New RAM, more stability problems-help!

C

CalvinGT

Limp Gawd
Joined
May 7, 2007
Messages
251
This is a follow up to this thread:
Right here.

So, today my new Crucial Ballistix PC2 8000 DDR2 1000 2x1 GB RAM arrived to replace my bad Mushkin sticks. I was excited. First I had an issue getting the chips grounded-no big deal, finally did it. Thats where the good experiences ended.

So, I tried to get the motherboard to detect the RAM by default. No go-it put it in auto mode at 800mhz and at 5-8-8-23 or something absurd. I set the stuff manually. 2.1v 1000mhz, 5-5-5-15 timings (the rated timings). Set my voltage for the cpu high-like 1.4 and the speed at 3.3 (C2D E6600). I go into Memtest, and I get a freeze. Not errors, a freeze. Reboot, try again. This time, still no errors but memtest goes crazy. I raise the CPU voltage and the memory voltage. This time it crashes after 10 minutes-just freezes and hard locks. Raised the mem to 2.3 and the CPU to 1.45. About 10-18 minutes in it just hard freezes, every time. No errors, just freezes.

Finally, I go out to dinner and let it run stock voltages, stock CPU speed, RAM at rated timings. 3 hours plus and the damn thing is still running and ZERO errors.

I try to OC up to 3.2 ghz again, get some nice voltage for insurance, 6:30 into it, hard freeze again!!! I have tried everything and I cannot figure this out! I have yet to try this in Orthos at that speed-why? Because for some reason Windows hard froze, yes hard froze, twice trying to get into windows at 3.3 at 1.45v and 2.3v RAM at the freaking default rated speeds!! I have never, not once in 3 weeks with my new CPU had it crash loading windows at that low speed. Hell, the sucker was stable at about 1.375 at 3.3ghz right out of the box. I have NO freaking clue whats going on, and right now I cannot tell you all how frustrated I am.

Please offer any help and advice you can, and I am willing to follow detailed instructions to get you more information to help. Thank you in advance!!!

Core 2 Duo E6600
Tuniq 120
EVGA 680iLT A1
2x1GB PC2 8000 Crucial Ballistix
EVGA GeForce 8800GTS 320 SC
320GB Seagate 7200.10 16mb cache SATA HDD
 
Brand new Mushkin 550W
http://www.newegg.com/Product/Product.aspx?Item=N82E16817812002

Not the most expensive PSU in the world, but I read good reviews of it and I was told by people that it would be more than enough for my rig and Mushkin has always been reliable-plus, these issues are new and the computer and PSU are only 3 weeks old and I haven't changed anything but the RAM following the problems in this thread and the other linked.

Thanks for reading, any ideas?
 
To ghost6303:
CalvinGT said:
My PSU is the newest model Mushkin 550 True Wattage. I will check the CPU Batch in CPU-Z after work and post. I also have not tried memtest because I didn't think this could be a memory error-could forcing the RAM to run in 4-5-4-11 at OC speeds do this? I figured it wouldn't since these Mushkin sticks are very highly regarded.

Thanks guys, I will run those tests and get that info after work!
Click to expand...




To op Well it could be the motherboard or you could have bad luck with ram again. Which bios are you running?
 
LostInRehab said:
To ghost6303:





To op Well it could be the motherboard or you could have bad luck with ram again. Which bios are you running?
Click to expand...

I am not sure the number and cant restart to check for a bit because I have orthos running, but its the very latest revision from EVGA. Probably from late April? It seems to be running stable in both memtest and orthos at stock CPU speeds and stock RAM speeds however, which makes the OCing mystery that much more frustrating (I say seems cause I tested RAM for about 3 hours at stock but orthos only for 30 mins).

