d9gmh weakening?

[ziddey]

I bought a 2x1gb buffalo firestix 800 at the end of last year when they were still d9gmh. Up until a week or so ago, I was able to run 800 3-3-3-8 and 1000 4-4-4-12, but recently, I haven't been able to anymore. I thought it was because of a different motherboard I was using, so I went back to the original, where it was a known good. Same results. I'm at 900 right now and 4-4-4-12 most times won't even post anymore. I have to ease timings to 5-5-5-15 to achieve stability. It used to be rock solid memtest 24+ hrs. I've only pushed 2.3v max (most of its life has been spent at 2.0v and 2.2v). I don't have a fan directly on the ram, but case airflow goes through the ram since I have a top mounted 2x120 rad intake. Anyone else have this happen?

[bofors]

I bought a 2x1gb buffalo firestix 800 at the end of last year...

Do you really mean the end of 2006 (or 2005)?

[ziddey]

Ah, sorry. 2006. I got the sticks around mid-november. And seeing how tight the timings were initially, I assumed they'd be d9's

[bofors]

So, you have had these RAM for only two months. That doesn't like enough time to "wear out" anything at the voltages you are talking about.

I think it is more likely that the RAM has a specific defect which has now become the limiting factor.

[ziddey]

Lol that's definitely not what I wanted to hear. Come to think of it, the only time I even ran 2.4v was with a 120mm over it when I had it on the bench. It's never seen more than 2.3v in a case. Aww shucks

[SPL15]

I'm having the EXACT same issues with my crucial ballistix DDR1000. I thought I was just losing my skills... I've been going absolutely nuts over this trying to figure it out!!

When I first got my memory it did 1000 and no higher with 5,5,5,18 with 2.28 Volts on my XBX2 on 1066 strap using 800Mhz ram speed (1:2 ratio). That was my highest overclock with this memory and it was rock solid stable. At the factory timings with factory rated volts it was not stable at all. I have an 80mm fan directly on top of the memory for cooling and the motherboard says motherboard temps never get higher than 37C at the temp probe between the memory banks. Memory runs luke warm to the touch

After i tested everything to make sure things werent DOA and ran at spec and all that stuff, I started overclocking. I set RAM speed to 533 (1:1 ratio) and I could run 425Mhz fsb (ddr850) perfectly fine all day 12 hours orthos stable etc. using 2.24 volts and with 5,5,5,15 timings. Also bus voltages were 1.375 and 1.675 on 1066 strap.

After running prime95 for several days i started getting errors and redid my overclocking settings down to 417Mhz (DDR834) at 3,4,4,6 timings @ 2.28volts fully stable (12 hours othos etc.) I started running prime again and for about 3 days it lasted 100% stable, but I started to get errors again.

So I boosted the 1.375 volt bus voltage to 1.4 volts and it was stable again for about 1 day. So then I had to change timings to 5,4,4,12 @2.28 volts to get halfway stable again. Which lasted about 3 days.

Currently I have to run @ 5,5,5,14 timings at 2.28 volts in order to get it stable @ 417fsb. It absolutely will not do DDR1000 anymore no matter what the timings are. And, even at the current crappy timings I am using, I get a random error every couple of days.

Definitely NOT happy with Crucial and Newegg right now. Crucial does not respond to my calls or emails and Newegg says I need confirmation from Crucial to send back my memory... I told them they are full of "it" with "sh" in front of "it". This memory is going back with or without crucials approval. Newegg also said that they would charge 15% restocking fee to switch this crap for something that works at the speed it is specified to work at, but that offer ended 3 days ago... I hate RMA's

Edit: Memory was purchased 12/18/06, so this degradation happened over the course of 1 month. I currently can run 100% stable again with 2.32 volts 415fsb 5,5,5,18 timings. my old 417fsb is no longer stable not matter the volts, timings, strap, etc.

Edit: Memory is BL2KIT12864AA1005 purchased from Newegg. If you look at customer reviews, I'm "SUPP" and that was my review several weeks ago. Also this memory has never seen higher than 2.32 volts in Bios. Actual measured voltage was 2.306volts on $250 calibrated DMM I have for work.


Edit: I've also tried BIOS 2330, 2333,2395, and latest BIOS from Intel. all give the same results

[Revv23]

newegg's rma policy is pretty clear, its not thier fault your crucial is going bad. theyt will exchange them without a fee. you just cant get your money back.

[bofors]

When I first got my memory it did 1000 and no higher with 5,5,5,18 with 2.28 Volts on my XBX2 on 1066 strap using 800Mhz ram speed (1:2 ratio). ...
At the factory timings with factory rated volts it was not stable at all.

You clearly received defective memory, I would have sent it back then.

