Who's Killing Your RAM?

[One4yu2c]

Getting ready to build that kickass Core 2 rig you’ve been lusting over? If so, chances are you’ve been eyeballing nVidia’s 680i chipset as a foundation, and with good reason. NVidia knows a thing or two about chipsets, just ask last generation’s legion of socket 939 adopters how they felt about their nForce4 X16 SLI based setups. Even I’m admittedly reluctant to fully retire my now aged MSI K8N Neo4 Platinum / 4400+ X2 rig. But alas, it’s a dead socket, and I too have considered pairing my sexy Core 2 processor with a 680i based motherboard capable of scorching through the benchmarks. But is that all it’s scorching through? - Paul "One4yu2c" Lilly

Read More...

[RangerXLT8]

That is like a grand worth of memory, in a FRYING PAN!

[Panda Man]

Originally Posted by RangerXLT8

That is like a grand worth of memory, in a FRYING PAN!

I'd like him to put vegetable oil in there and turn the stove on

Nice write-up paul! Since the only complaints I've really heard about are memory failures on the 680i chipset and not others, I think I will stay away for the time being, unless nVidia is correct (and another source verifies it) that it is indeed the manufacturer's fault.

[simple_inhibition]

nice article. definitely something to keep my eye on for when i make the leap to DDR2.

i know what killed my OCZ VX..... it involved 3.2v, alcohol and a rubber chicken in an overclocking orgy one night...

[gvblake22]

I went through the registration process at the JEDEC website and dug around in their mass of PDF documents and finally found what I was looking for. The document is the "DDR2 SDRAM Specification dated May 2006 (JESD79-2C) for those who want to hunt it down. Page 56 lists the "Absolute Maximum DC Ratings" of DDR2 memory, and all voltages are listed up to 2.3V. It also notes under the table that "Stresses greater than those listed under “Absolute Maximum Ratings” may cause permanent damage to the device. This is a stress rating only and functional operation of the device at these or any other conditions above those indicated in the operational sections of this specification is not implied. Exposure to absolute maximum rating conditions for extended periods may affect reliability."

I would post a screen shot of it, but I think that would violate their content copywright permissions, so you'll have to look it up for yourself if you want more information.

So in short, Paul is right, it's not solely the chipset maker's fault. Memory makers are pushing this stuff to the limits, and users need to take initiative and figure out what is safe and what is not. Just like in a court of law, you can't plead ignorance when you do something wrong...

[Capper]

Originally Posted by Capper

I honestly don't think the 6XX chipset is any more popular than any other.....some people got excited by Linkboost, some by EPP memory......the truth of the matter is its been a fairly problematic chipset thus far......not that it won't get better as it matures....you had the RAID issues, memory issues, LinkBoost issues, and BIOS issues.

As to the memory, I wasn't trying to pass blame to memory makers.....I was just trying to say that its irresponsible of us as people who know and who are supposed to help others to allow posts where people talk of "reports" or who spread rumors (not that anyone was).....In the end, I think it will prove to be a combination of voltage fluctuations with the board, along with memory just pushed to its limit.

680i chipset burning up RAM? Nvidia responds!

[One4yu2c]

Originally Posted by gvblake22

Memory makers are pushing this stuff to the limits, and users need to take initiative and figure out what is safe and what is not.

A quick and easy to administer test is to simply touch the RAM's heatspreader when stressing the memory. Some of these kits get surprisingly hot, and has given me pause to consider implementing an active cooling solution.

[wtcnbrwndo4u]

Good blog post Paul.

Very interesting; I wasn't aware of the issue at hand. If you refuse to change RAM, it seems the only probable route is water cooling, which isn't the most cost-effective plan (unless you ghetto-rig a fan setup, which still wouldn't be as effective).

Is this issue recorded on the lower models of the 6XX chipset?

[One4yu2c]

Originally Posted by wtcnbrwndo4u

Is this issue recorded on the lower models of the 6XX chipset?

That's the thing, it's not really "recorded" anywhere, but there has been enough stink brought up in enthusiast communities to warrant an official response/stance/comment from nVidia.

To answer your question though, I haven't seen users complain of similar circumstances on the 650i chipset. This could mean that (A) I haven't looked enough (very possible) or (B) users with 650i chipsets aren't pushing and/or overvolting their RAM (or using the same high rated kits) to the same extent as 680i owners.

[Capper]

A couple of things worth pointing out is that most of the 650i boards have limited BIOS options, not nearly as thorough as the 680i BIOS. Also consider that many enthusiasts aren't going to invest in an enthusiast board , they are usually going to grab the big boy. Also, most of the 650i BIOS's I've seen are still a little rough around the edges, even the ASUS versions are a little problematic.