Thanks again!!
 
asssuming I got the right bios info for your board there is a new one out as of 5/7
Bios Version : P28
BIOS Date : 5/7/2007
Bios File : NF68_P28.exe (filesize N/A)
Notes :
The following was updated in release P28:
Improve S3 resume functionality
USB flash key functionality improvements
Overclocking and memory improvements
WHQL-related HPET fixes

Also includes these updates from bios P20, P21, P23, P24, P25, P26, and P27:
Quad-Core OC Improvements
Additional support for 1333MHz CPU's
Wireless PCI card fixes
Vreg fan header default set to on
Memory performance improvements
Vista WHQL Certified
USB Floppy improvements for RAID installs
Correct CPU temp and voltage system monitor displays in BIOS
Correct default CPU multiplier setting
Fix "Code 50" hang received under certain VGA configurations
Fix for SATA disk drives
Improvements to memory stability and overclockability
Disabled spread spectrum tables for improved overclockability
Improve POST screen CPU speed reporting
Improves X-Fi and internal audio functionality
WHQL Certified BIOS
Improvements to overclocking
Enhancements for Quad-Core CPUs
Adds ability to enable splash screen

link http://www.evga.com/Support/Drivers/default.asp?switch=2
 
I have not tried both sticks separately yet, sorry, had to do boyfriend duty and hang out with the woman! I tested the ram at stock and the CPU at stock while hanging with her. I got through 1.5 hours of orthos with zero errors or crashes and almost 3 hours of memtest with no crash/freeze at all.

However, just 5 minutes ago I attempted to boot up at a modest 3.2 with 1.40v and the RAM at the defaults specs. Twice it froze within a minute.

I will try both sticks one a time in the next hour or so-but if one of the sticks was faulty why would it only show up when the CPU alone was being overclocked? And wouldn't it display an error in memtest? I have never had memtest do anything but give me errors-this is the first time I have ever seen it simply freeze (but it only does it when the CPU is overclocked).

I will test out each RAM stick and post back. Thanks for any help you can think of-
Lostinrehab I will also attempt to flash the bios tonight as well.
 
Ok figured out why I can't flash it. The LT motherboards have a specific bios separate from the 680i. I need to find it in .ISO format so I can boot off the cd since I have no floppy, and then I will be able to see how that works.
 
Ok update on the RAM. Tested each chip individually, in every slot on the motherboard.

Each 1gb chip passed one full passthrough of Memtest in each slot on the motherboard individually, while CPU overclocked.

I then put both chips in. 8 minutes later crash-just hard freeze. I moved them to slots 2-4 rather than 1-3. Crash, one time after 10 minutes, one time after 1hr 28 minutes. So preliminary testing shows the RAM works fine individually, but for some reason causes massive instability (yet shows NO errors in memtest) with both chips of RAM in and CPU overclocked.

At stock speeds everything works fine.

For all thats holy, whats going on?? :eek:
 
Well if you still have your other ram you could try testing it one stick at a time as well to see if they cause problems. I'm having a hard time understanding how your ram has problems since it works fine at default settings DDR2 1000 and 5-5-5-15 at v2.2 ... Unless you're running the ram with a 2:3 divider the ram should not be running higher than spec. If you havent tried it already try running the ram with a 1:1 divider. If you're still having problems it could be either one of 3 things PSU, motherboard , ram again. A good psu can mean the world of difference in overclocking and I'm not sure your Mushkin 550W would rank among the most reliable. You could have a dud motherboad that could be bad with ram or even killing it o.o.
 
Ok, I am sorry, I want to make sure I am not being confusing. I am having issues solely with stability overclocking. My first RAM, 2x1 GB Mushkin DDR2 800 went bad in a few weeks-massive errors detected in memtest. I RMAd that and bought the Ballistix from Buy.com in the meantime.

