Looking for non-D9 based DDR2

[RJARRRPCGP]

There's no updated list for DDR2 chips based on RAM builder. :(

[Suprchargd]

You want any non-D9 RAM or a specific one? ProMOS or Elpida, etc.?

[RJARRRPCGP]

Looking for either one mentioned, because it appears that Micron dropped the ball with D9, with RAM chip reliability. (RAM that fails at stock! Give me a break!)

[JamesXP]

GSKill DDR2 2x2GB PC8000, doesnt overclock much over 1000mhz but it isn't D9.

[Reefa_Madness]

Here is a cheap alternative with some pretty good potential...Hynix based ram.

Original story that I read on Hynix is here...these are TwinMOS modules rated as PC2-6800 (yes, 6800, not 6400):

http://www.vr-zone.com/articles/TwinMOS_Twister_X_-_2_x_512MB_PC2-6800/4638.html

This thread tests the Super Talent with the Hynix DDR2-667 (not DDR2-800) with pretty good results:

http://www.xtremesystems.org/forums/showthread.php?t=168686

another one with some Super Talent results, but no screenies:

http://forums.anandtech.com/messageview.aspx?catid=83&threadid=2126432&enterthread=y

...and here they are for sale...they have actually dropped about a $1 since the last time I checked the prices:

http://www.ewiz.com/detail.php?name=T667UB1G_H

Hynix DDR2 Product Coding:

http://www.hynix.com/products/consumer/consumer_sub.jsp?menuNo=1&m=2&s=1&menu3=01&RK=03&RAM_NAME=DDR2%20SDRAM&SUB_RAM=512Mb (http://www.hynix.com/products/consumer/consumer_sub.jsp?menuNo=1&m=2&s=1&menu3=01&RK=03&RAM_NAME=DDR2%20SDRAM&SUB_RAM=512Mb)

Do note that the TwinMOS are a 2x512MB kit, while the Super Talent are 2 individual (not kit) 1GB sticks.

[Suprchargd]

I can confirm those Super Talent 667 Hynix sticks do fairly well with overclocking.

[Gillbot]

My ocz platinum isn't d9.

[RJARRRPCGP]

Why does there seem to be less choices for DDR2?

[Moto7451]

D9 happens to be good memory, hence why it's used a lot. I think a lot of companies/users may be pushing the memory too hard with the newer revisions. It could also be something similar to SNDS.

I'm running KVR that does 1000Mhz with little tweaking. Not D9 based. Also I think if you go for OEM Micron chips you'll have better success. I haven't seen any non-enthusiast D9 based products (like my Micron branded D9GMH PC5400 DIMMs that do 1000MHz as well) die.

[ALiEN2953]

the reason D9 is failing a lot is because people but too high voltage through them, and because they are rated to take far higher voltages then they can really handle stably 24/7

[Suprchargd]

If you're pumping over 2.1v to D9 RAM you better have a dedicated RAM fan on top of them, to be safe.

[DumpALump]

I think a lot of the failures are due to inadequate cooling. If you're pumping over 2.1v to D9 RAM you better have a dedicated RAM fan on top of them, to be safe.

So why have 3 of my sets failed when they had a 120mm yate loon blowing on them? What about others who zip tied fans directly to their D9s?

Why are there posts on a lot of forums about D9s dying, but not other chips?

I'm switching over to these hynix or promos chips.

[Suprchargd]

I've read through some pretty big threads, and I'm not disagreeing with you at all that D9 RAM is being falsely rated for voltages that it can't reliably be run at. My post came off in a way I didn't mean it to.

On the other hand, one person I know that is local to me has been running some D9 RAM at upwards of 2.45v for a very long time, and it hasn't given him any problems. Seems to be luck of the draw.

Edited my original post.

[Moto7451]

Why is mostly D9 dying? Because thats what's mostly used and thats what people like tormenting. This thread is nearly a week old and there are only a couple specific examples of alternate memory listed.