Just get another set from NewEgg now, you will likely be very happy with normal Ballistix.

[The Stilt]

I´ve seen these D9GMH Buffalo Firestix degrading too many times :mad:

I don´t know if it is the pcb (Dynamic 2DUE28F-AA) which causes this or just low quality chips.

I have sold a lot of them and almost ~ 50% of them have degraded in few months. That is not because of high voltage since no more than 2.2V - 2.3V have been used.

[SPL15]

newegg's rma policy is pretty clear, its not thier fault your crucial is going bad. theyt will exchange them without a fee. you just cant get your money back.

I know Newegg has no connection with the manufacturing of Crucial Memory. The two people I chateed with online at Newegg told me that in order to RMA them, I have to get confirmation from Crucial to return them for money back with no 15% restocking fee OR to RMA them for an exchange. Basically they both told me I needed to get verifiable confirmation from Crucial that these modules are in fact bad in order to send them back for an exchange or refund without 15% restocking fee.

Since I cannot get a response from Crucial, it makes it a bit difficult to get verifiable confirmation from the manufacturer that these modules are defective. Newegg's return policy has no statement requiring the manufacturer to authorize an RMA return. That's why I'm a little bitter at Newegg at the moment. I've never had to get manufacturer permission to send back defective products with any vendor I have dealt with.

I only buy form Newegg cause I've had issues with every other distributer I have dealt with. Prices may not be the absolute best, but I know if they say something is in stock, it is instock and ready to go and will be at my door in 3 days using ground service.

Newegg's return policy is very clear indeed. No where does it say I need Manucacturer authorization to return defective products.

[bofors]

I´ve seen these D9GMH Buffalo Firestix degrading too many times :mad:

I don´t know if it is the pcb (Dynamic 2DUE28F-AA) which causes this or just low quality chips.

I have sold a lot of them and almost ~ 50% of them have degraded in few months. That is not because of high voltage since no more than 2.2V - 2.3V have been used.

So this is a Firestix issue per se, and not necessarily a D9GMH problem either.

[bofors]

The two people I chateed with online at Newegg told me that in order to RMA them, I have to get confirmation from Crucial to return them for money back with no 15% restocking fee OR to RMA them for an exchange. Basically they both told me I needed to get verifiable confirmation from Crucial that these modules are in fact bad in order to send them back for an exchange or refund without 15% restocking fee.
...
No where does it say I need Manucacturer authorization to return defective products.

It sounds like you are letting NewEgg jack you around. I bet you are giving them too much information, details about overclocking the RAM and such. All you have to do is say that the RAM does perform within the spec. and demand a RMA# for exchange. If you do not get an RMA# immediately, ask for a manager. Be polite, but make sure they understand that you are pissed.

[afireinside]

I had a set of 6400 ballistix that went from being crappy out of the box to not posting 500mhz 5-5-5-15 2.5v in about 3 weeks. I ended up RMAing them and selling. Seems some defective d9/pcbs are making it off the production line.

[SPL15]

I´ve seen these D9GMH Buffalo Firestix degrading too many times :mad:

I don´t know if it is the pcb (Dynamic 2DUE28F-AA) which causes this or just low quality chips.

I have sold a lot of them and almost ~ 50% of them have degraded in few months. That is not because of high voltage since no more than 2.2V - 2.3V have been used.


My guess would be poor quality chips. This sounds like classic Electromigration problems from poorly designed (or slightly defective) IC's. I deal with this kind of stuff on a daily basis at work (not computer related). We found a cheap source for a 8bit processor and initial testing and verifying showed they worked just as good as our original spec parts even after our strenous 75C temp test and 20,000volt surge tests, but we had a 12.5% 1 year failure rate in the field compared to 1.75% with our previous IC vendor.

Sent the defective chips in for die analysis and electromigration from poor fabrication was the culprit. we sent the same IC but one that was in our test fixture and still good had a little electromigration, but not bad. This would mean that some of the IC's were actually good, but a high percentage of them had small defects that would only surface over time. Even the bad ones still worked, buit would work unreliably and had major heat issues.

Basically what would happen in the chip was that transistors would lose tolerance due to poor fabrication and develop excessive leakage currents which would gradually cause very small pits to develop in the circuit paths in the IC and cause parasitic resistances which drastically reduced slew rates thus making signal integrity very poor. The ICs would still work, but unreliably and not to the specified speed. Also, the IC's measured current consumption was 3 times normal and the chips would get hot to the touch.

These ICs were from a very reputable manufacturer, when we contacted the manufacturer, they said they had a new process that reduced silicon usage by 14% in most of their Programmable microcontroller IC's and up to 18% in normal logic IC's....

