV]

More on the MB from hell.....

Manually setting the first four RAM timings is a good step as this will help to stabilize the RAM.
Just leave the rest of the RAM timing settings on 'Auto'.

Can you take a CPUZ snip of the Memory tab with all four RAM cards installed in the 'Single Channel' configuration.
This should actually be in Dual channel mode, with four RAM cards installed.
From your motherboard manual.
View attachment 150124 View attachment 150122

Your 'Single channel' mode installation is with the <the Kingston in A1 and B1 and the OCZ in A2 and B2> both Kingston and both OCZ cards installed in the same color slots, correct?

If this is right, it is the correct installation for your RAM, and should be in Dual channel mode.

Let's confirm if in your 'single channel' mode they are actually running in single channel mode.
With four RAM cards installed it should be in 'Dual channel' mode.

Not quite sure what 'WINDOWS Boot repair tool' is.
Is it Start up Repair?

My apologies for not getting back to you recently, had some school exams and a job interview in the past few days so it's been a bit busy around here.


Yes Start up Repair is what I was thinking of. I don't need it very often so don't remember the actual name. As far as the ram and the single/dual channel mode. BIOS tells me right at the beginning of post if it's going into single or dual channel mode, based on how I have the ram inserted into the slots. Right now I have the Kingston sticks in A1 and A2, and the OCZ sticks in B1 and B2. This posts and tells me it is running in Single Channel mode. This is the longest run screenshot so far with my memory in that configuration.

View attachment 151050

Until I stated looking for problems I had the memory running in dual channel with the kingston sticks paired in A1,B1 and the OCZ sticks paired in A2,B2. I believe it must be a mistake in the manual or a mistranslation from Chinese where it is saying the Channel pairs are A1/A2 B1/B2. You can see by the socket coloring and how it's paired that the #'s and not the letters are the Dual Channel pairs. And in actual usage this is what is correct. CPUZ shows single channel mode currently also. Although strangely enough, or perhaps not so strange, CPUZ's Slot #'s in the SPD tab line up as follows A1=Slot 1 B1=Slot 2 A2=Slot 3 B2= Slot 4. At least as far as CPUZ sees things.

View attachment 151053

So far with the memory in the current single channel config this is the LEAST errors I am getting in Memtest86 over the longest period, and as you can see from the screenshot, or actual not since I can't scroll back over all the errors, they are all on the same 2 tests in the same spots. From what I can tell setting the NB voltage did help some with the stabilization in single channel mode. But somehow made the problem worse in dual channel mode, as running the ram in a few dual channel configurations either results in MANY more errors, ie over 70 errors before the first pass is even completed, or the memory test locks up on one of the tests in the first pass.

As I said before we even changed the NB voltage, when I started seeing errors in Memtest86 I did run all the ram sticks in single installations and came back with no errors at all thru several passes, both with the ram in A1 and with one of each type of the ram in all other slots singly. At this point I am at a loss as to what to do other than kick myself for not being able to get 4x1GB when I decided I wanted to run at 4GB when I upgraded to Win7. :sarc: Because either this is an error with Memtest86 being able to read/write to my ram in Dual Channel mode because it's internal data has no info for my ram or MB apparently, or my MB is simply far more picky about what it considers good pairs as the comment in the manual about making sure the memory size is = and CAS latency is = is apparently BS.

I am curious if there is a good reliable memory test program that can be run from inside Windows since Windows itself has a better understanding of how my memory is configured and the identification of the MB and drivers for everything that apparently Memtest86 does not. I realize this is probably a bit of wishful thinking on my part since obviously the test would need to be running with Windows AND the test using part of the memory it was testing but it's a nice thought. I should note that the Windows memory test, for all it's uselessness about giving any direct information, does say there is a problem with my memory, but unless you know how to make that test run in a more detailed mode I have no idea what it thinks IS wrong.

 

[Dave76]

Understand your frustration, but this issue is well known, RAM that is not matched is very difficult to get to run stable.
You can pick up two identical packages of, let's say 2x1GB Kingston RAM, and they won't run together, could be from different manufacturing batches or a small change in components or internal settings.

Sometimes it just won't run.

The above tests show that the RAM cards are good, your motherboard slots are good.
The two sets of RAM won't run together at these settings.

RAM runs from BIOS settings not from Windows, I'm afraid Memtest86+ is the best memory testing program available.
It's more likely that Windows will cause problems with your RAM, which is why memtest86+ runs without Windows.

Your best option is to get new, matched RAM.

Next best option is to start adjusting the timings and voltages and try to find a setting that will run stable. This will take a lot of time and memtest86+ runs.