I've seen people complaining that their board was limiting their voltage to 2.4v and that this was unacceptable :D. Someone here on the forum (or was it XS) had burn marks from the LEDs on their ram slots (DFI motherboard) from running high voltage through their DIMMs. If you run 2.4+v through Hynix or Promos I'm sure they won't be very happy either.

[Neuromancer]

Moto

alot of intel SNDS had to do wit hthe fact that OEMS were marketting 800MHz 1066MHZ and beyond memory sticks before Micron actually made them.

They were just binning some decent ICs and praying they lasted at Overclokced ratings. Not thinking I guess that overclockers would then say "Oh if its rated 1066 I should be able to hit 1200...

All on a 533 or 667 natural IC.

I dont see any AM2(+) users complaining about dead d9s though.. so I always assumed it was a intel chipset thing. (I have had more dead hardware in the last 6 months since swithcing to c2d then I have in the previous 6 years, which that was all AMD or pre p4 intel)

[Moto7451]

SNDS is when a Northwood P4 dies and that happened during DDR1.... not sure the connection you were trying to make?

My example was that Northwoods started all of a sudden dying when a certain voltage was used. It could be after a week a month or a year, without any warning hence sudden. So long as you kept it under that voltage is A-OK. I think that there might be a SMDS - Sudden Micron Death Syndrome happening over 2.3-2.4v ;) (is that what you meant above?).

[MadMan007]

the reason D9 is failing a lot is because people but too high voltage through them, and because they are rated to take far higher voltages then they can really handle stably 24/7

I wonder about the ratings the makers give them. Even Crucial Ballistix are given a stock voltage of 2.2V. I tried undervolting mine at the same speed they run at 2.2V and they failed memtest quickly with a mere .05V drop but that was after running them at rated voltage for a while. So does that mean Micron itself (aka Crucial at retail) has too high a recommended voltage?

Somewhat related I wonder how much airflow people think is necessary to help cool Ram. Small-ish fans running fast are too loud for me so I usually undervolt them, a 60mm needs about 6V to sound reasonable to me inside a case. It still moves air but not as much of course. I suspect it wouldn't matter too much unless you're pushing crazy voltage or speeds as long as there is some directed airflow, just looking for other opinions.

FWIW I've never had D9 die but I don't push crazy voltage (2.2-2.25V) and I'm also careful to fully power down the mobo by switching off the PSU. I know some stuff was dying on NV boards because of that.

[freeagent]

SNDS is when a Northwood P4 dies and that happened during DDR1.... not sure the connection you were trying to make?

My example was that Northwoods started all of a sudden dying when a certain voltage was used. It could be after a week a month or a year, without any warning hence sudden. So long as you kept it under that voltage is A-OK. I think that there might be a SMDS - Sudden Micron Death Syndrome happening over 2.3-2.4v ;) (is that what you meant above?).


seems so long ago... intel guys were worried about 1.65v or so, while i hammerd 2v into my athlons and giggled :)

[Neuromancer]

SNDS is when a Northwood P4 dies and that happened during DDR1.... not sure the connection you were trying to make?

My example was that Northwoods started all of a sudden dying when a certain voltage was used. It could be after a week a month or a year, without any warning hence sudden. So long as you kept it under that voltage is A-OK. I think that there might be a SMDS - Sudden Micron Death Syndrome happening over 2.3-2.4v ;) (is that what you meant above?).


Hah hah

Yah I thought SNDS was a memory death thing.

I was thinking N was short for NAND or something (some kinda memory chips).

I assumed and went ahead and made one outta myself lol

[Neuromancer]

I wonder about the ratings the makers give them. Even Crucial Ballistix are given a stock voltage of 2.2V.


All microns are rated at 1.5 or 1.8volts it looks like to me. Everything else is done at the OEMs discretion.

My theory is....when you build 2x1GB kits for 15 or 20 dollars, and then bin them (for maybe another 10 bucks) Then charge 125 dollars for the kit. You can afford to replace a few kits so thats why they can lifetime guarentee them.

[hank123]

All microns are rated at 1.5 or 1.8volts it looks like to me. Everything else is done at the OEMs discretion.