When Semiconductor manufacturers "improve" their current product line, it usually means they found ways to reduce the cost of manufacturing, not improve performance. You gotta realize, companies arent here to provide customers with great products, they exist to please investors who associate "improved" with reduced costs. Customers are just a means to an end for publically owned corporations. If a new reduced cost IC has 10% higher falure rate, but reduces costs by 30%, this is improvement according to investors and that's who companies care about. failure rates only matter to companies when they effect sales.

D9 Fatbodies only went out of existance because of costs, performance was there, but reduced costs are what their true customers (investors) want.

[SPL15]

It sounds like you are letting NewEgg jack you around. I bet you are giving them too much information, details about overclocking the RAM and such. All you have to do is say that the RAM does perform within the spec. and demand a RMA# for exchange. If you do not get an RMA# immediately, ask for a manager. Be polite, but make sure they understand that you are pissed.


Oh dont worry, I'm well aware not to disclose OCing information. I played dumb, but smart. Basically told them memory doesnt work at specified timings and voltage, and it ONLY works at DDR 800 with 2.2 volts at 5,5,5,15. That is as far as I got into my system and all of it is true. They said they would be happy to replace or refund IF i could email or fax them confirmation from Crucial that these modules were bad. Got that from 2 people.

I'm calling them today, my system isnt even stable with 5,5,5,18 anymore. and this has only been a couple of hours today of prime95. These modules are going down hill very quickly now.

[ziddey]

Eeeeeh. This is getting pretty bad. I'm at 900 5-5-5-15 2.3v and even with a known good motherboard (where I was able to do 500fsb 1:1 4-4-4-12 2.3v rock solid), it won't pass through more than an hour or so of large fft. I had this exact same problem on a different motherboard which was why I went back to an older one to verify. I did test last night in the other motherboard at 900 5-5-5-15 2.1v and it passed all night fine. I even did 3 hours loop of test 5. I don't get it.

Will try jacking up vcore to 1.59v again, even though cpu was previously tested a-ok at 1.44v for both small and blend, and still tests ok on small.

[bofors]

Sent the defective chips in for die analysis and electromigration from poor fabrication was the culprit. we sent the same IC but one that was in our test fixture and still good had a little electromigration, but not bad. This would mean that some of the IC's were actually good, but a high percentage of them had small defects that would only surface over time. Even the bad ones still worked, buit would work unreliably and had major heat issues.

This is basically what I suspect is going on with these Firestix, they are probably low grade D9GMH (with lots of non-fatal defects that quickly become exposed). But the DDR2-1000 Ballistix are supposed to be high bin D9GMH, mine certainly have not degraded in the two months that I have had them. You just happened to get a bad pair.

[bofors]

They said they would be happy to replace or refund IF i could email or fax them confirmation from Crucial that these modules were bad. Got that from 2 people.

I sent some RAM that failed back to NewEgg in November, I did not have to do anything special. I just told them the RAM would not POST and gave a BIOS beep error code. NewEgg gave me an RMA #, credit for return shipping and a new set in exchange.

[Revv23]

I know Newegg has no connection with the manufacturing of Crucial Memory. The two people I chateed with online at Newegg told me that in order to RMA them, I have to get confirmation from Crucial to return them for money back with no 15% restocking fee OR to RMA them for an exchange. Basically they both told me I needed to get verifiable confirmation from Crucial that these modules are in fact bad in order to send them back for an exchange or refund without 15% restocking fee.

Since I cannot get a response from Crucial, it makes it a bit difficult to get verifiable confirmation from the manufacturer that these modules are defective. Newegg's return policy has no statement requiring the manufacturer to authorize an RMA return. That's why I'm a little bitter at Newegg at the moment. I've never had to get manufacturer permission to send back defective products with any vendor I have dealt with.

I only buy form Newegg cause I've had issues with every other distributer I have dealt with. Prices may not be the absolute best, but I know if they say something is in stock, it is instock and ready to go and will be at my door in 3 days using ground service.

Newegg's return policy is very clear indeed. No where does it say I need Manucacturer authorization to return defective products.


No where on newegg's website does it say that they will accept a return without a restocking fee under any circumstances, so by taking them back without restocking fee, i can understand that they want proof that the sticks are bad, and that you arent just trying to aviod the restocking fee. This is the first i have ever heard of newegg waving the restock fee and i think its pretty cool of them to try and meet you halfway on this.

Like i said you can still exchange them without manufacturer authorization, you just can't refund them in full.

as for the reason the sticks are dieng, my bet is on the PCB, if it were the IC we would see tons of dead or dieing GMH in all brands and not just in these firesticks. That being said, that bad PCB could be causing the electron migration, hence the symptoms.