Currently, I have not found stability issues running everything at stock speeds, both RAM and CPU. However, when I try to overclock, I get problems that I have never had before. In summary:
1. When I have both Ballistix stick in my system, at any moderate overclock (3.0-3.375ghz) with voltages jacked, I get hard freezes in both windows and memtest. I do not, however, get any errors in memtest prior to freezing. No amount of moving the two sticks around changes anything.
2. When using only ONE stick in my system at 3.0-3.375ghz, I am stable and have yet to crash. This is at both 1000mhz 5-5-5-15 and the less impressive 800 at 5-5-5-15 just for example.
3. When testing each 1gb stick individually in Memtest for over 2 hours each, I have not had any Memtest errors.
4. When testing both sticks 2x1gb in memtest, after 7 hours I have no errors (at stock CPU speeds).
5. As soon as I overclock CPU with both chips in the board, I get near constant hard freezes.
6. Prior to my Mushkin RAM going bad, I never had any issues of hard freezing, in fact I think my computer did it once at 3.7ghz OC and that was it.

So I hope that clarifies my issues-its basically completely unstable at CPU overclocks, and I don't understand why. Could my PSU have gone bad in the last week or something? Also-how do I make sure the RAM is running 1:1?

Thanks a ton lostinrehab, I owe you one!
 
No one likes to hear this but DDR2-1000 memory modules are usually DDR2-667 ICs that have been over volted and factory overclocked.

The best way to trouble shoot is to lower the speed you are trying to run your memory at.
CL5-5-5-15 and DDR2-800 at 2.2 volts would be a good place to start.

Before you do any overclocking of your CPU you want 100% stability. That means zero errors while running MemTest86+, preferrably over night. Use default CPU MHz and default CPU core voltage of 1.325 volts.

I'd also run Orthos for 4 hours if you can. You want to be able to say, I'm 100% sure that my memory is capable of this speed.

Raise the MHz of your CPU while maintaining your memory at the speed that you have proven it can run at. Don't raise the CPU voltage until you need to. Personally I like using a simple benchmark like SuperPi mod. Even a quick 1M or 2M test will produce errors when you have gone too far with your overclock and need to add some more voltage. When you reach your overclocking goal and are SuperPi stable, then you can run Orthos and start checking for stability with that. You will usually need to bump cpu voltage again for Orthos stability.

The key is to isolate memory and cpu so you can test each one individually. Your DDR2-1000 might not live up to the promise and you might have to RMA it but by running it much lower than what Crucial rates it at, you might be able to find out what's really holding you back.

Edit: CPUz shows you what MHz your memory is running at and SpeedFan is pretty accurate on most boards for reporting CPU voltage. If you have a digital multimeter then check for voltages on all 3 rails at idle and at full load as well. People are sometimes surprised when they take their power supply for granted.
 
Unclewebb-
I am going to try everything you just listed, but I had already tested the RAM at 800mhz at 5-5-5-15 and still experienced some freezing. Also, two nights ago I ran Memtest with everything at stock speeds for the 6 hours I was asleep at 1000mhz and had zero errors. So that would seem to indicate that the RAM does run at its rated speed, at least at stock. However, I will run more tests to make sure of this before deciding that the RAM is just not capable of overclocking.

If worse comes to worse when my Mushkin RAM comes back I will plug that in. Before I had all the errors I found it both fast and stable!

Also, I don't have a method of testing PSU wattage, so replacing that will be my hopefully unnecessary last step.
 
CalvinGT said:
However, just 5 minutes ago I attempted to boot up at a modest 3.2 with 1.40v and the RAM at the defaults specs. Twice it froze within a minute.
Click to expand...

I think that is your problem, you are pushing your cpu beyond its limits at the voltage your running.

Every CPU is different, and what you see others getting is no guarantee that you will get the same results, even with the exact same hardware and software. There is no way my cpu will do 3.2GHz on @ 1.4v

For example my cpu will run up to 2.90GHz stable at default voltage. But If I want 375x8= 3.0GHz orthos stable I need to raise the v core to 1.4v. And if I want 3.20GHz orthos stable I have to raise it all the way to 1.5 v core before I am stable. My cpu is just not a great oc'er. No problem for me because I am satisfied were I am at.

I know you posted earlier that you were able to get to "stable at about 1.375 at 3.3ghz right out of the box". But It could be that your cpu was not very happy there and its just showing up now. 1.375v would not get me to orthos stable at 3.0GHz.