My theory is....when you build 2x1GB kits for 15 or 20 dollars, and then bin them (for maybe another 10 bucks) Then charge 125 dollars for the kit. You can afford to replace a few kits so thats why they can lifetime guarentee them.

oooh thats a nice starting point. i might have to go dig and see if i can find anything to back this up.

i lost a stick of d9 at stock voltage. i been thinking about running something else to. i have some none d9 G. skill PC6400 that will OC about 34Mhz and thats it.
so u can get like 2 gigs of that red G. Skill PC6400 for under 50 bucks shipped. and there NOT D9s.

[MadMan007]

All microns are rated at 1.5 or 1.8volts it looks like to me. Everything else is done at the OEMs discretion.

My theory is....when you build 2x1GB kits for 15 or 20 dollars, and then bin them (for maybe another 10 bucks) Then charge 125 dollars for the kit. You can afford to replace a few kits so thats why they can lifetime guarentee them.

Yea I was aware the chips are rated for 1.8V standard, that's the JEDEC spec as well though which all chips are supposed to follow for JEDEC speeds. My question was more why would Crucial Ballistix sticks in particular have stock 2.2V if Micron knows it kills sticks? Your price example makes sense, who knows what the exact numbers are, but it still seems awfully irresponsible for the manufacturer's retail brand to overvolt so much. Why have the stock voltage set to a point you know there will be more RMAs, an RMA replacement is still a loss.

[Evilsizer]

we can come up with a number of things that we can say kills the high speed D9's. you can look at things how you want with amd vs intel relation to D9's dieing. i still think something in the process of manufacturing the IC changed. which even at stock 2.1v or 2.2v is causing them to do die, this is even at ddr2-800 speeds. i have seen a number of them with intel systems and i have only seen one AMD rig with them die. some say its the Dram drive strenght is the problem yet we dont know. NV a while back when people were blaming the 680i for killing the ram. that anything over 2.2v without active cooling kills the ram. yet i had active cooling on my PC2-6400-HZ 2gig kits(D9GMH) based die after about 3months with only 2.25v and they had a 80mm fan on top. i have since switched to HK's which are promos based same fan and havent died yet. while i did push the D9 to ddr2-1118 speed, the promos stuff hasnt seen that kinda speed YET. since i dont game like i use to, i dont plan on pushing the ram unless its a bench session with a new cpu. none the less the PK's are running near stock timing at DDR2-960 with stock voltage, since my board doesnt like the tight timings they can do(4-4-4-10 instead of 4-4-3-5).

like i said we can guess all we want on the reason why ram is dieing... it could even be a ploy by intel to get us to buy into DDR3 for all we know....

[hank123]

like i said we can guess all we want on the reason why ram is dieing... it could even be a ploy by intel to get us to buy into DDR3 for all we know....

hahahahaha thats the funnest thing i have heard all day.

I think Super Talent or some cheap value ram would be the way to go.

[Moto7451]

we can come up with a number of things that we can say kills the high speed D9's. you can look at things how you want with amd vs intel relation to D9's dieing. i still think something in the process of manufacturing the IC changed.....

Well "D9" afterall is just their product family. It's like saying P4 (going back to SNDS). D9GMH and D9GKX are the same die but different speed bins. D9DQK & DCD (two other types I have) are different memory chips entirely. They have different packages (pin count/size) and different ICs (width is different). Fat bodies are different this way also. They may all use the same basic design of the memory cell, but each revision probably should be treated as a different beast. If they've changed the IC, i.e. a die shrink (quite likely in order to boost margins), then the problem isn't the change itself, but perhaps that the new die shrunk memory reacts differently/doesn't like the same level of voltage. This would be annalogous to AH-5/6-BH5/6-CH5/6-UTT deal. Same basic design with different binning and processes that netted various results from wonderful to downright awful (though the latter most category was mostly reserved for AH series chips ;)).

Edit: Pretty sure the 1.5v stuff is for use on embedded stuff, graphics cards, and possibly notebooks.