[irev210]

i've used a few sets of d9gmh firestix, all without issue *shrug*


also, newegg RMA for replacement... just do it online, no cost.

[bofors]

as for the reason the sticks are dieng, my bet is on the PCB, if it were the IC we would see tons of dead or dieing GMH in all brands and not just in these firesticks. That being said, that bad PCB could be causing the electron migration, hence the symptoms.

I am not betting on the circuit board here at all, its traces and connections are hundreds if not thousands of times wider and thicker than the microcircuitry on the D9GMH IC's. That means the circuit boards should be much more durable and much less likely to have fabrication defects which degrade performance.

My bet is that Micron had a very crappy batch of D9GMH (with an unusually high defect density, perhaps do some equipment failure during fabrication, like maybe a vacuum pump died) which they sold to Buffalo.

PS: The phenomona of atomic flow under voltage is "electromigration" not "electron migration" (which implies that electrons are moving and obviously is not a problem): http://en.wikipedia.org/wiki/Electromigration

[syne_24]

I know Newegg has no connection with the manufacturing of Crucial Memory. The two people I chateed with online at Newegg told me that in order to RMA them, I have to get confirmation from Crucial to return them for money back with no 15% restocking fee OR to RMA them for an exchange. Basically they both told me I needed to get verifiable confirmation from Crucial that these modules are in fact bad in order to send them back for an exchange or refund without 15% restocking fee.

Since I cannot get a response from Crucial, it makes it a bit difficult to get verifiable confirmation from the manufacturer that these modules are defective. Newegg's return policy has no statement requiring the manufacturer to authorize an RMA return. That's why I'm a little bitter at Newegg at the moment. I've never had to get manufacturer permission to send back defective products with any vendor I have dealt with.

I only buy form Newegg cause I've had issues with every other distributer I have dealt with. Prices may not be the absolute best, but I know if they say something is in stock, it is instock and ready to go and will be at my door in 3 days using ground service.

Newegg's return policy is very clear indeed. No where does it say I need Manucacturer authorization to return defective products.


I agree with you about the authorization needed from the manufacture is rather abnormal procedure for newegg. AFAIK I have never needed approval from any other vendors from newegg before. Perhaps maybe this is on Crucial end because they are strict on their RMA policy. I would try calling again to a different sales rep and maybe they can hook you up. My friend recently RMA a pair of Ballistix that was over 30 days and he got no restocking fees or needed approval from Crucial.

[Revv23]

I am not betting on the circuit board here at all, its traces and connections are hundreds if not thousands of times wider and thicker than the microcircuitry on the D9GMH IC's. That means the circuit boards should be much more durable and much less likely to have fabrication defects which degrade performance.

My bet is that Micron had a very crappy batch of D9GMH (with an unusually high defect density, perhaps do some equipment failure during fabrication, like maybe a vacuum pump died) which they sold to Buffalo.

PS: The phenomona of atomic flow under voltage is "electromigration" not "electron migration" (which implies that electrons moving and obviously is not a problem): http://en.wikipedia.org/wiki/Electromigration


That makes sense, thanks for the clarification. :) :toast:

[Revv23]

I agree with you about the authorization needed from the manufacture is rather abnormal procedure for newegg. AFAIK I have never needed approval from any other vendors from newegg before. Perhaps maybe this is on Crucial end because they are strict on their RMA policy. I would try calling again to a different sales rep and maybe they can hook you up. My friend recently RMA a pair of Ballistix that was over 30 days and he got no restocking fees or needed approval from Crucial.


You can RMA without authorization, however you cant get a 100% refund without authorization.

[SPL15]

You can RMA without authorization, however you cant get a 100% refund without authorization.


Newegg's return policy says that you can get 100% refund IF defective. If your just returning product because you dont like some particular aspect of it, then the 15% restocking fee applies.

I think it is worded something like all returns subject to 15% restocking fee unless defective.

My case the memory is defective.

I still think it is lame to require manufacturer authorization for a return, unless your policy specifically states that it is needed. And even then, I think it is still lame to require authorization from the manufacturer to return product unless your actually shipping the product to the manufacturer.

I got put on hold for 15 minutes yesturday waiting to talk to someone @ Newegg and didnt have time to wait longer, I'll try again Monday.

Thanks for your guys' input, I'll make sure I get a fair deal on this memory.

Sorry for the thread Jack also. I didnt mean to make this thread about my issues with Newegg and Crucial.

[ziddey]

lol it's ok. I don't think the thread would have gotten much attention otherwise ;)

[SPL15]

I just did the online RMA process.... Geez I shoulda just did that to begin with. thanks for the tip irev210.

Saved me 30 minutes on my cell minutes. Plus i didnt have to get nasty with anyone and make me not want to use Newegg anymore.