Try 3.0GHz at 1.40v or even 1.45v and see if your stable there, Just to rule out the cpu being the limiting factor.
 
One more thing. If your motherboard is running your memory at a command rate of 1T then change that to 2T. CPUz should show you what it is running at.

You lose a little performance at 2T but you might get a lot better stability. Some memory just doesn't like running at 1T, especially at high MHz.
 
CalvinGT said:
Also-how do I make sure the RAM is running 1:1?
Click to expand...

Are you running the FSB and Memory clock mode linked or unlinked. If you are running it linked then after you set the memory speed to 1000 and start overclocking the fsb then you're actually overclocking the ram higher than ddr 2 1000. If you are running it unlinked then there shouldn't be a problem but try it linked and set your fsb to 266 and your ram to 533 so that it runs on a 1:1 ratio. After that you can try overclocking the fsb to 356 this would make your ram speed DDR2 712.
 
CpuMan said:
I think that is your problem, you are pushing your cpu beyond its limits at the voltage your running.

Every CPU is different, and what you see others getting is no guarantee that you will get the same results, even with the exact same hardware and software. There is no way my cpu will do 3.2GHz on @ 1.4v

For example my cpu will run up to 2.90GHz stable at default voltage. But If I want 375x8= 3.0GHz orthos stable I need to raise the v core to 1.4v. And if I want 3.20GHz orthos stable I have to raise it all the way to 1.5 v core before I am stable. My cpu is just not a great oc'er. No problem for me because I am satisfied were I am at.

I know you posted earlier that you were able to get to "stable at about 1.375 at 3.3ghz right out of the box". But It could be that your cpu was not very happy there and its just showing up now. 1.375v would not get me to orthos stable at 3.0GHz.

Try 3.0GHz at 1.40v or even 1.45v and see if your stable there, Just to rule out the cpu being the limiting factor.
Click to expand...

The thing is, the difference in stability happened after I switched RAM, which surprised me. I totally understand what you are saying and know that you are 100% right and it simply might be my CPU, but I did 4 hours stable on Orthos at 3.3 @ 1.4v yet I can barely get into Windows with those settings with this new RAM, literally a week later. Such a large divergence based on only one new or changed part seems awfully suspicous to me you know?

Thanks for the thoughts!
 
LostInRehab said:
Are you running the FSB and Memory clock mode linked or unlinked. If you are running it linked then after you set the memory speed to 1000 and start overclocking the fsb then you're actually overclocking the ram higher than ddr 2 1000. If you are running it unlinked then there shouldn't be a problem but try it linked and set your fsb to 266 and your ram to 533 so that it runs on a 1:1 ratio. After that you can try overclocking the fsb to 356 this would make your ram speed DDR2 712.
Click to expand...

Ok normally am running unlinked. However, for testing earlier I tried it linked and the stability problems continued. I have spent the last 4 hours running various tests, and from what I can see, my preliminary conclusions are that this RAM/CPU/MOBO combination can support no higher than 3.15 ghz stable in memtest and Orthos without hard freezes. I upped voltages to absurd numbers (1.5cpu and 2.4ram) to try to get stable in orthos or memtest at 3.25 and 3.3 and 3.375 and it simply would not work. Period. I think (I hope) that this is just a bad combo of RAM/CPU/MOBO. The mushkin's while clearly DOA for their recommended specs, at least never froze the system like this, they either showed errors in memtest or caused orthos to "beep" and stop.

So, my plan as I go out for the day-run Orthos and Memtest at 3.15 and try to determine stability, and then overnight run one and then the other at STOCK speeds for both and see what results I get. Worse comes to worse I will run this sucker stock until the Mushkin RMA comes back and then compare. If it still doesnt work then I will know it might be mobo/cpu.

Is this a good plan? Thoughts?

BTW, HardForums are one hell of a community. Thanks guys!
 