If you want to try this, start with adjusting the RAM timings, leave it at 400 MHz for now, change the first four timings to 6-6-6-20 and leave the rest on auto.
Run memtest86+, if you get any errors stop the test and change the settings to 6-6-6-21, run memtest86+ again. Any errors stop the test and let us know when the errors came up and approx how many. More errors earlier than the first test or less.

This is the general procedure, sometimes loosening the timings will help.

Let us know the results.

 

[RangermanLV]

Next best option is to start adjusting the timings and voltages and try to find a setting that will run stable. This will take a lot of time and memtest86+ runs.

If you want to try this, start with adjusting the RAM timings, leave it at 400 MHz for now, change the first four timings to 6-6-6-20 and leave the rest on auto.
Run memtest86+, if you get any errors stop the test and change the settings to 6-6-6-21, run memtest86+ again. Any errors stop the test and let us know when the errors came up and approx how many. More errors earlier than the first test or less.

This is the general procedure, sometimes loosening the timings will help.

Let us know the results.

Ok what i'll do is change to the 6-6-6-21 and run the tests in dual channel and single channel and see if I can get some kind of stability or at least error reduction.
As far as the real original problem with my video driver resetting over and over in certain games and certain situations, I know the ram errors I'm having definitely arnt helping the problem, but I'm wondering at this point if the video driver problem itself is more likely some kind of internal Windows error as far as a compatibility issue with some recent addition or change Windows did and the type or types of video modes these particular games want to use? It's such a confusing problem because it seems to be almost totally random in that when it does occur, sometimes it will happen 5 times within 10 mins of play, and sometimes it wont happen for over an hour or more after I've started the game, then happen several times, then stop again for a while.

I have noticed that as far as the AD&D online game it does SEEM to be triggered, by some particular areas of the game that use perhaps a more intense or additional feature of my graphics card. But there have been a few times when i've gone thru these areas and not had one driver reset so I really don't know what to make of that.

In any case thanks for the continued help, hopefully at some point i'll be able to get some stability out of these two pairs together, or i'll get a job and just buy a matching pair of Kingston 2g sticks.

 

[Dave76]

Let us know how the RAM timing tests are going.

Hope this will work for you.

 

[RangermanLV]

And so it goes.....

Well I've tried several configureation but there are still many more to try. I should be comeing into some money soon, enough to buy 2x2GB paired Kingston memory sticks that are the fastest speed my system can handle so I may just give up trying to get the Kingston and the OCZ to run together in either one of the modes. I have the two Kingston 1GB sticks running in dual channel mode and I ran it thru about 20 passes with Memtest86+ the other night with no errors so I may just leave it like that for now till I get the new memory.

On a somewhat more productive note I BELIEVE I've found the main problem as to what is/was causing my graphics driver to trigger TDR's all the time. I have wondered and was told by someone I had reason to belive in the past that 70C is a nominal operateing temp for newer video cards while running fairly recent game software. I.E. Witcher 2, COD, WoW with high video settings...etc...etc.. However I also always wondered why I never heard my GPU cooling fan running any faster, I just assumed it was very quiet even at 100% speed.

However I recently downloaded the Nvidia System Tools which give you quite a bit more control over what your video card does and how things operate. So just for kicks I set the GPU fan speed to Manual and moved it to 100%. Ok well NOW I can hear it. It's not really loud but it is noticeable so I am wondering how come even if my card temp is getting to 70C I've never heard it run faster before?

So what I've done is manual set the GPU fan speed for 80%. Fairly fast but not really audible at all. Now running the games I've been having TDR problems with before I suddenly am not having these problems anymore. I'm going to have to do some more testing with this to see if all the games I was having problems with before that setting the fan speed manually myself has now resolved, but so far this seems to be working well.

So now my new question to anyone who has a clue, why was my GPU fan speed never running faster when the card gets hot on it's own? I am assuming this is something written into the firmware of the video card that correlates GPU temp to increasing fan speed but apparently it's never been working on it's own. Anyone know how to check if some other program is trying to control the GPU fan speed and if it's working or not. Also does anyone have an opinion on just leaving the fan speed manually set at 80% constantly? With it set at 80% I can run Witcher 2 with some fairly decent graphic settings and the card never goes over 52C. As I said I still need to test this with the other programs I was having problems with but so far it's working great.

Thank you all for your opinions and information in this matter.

 

[Dave76]

That is the best plan for your RAM, getting different brands/spec RAM to run together is not easy.

For your graphics card issue, might be FW, have you checked for updates?
Have you updated the drivers?
Sometimes you may need to go back to a previous driver if the new one isn't working for you.
Does the Nvidia System Tools have any options for 'Auto' or any more detailed settings.
Some utilities have variable settings that can be adjusted.
Running the fan manually at 80% is better than the alternative.

You might want to start a new thread in the Graphic Cards forum.
There are more graphics card experts in that forum.