For anyone wanting to know, Newegg has been 100% reliable for me in the past, fast shipping, instock items actually instock, etc., etc.

This memory is doing very odd things now. if I have my system off over night it works good at 415mhz 5,5,5,15 @2.2 volts, but orthos will fail after about 30 minutes, then it goes down hill from there until I reduce fsb to around 400mhz.

Doubt it is my cooling, I've got fans on my north & south bridge and 1 inch above the memory.

Another wierd issue that I get is that I get the 3 beep memory error from my XBX2 when I change BIOS. I have to remove battery and power for several hours before it will boot up again.

One thing I was wondering, I use suspend to ram (S3 suspend) a lot since this computer is my HTPC and I like the almost instant boot up. From what I just read S3 suspend loads everything into ram and shuts down the rest of the computer including fans. The ram is still running and has the OS data on it until power up. Basically it is acting like hibernation except instead of righting it to the disk for recovery it just keeps everything in ram. I also use intel quick resume technology (although I fail to see what it is actually doing since it seams like a normal S3 suspend.)

Since I was using 2.28 volts in a pretty tight case, could this be causing the memory to over heat since it is still running, but with no fans running? Does S3 suspend change RAM voltage/timings, refresh rate?

I've felt these modules without a fan on them with 2.2 volts and they get pretty hot. With 2.28 volts with no air flow for 12 to 24 hours could this be causing these modules to overheat and thus degrade in such a small time frame?

If anyone has better input on S3 suspend please speak up.

[GPSeek]

I just did the online RMA process.... Geez I shoulda just did that to begin with. thanks for the tip irev210.

Saved me 30 minutes on my cell minutes. Plus i didnt have to get nasty with anyone and make me not want to use Newegg anymore.

For anyone wanting to know, Newegg has been 100% reliable for me in the past, fast shipping, instock items actually instock, etc., etc.

This memory is doing very odd things now. if I have my system off over night it works good at 415mhz 5,5,5,15 @2.2 volts, but orthos will fail after about 30 minutes, then it goes down hill from there until I reduce fsb to around 400mhz.

Doubt it is my cooling, I've got fans on my north & south bridge and 1 inch above the memory.

Another wierd issue that I get is that I get the 3 beep memory error from my XBX2 when I change BIOS. I have to remove battery and power for several hours before it will boot up again.

One thing I was wondering, I use suspend to ram (S3 suspend) a lot since this computer is my HTPC and I like the almost instant boot up. From what I just read S3 suspend loads everything into ram and shuts down the rest of the computer including fans. The ram is still running and has the OS data on it until power up. Basically it is acting like hibernation except instead of righting it to the disk for recovery it just keeps everything in ram. I also use intel quick resume technology (although I fail to see what it is actually doing since it seams like a normal S3 suspend.)

Since I was using 2.28 volts in a pretty tight case, could this be causing the memory to over heat since it is still running, but with no fans running? Does S3 suspend change RAM voltage/timings, refresh rate?

I've felt these modules without a fan on them with 2.2 volts and they get pretty hot. With 2.28 volts with no air flow for 12 to 24 hours could this be causing these modules to overheat and thus degrade in such a small time frame?

If anyone has better input on S3 suspend please speak up.
I don't know more about S3 than you do. I just want to add that I also think that S3 can be a real problem to the ram sticks.

I fried a Crucial Ballistix kit running at 400MHz 3-3-3-8 at 2.32v in about 3 weeks.:slapass: I only left the machine in suspend mode for about 4 hours in an afternoon. The machine returned to Windows without any problems when I came back to it. After working with it for a while (1 hour or so I forgot), I got a BSOD with a system dump. After the BSOD, the machine could not boot with these sticks at all. I tried these sticks in a intel 965 board to make sure it was the memory problem. It booted once out of 6 or 7 trials. However, the windows crashed right after loading.

So I thought it might be the heat in the fanless suspend mode that killed the sticks. :slapass:

I never let it run into suspend mode again. I may do so after I figure out how to keep the fan blowing to the memory running in the suspend mode.
I also ordered another pair of Ballistix in the hope that not all of the 4 sticks die together.:)

[GPSeek]

The Crucial Ballistix PC8000 and Crucial Ballistix Tracer PC8000 are good performers anyways.:fact:

However, they are double sided sticks, that's not so good in terms of heat dissipation IMHO.:slapass:

[SPL15]

The Crucial Ballistix PC8000 and Crucial Ballistix Tracer PC8000 are good performers anyways.:fact:

However, they are double sided sticks, that's not so good in terms of heat dissipation IMHO.:slapass:


I forget who did the tests, but I've read that the dual sided modules outperform single sided by a decent margin. The test I read did single vs dual channel and single sided modules vs double sided. Of course the dual channel mode beat single channel, but it showed measurable performance boost with using fully populated memory modules in dual channel mode.