CalvinGT said:
Ok normally am running unlinked. However, for testing earlier I tried it linked and the stability problems continued. I have spent the last 4 hours running various tests, and from what I can see, my preliminary conclusions are that this RAM/CPU/MOBO combination can support no higher than 3.15 ghz stable in memtest and Orthos without hard freezes. I upped voltages to absurd numbers (1.5cpu and 2.4ram) to try to get stable in orthos or memtest at 3.25 and 3.3 and 3.375 and it simply would not work. Period. I think (I hope) that this is just a bad combo of RAM/CPU/MOBO. The mushkin's while clearly DOA for their recommended specs, at least never froze the system like this, they either showed errors in memtest or caused orthos to "beep" and stop.

So, my plan as I go out for the day-run Orthos and Memtest at 3.15 and try to determine stability, and then overnight run one and then the other at STOCK speeds for both and see what results I get. Worse comes to worse I will run this sucker stock until the Mushkin RMA comes back and then compare. If it still doesnt work then I will know it might be mobo/cpu.

Is this a good plan? Thoughts?

BTW, HardForums are one hell of a community. Thanks guys!
Click to expand...

So you were running unlinked. Try dropping the ram voltage to 2.1 I use a pair of crucial tenth anniversary ram at 950 and I know that my sticks don't like v2.2 or more. Also like another person said make sure the command rate is t2.
 
LostInRehab said:
So you were running unlinked. Try dropping the ram voltage to 2.1 I use a pair of crucial tenth anniversary ram at 950 and I know that my sticks don't like v2.2 or more. Also like another person said make sure the command rate is t2.
Click to expand...

I have been running at T2 all along, I will try the RAM at 2.1 though. I know the specs say 2.2, but that doesn't really mean much in the long run-I will not be able to try until I get home though!
 
Final update of the night. Tried all suggestions, run the computer doing various tests for almost 12 hours today.

Bottom line: at no combination of voltages, etc. can I get the computer to run without hard locks/freezes past 3.1ghz with both 1gb RAM sticks installed at the stock RAM speeds. I am not getting orthos errors indicating voltage issues. I am not getting Memtest issues (ran memtest 6 hours plus at ALL stock speeds with zero errors). I am getting hard lock/crashes that lead me to believe that when both sticks of RAM are in my computer, it causes some sort of very odd hardware conflict and just will not overclock very much. I have no idea why, but I have tested with just one stick with better results.

At this point, absent any further suggestion from you guys, before I worry that the motherboard and or PSU might be an issue, I am going to wait for my Mushkin RMA RAM to come back (hopefully the end of this week) and run some tests with that and compare results. For now I am stable running stock speeds and can live with that.
 
Final update. Have replicated all steps suggested and replicated all of my original testing.

This RAM just will not allow me to overclock at all past roughly 3.1ghz, and even that is not guaranteed long-term stability (ie, 8 hours plus). Hoping I will see my RMAd RAM by this friday, but that might be wishful thinking-at that time I can test it and see if it might be a mobo or psu issue.

Any final thoughts?
 
Could be your northbridge. I've had the exact same problem where it would not show any errors in memtest but would crash out even with the CPU at stock speeds. Also the fact that the memory works in single channel but not dual channel is an indication of a problem with the memory controller on the northbridge. Have you tried upping the MCP voltage? Also replacing the northbridge goo with A5 could help.

Mine runs 24/7 orthos stable at 7x450 = 3.4 GHz with MCP @ 1.65V, CPU @ 1.45V w/ FSB @ 1.5V, Crucial Ballistix 2x1GB synced @ 2.3V, but ONLY after I upped the MCP volts to 1.65 and applied AS5, before that I could not get past 410 FSB.

Normally I run it at 8x400 = 3.2 though (with much lower volts), I really can't tell the difference outside of benchmarks :)
 
Thanks for the info man, two questions: what is the MCP, and how much above stock should I mess with the voltage? Also, what timings do you run the Ballistix at that overclock?

The mushkin should be here end of week, so I can test if this is a fault memory controller and all the other worst case scenario stuff.
 
You must log in or register to reply here.
Back
Top