I dont think heat should be an issue when my system is on. I've looked into the S3 suspend a bit deeper and the only other thing I can see is that the computer puts the memory into a slow refresh state. If I remember right, DDR and DDR2 are a use SRAM technology which stands for STATIC ram which uses static charges as the logic 1 and 0. SRAM is quick, but it needs to be constantly "refreshed" in order to keep the data.

If the my motherboard is reducing refresh rate, but is still running the modules at 2.28volts, I could definitely see how overheating could be an issue. I've only had one instance where my computer didnt boot out of S3 correctly, I didnt think too much of it at the time, but I'm starting to wonder how safe S3 suspend is when using elevated voltages on high performance ssystems.

I'll do tests when I get home. its a real pain to get to my memory in my case. I use a Silverstone HTPC case with watercooling, 2 optical, 4 WD raptors, and a 8800GTX... I've gotta half way disassemble my computer just to reach the stupid CMOS jumper.

[GPSeek]

I've gotta half way disassemble my computer just to reach the stupid CMOS jumper.
you can install a switch to replace the jumper as other members do:)

[ziddey]

I don't know more about S3 than you do. I just want to add that I also think that S3 can be a real problem to the ram sticks.

I fried a Crucial Ballistix kit running at 400MHz 3-3-3-8 at 2.32v in about 3 weeks.:slapass: I only left the machine in suspend mode for about 4 hours in an afternoon. The machine returned to Windows without any problems when I came back to it. After working with it for a while (1 hour or so I forgot), I got a BSOD with a system dump. After the BSOD, the machine could not boot with these sticks at all. I tried these sticks in a intel 965 board to make sure it was the memory problem. It booted once out of 6 or 7 trials. However, the windows crashed right after loading.

So I thought it might be the heat in the fanless suspend mode that killed the sticks. :slapass:

I never let it run into suspend mode again. I may do so after I figure out how to keep the fan blowing to the memory running in the suspend mode.
I also ordered another pair of Ballistix in the hope that not all of the 4 sticks die together.:)
Hmm, yikes. I've got a ds3 which is more or less the same thing. I used to also run my d9gmh's at 400 3-3-3-8 2.3v for a while. Never had any problem, though I can't do it anymore. I did use a DMM to ensure ram was seeing 2.3v.

[SPL15]

you can install a switch to replace the jumper as other members do:)


Yea, I'm lazy and dont have SPDT switch and header right now... Maybe I can mistakenly buy the parts in my next order at work and make one. it would sure as heck make things easier.

Since this is a HTPC Media Center I was kinda planning on a "set it & forget it" type of system, but this new computer has gotten me back into Overclocking and pushing limits again...

I just did a little bit of testing and the memory runs almost cool to the touch when in S3 suspend for an hour after running orthos for 15 minutes to heat things up first. The motherboard must throttle voltage too when it is in slow refresh mode because Temperature is definitely not elevated in S3 Suspend.

You can tell there is a little heat being generated, but absolutely nothing to worry about. When everything else is cold to the touch, the memory is barely luke warm after being on Suspend for an hour. I would feel safe putting a blanket on top of the memory it runs that cool.

Since this memory is going back tomarrow, I'm doing some elevated voltage tests and so far:

Unstable - 417Mhz - 4,4,4,12 @ 2.44V
Stable - 417Mhz - 4,4,4,12 @ 2.48V
Unstable - 435Mhz - 4,5,5,15 @ 2.48V


My fan keeps these modules luke warm at these voltages, but i dont think I'll go over 2.5V. I'd feel guilty destroying these modules and then returning them for RMA.

[GPSeek]

I just did a little bit of testing and the memory runs almost cool to the touch when in S3 suspend for an hour after running orthos for 15 minutes to heat things up first. The motherboard must throttle voltage too when it is in slow refresh mode because Temperature is definitely not elevated in S3 Suspend.

You can tell there is a little heat being generated, but absolutely nothing to worry about. When everything else is cold to the touch, the memory is barely luke warm after being on Suspend for an hour. I would feel safe putting a blanket on top of the memory it runs that cool.

Good to know this. So my fried kid was died of defective material instead of heat.:toast:


Since this memory is going back tomarrow, I'm doing some elevated voltage tests and so far:

Unstable - 417Mhz - 4,4,4,12 @ 2.44V
Stable - 417Mhz - 4,4,4,12 @ 2.48V
Unstable - 435Mhz - 4,5,5,15 @ 2.48V


My fan keeps these modules luke warm at these voltages, but i dont think I'll go over 2.5V. I'd feel guilty destroying these modules and then returning them for RMA.

Newegg RMA for replacement is a slow process.:slapass: You probably end up with 2 week turn-around.:slapass: I first applied RMA for replacement but changed to RMA for refund and ordered another pair at the same time which is about 1 week faster.:)

[SPL15]

Newegg RMA for replacement is a slow process.:slapass: You probably end up with 2 week turn-around.:slapass: I first applied RMA for replacement but changed to RMA for refund and ordered another pair at the same time which is about 1 week faster.:)

Yea I'm kinda SOL on that one. I'm past 30 days and I can't return for refund any longer. Also I already filled out the RMA exchange and all that stuff so these modules are going out tomarrow.

I was thinking about buying something else for now and just selling the Ballistix when they get back... I dont know yet though. I told myself I wasnt gonna do the buy and sell for less kinda thing again like I did with my old P4 system.

I've had orthos running at 2.48v on this memory using 417FSB @ 4,4,4,12 and it was stable for several hours. Moved the timings to 4,4,4,10 and I get an error after several minutes. 450FSb unstable no matter what, but it is better with the higher volts. If DDR850 with 6,5,5,18 timings with 2.48 volts aint stable then this memory is definitely defective.

Thanks for the input everyone.

[ziddey]

Hmm, I can't remember best buy's return policy on ram, but I'd imagine it's same as psu's. You could buy some ram, and hopefully your rma turns around in under 30 days, and you can return for 100% refund. If the rma runs long, you could purchase another set of the same ram, and immediately use the old receipt to return the ram. Then when you can swap out again, return the old ram with the new receipt. :banana::banana::banana::banana: best buy

[GPSeek]

I was thinking about buying something else for now and just selling the Ballistix when they get back... I dont know yet though. I told myself I wasnt gonna do the buy and sell for less kinda thing again like I did with my old P4 system.


I was also thinking about to get something else when I RMA'ed the defective Crucial kit as another XS member did. However, I did not find anyting as competitive as Crucial Ballistix PC8000 at that time. The others were either too expensive or binned too much to have much less oc potential.:slapass:

And I grabbed another pair of Crucial Tracer from a newegg sales days ago. It's very unlikely that 3 or more sticks die together. Your M/C needs ram anyways.

All the 3 Crucial kits (including 1 tracer) I have/had showed very consistent performence. :fact: They all run at 400MHz 3-3-3-8 at 2.32v without any other efforts.:toast:

[SPL15]

Hmm, I can't remember best buy's return policy on ram, but I'd imagine it's same as psu's. You could buy some ram, and hopefully your rma turns around in under 30 days, and you can return for 100% refund. If the rma runs long, you could purchase another set of the same ram, and immediately use the old receipt to return the ram. Then when you can swap out again, return the old ram with the new receipt. :banana::banana::banana::banana: best buy

Worst Buy has 15% restocking fee for anything not defective I believe. I try to stay outta that place unless I'm looking at big screen TV's I plan to purchase elsewhere.

I was also thinking about to get something else when I RMA'ed the defective Crucial kit as another XS member did. However, I did not find anyting as competitive as Crucial Ballistix PC8000 at that time. The others were either too expensive or binned too much to have much less oc potential.

And I grabbed another pair of Crucial Tracer from a newegg sales days ago. It's very unlikely that 3 or more sticks die together. Your M/C needs ram anyways.

All the 3 Crucial kits (including 1 tracer) I have/had showed very consistent performence. They all run at 400MHz 3-3-3-8 at 2.32v without any other efforts.

Yea, i've been reading other peoples experiance with the Ballistix PC8000 and I would like to stay with them if I can get that kinda results or even close.

My system right now is memory limited to at least 3.87Ghz with my CPU @ only 1.4 volts. I first tested this speed with 5,4,4,12 timings at 2.28 volts and it was Orthos stable for about 5 hours and then I gamed for about 3 days straight and then primed for 1 day before I got couple of lockups then you know how it went after that. I think I got a real good CPU on my hands and I'd rather have something that will let me be CPU limited in speed rather than memory limited.

I'll defintely use the ballastix if they get me the results I am lookin for at decent timings. At DDR8xx I definitely want to be doing 4,4,4,12 and no higher.

I just hope that the modules i currently have really were defective to begin with and that it isnt something in my system that caused this problem.

[ziddey]

bby isn't 15% on everything. I got a psu and returned it in 3 weeks with no charge. i believe it's radar detectors and computers being the main 15%

[syne_24]

All the 3 Crucial kits (including 1 tracer) I have/had showed very consistent performence. :fact: They all run at 400MHz 3-3-3-8 at 2.32v without any other efforts.:toast:

The tracer was the deal of century from newegg. I ran memtest and orthos for 8 hours at 400 @ 3-3-3-8 w/ 2.2v and it's rock solid. I would oc more but this board sucks, I'll push it more when the abit gets here. :)

[Hutch]

The Gigabyte S3 C1 has killed 2 of my sticks of ram running stock volts or +.1v and an 80mm fan over them... Both sticks were killed in the same slot (DDRII2 slot) and both times they died while the system was in windows and both times i got a bsod... After I rebooted the system would not boot with the stick in... (Two seperate incidents.)

Whats strange is that your sticks have declined in performance after putting them in a C1 DS3/S3 board... Not only did mine decline in performance the board actually killed two sticks... Sigh

I'm willing to bet it's the board.

[SPL15]

The Gigabyte S3 C1 has killed 2 of my sticks of ram running stock volts or +.1v and an 80mm fan over them... Both sticks were killed in the same slot (DDRII2 slot) and both times they died while the system was in windows and both times i got a bsod... After I rebooted the system would not boot with the stick in... (Two seperate incidents.)

Whats strange is that your sticks have declined in performance after putting them in a C1 DS3/S3 board... Not only did mine decline in performance the board actually killed two sticks... Sigh

I'm willing to bet it's the board.


If your talking about my case, I'm using an Intel XBX2. The S3 I was refering to is the S3 suspend state (suspend to RAM).

My Mobo does act a little funny when changing OC settings and BIOS flashes, but I think it is more a design issue with the XBX2 and not defective... At least I hope...

[ziddey]

No I think he was talking about me. I went from a rev2 c2 ds3 to a rev1 c1 ds3. After the switch, I was no longer able to run my tight timings (400 3-3-3-8 / 500 4-4-4-12). Then, when I switched back to my rev2 c2 ds3, I still didn't have the ability to use those timings (even though I was once able to with ease on the board just a month ago)

[SPL15]

An update on my situation...

Bought another pair of Ballistix DDR1000 and same exact results as my other pair I thought were degrading. It's gotta be my XBX2.

I moved MCH voltage up to 1.7v and it is a bit more stable, but ORTHOS still fails after about 8 minutes with any FSB higher than 415Mhz. This is with 5,5,5,15 timings at 2.2volts. Increasing Memory voltage gets me very little stability boost and dropping Memory speed down to 4:3 ratio (ddr624) makes no difference in stability. Only dropping FSB and increasing bus voltgages gets me stability regardless of memory timings which leads me to believe I either got a failing northdridge or my CPU is going to hell. I doubt it is the CPU because I can run small FFT's at 3.850GHZ for an hour with no errors, but blend gives an error within a second or two.

I put all the modules in the XBX2 for 4GB of ram and I get that exact same results no change in stability for better or worse.

I tried moving the strap to 333 with 667 divider and now the stupid XBX2 wont post. Been trying to get it recovered for 3 hours now and it just lights up and nothing happens, no screen, no beeps, but power up and thats it.

Whenever I do a BIOS update it does the same exact thing. I have to remove the battery, unplug the PSU and let it sit overnight. I've tried removing memory and all the little magic tricks/voodoo people have said to do, but no go. Whatever happened to just removing the stupid jumper for a few seconds and magically restoring defaults.

If any one knows a way to restore factory defaults without removing the battery for half a day please let me know.

I'm about ready to take this stupid motherboard out back and blast it to pieces with my Kimber 45 and 357Sig P226. I'll probably finish off with vaporizing it with a 1/2 stick of TNT. I dont care about RMA, I want to pay this motherboard back for wasting my time and money.

[SPL15]

Another Update.

Finally got my system running again. Been playing around with my system and added a loud powerful 80mm fan directly above the chipset after I reseated the factory heatsink and now it is back stable (for now) at 417Mhz fsb. I'm being conservative right now and just running the ram at 5,5,5,15 @2.2volts to make sure things are stable on the motherboard.

I did down clock the system and use the 1:2 ratio to see how this memory does and I've gotten DDR1012 ORTHOS blend stable for 15 minutes.

So Basically to sum everything up, I didnt troubleshoot as well as I should have and blamed the memory related errors on RAM. I wonder how many other people have this same situation and are blaming the ram. I see many people on these forums with the XBX2 that are limited to 400 or less FSB with the E6600. The consensus seams to be that the Northbridge needs no additional cooling from the results of my thread in the Intel section and from other threads I've read. Maybe this info is wrong and maybe some people just get real good chipsets while others get mediocre.

[NickS]

Mines starting to do it. Won't run 1000MHz @ 4-4-4-12, 2.35v :(

Will do 1000 @ 5-5-5-15, 2.25v tho.