Micron D9 and high voltage, electromigration effects

[zsamz_]

saayaa can you tell me quick.

when i want to do suicide runs, like 2.65 V to achieve 700 mhz for FSB testing or something.
how do memory die?
do they instantly die due to high voltage just cause they dont like it
or do they die of the Heat made by 2.6v

as long as the board has nice vreg you should be good
i've lost many d9 2.24v but only on gigabyte boards

[Vivi]

thanks. zsamz, ive only used asus boards and as far as i know they are the best when it comes to ram power feed. might be why my ram is stil fine .
i can see in your sig you dont like gigabyte
how do the DFI's treat the memory?

[zsamz_]

thanks. zsamz, ive only used asus boards and as far as i know they are the best when it comes to ram power feed. might be why my ram is stil fine .
i can see in your sig you dont like gigabyte
how do the DFI's treat the memory?

what does 7 months of no dead d9 tell ya 1200+ 24/7 not even degraded
asus x38-48 vrm much better than asus p35 vrm

1 think for sure gigabut vrm suks lol no d9 lasted more than 8 weeks on them damn boards i had

[Vivi]

my 24/7 settings are 800mhz 4-4-4-10 1.9v lol.

i only bench at extreme timings. having ram fast 24/7 is pointless in my opinion

what is the MAX mhz your ram has achieved?

[zsamz_]

my 24/7 settings are 800mhz 4-4-4-10 1.9v lol.

i only bench at extreme timings. having ram fast 24/7 is pointless in my opinion

what is the MAX mhz your ram has achieved?

only did 1300 area @ 2.24v this set hates cas 4 i havent used above 2.24v on these sets
but my hyperx does 4-4-4-12 @ 1140 2.3v
pointless if you think i'll trade ya ocz plat for the ballistix they do 4-4-4-12 @ 900
when i'm in power save mode i put in the hyperx n run 3-3-3-8@825

[Calmatory]

I didn't read the thread really, but what I have understood, this applies to quite much all IC's, including CPUs, GPUs, Chipsets, VRAM, integrated chips like SATA/IDE/LAN/RAID/BIOS/Sound controllers and chips etc.

Obviously the clock frequency has great influence on the current, proportionally relative to the frequency. i.e. as the frequency doubles, so does the current, am I correct? Voltage then increases both voltage and current, obviously. This then generates alot more heat than just upping the MHz, and heat obviously accelerates the electromigration effect, right? Please enlighten me. :p

[zsamz_]

I didn't read the thread really, but what I have understood, this applies to quite much all IC's, including CPUs, GPUs, Chipsets, VRAM, integrated chips like SATA/IDE/LAN/RAID/BIOS/Sound controllers and chips etc.

Obviously the clock frequency has great influence on the current, proportionally relative to the frequency. i.e. as the frequency doubles, so does the current, am I correct? Voltage then increases both voltage and current, obviously. This then generates alot more heat than just upping the MHz, and heat obviously accelerates the electromigration effect, right? Please enlighten me. :p

the cleaner the current the safer it is
vdmm voltage regurators play big role
keepin heat down important too

[Calmatory]

the cleaner the current the safer it is
vdmm voltage regurators play big role
keepin heat down important too

Obviously. But what I am after is the influence of electromigration in other ICs aswell, not jsut DRAM chips.

Are DRAM chips more vulnerable to damage caused by electromigration? If the required current is defined by the frequency, wouldn't these high clockspeeds(+4GHz 24/7 use) and thus high currents flowing trough CPUs kill them very quickly? Or are CPUs just much more resistant/durable by default currents and temperatures? Could have alot to do with manufacturing process and methods (high-k & metal gates etc.) used.

[zsamz_]

Obviously. But what I am after is the influence of electromigration in other ICs aswell, not jsut DRAM chips.

Are DRAM chips more vulnerable to damage caused by electromigration? If the required current is defined by the frequency, wouldn't these high clockspeeds(+4GHz 24/7 use) and thus high currents flowing trough CPUs kill them very quickly? Or are CPUs just much more resistant/durable by default currents and temperatures? Could have alot to do with manufacturing process and methods (high-k & metal gates etc.) used.

well i can tell you my gigabut p35 killed promos chips too not just 6 sets of d9@2.24v
i think spiking current on some boards just killin the ram

[kathman]

How good is the P5E with ram??

[Revv23]

Obviously. But what I am after is the influence of electromigration in other ICs aswell, not jsut DRAM chips.

Are DRAM chips more vulnerable to damage caused by electromigration? If the required current is defined by the frequency, wouldn't these high clockspeeds(+4GHz 24/7 use) and thus high currents flowing trough CPUs kill them very quickly? Or are CPUs just much more resistant/durable by default currents and temperatures? Could have alot to do with manufacturing process and methods (high-k & metal gates etc.) used.

you are correct it applies to all IC's ever. Even chips run at stock speeds have a limited life. adding heat, mhz, and voltage simply speed up this process. There are sweet spots that won't shorten the life of an IC too much, which is what most people run thier 24/7 overclocks at.

[zsamz_]

you are correct it applies to all IC's ever. Even chips run at stock speeds have a limited life. adding heat, mhz, and voltage simply speed up this process. There are sweet spots that won't shorten the life of an IC too much, which is what most people run thier 24/7 overclocks at.

yup exactly best findin the sweetspot
i run 1200+@2.0v if i run cas4 1000mhz i need 2.22v
every stick is different some sticks prefer 1000mhz cas4 cause they need 2.3v to go 1200+

[mad_skills]

In my experience D9GKX tend to do CL4 overclock really well while D9GMH are better for CL5, meaning they need less voltage for CL5 and more for CL4 than D9GKX.

[Alex-Ro]

Well it depends,mine D9GMH will do 1000 cl4 with 2.08v and 1200 c5 with 2.1.

[mad_skills]

I am aware of that, i just mentioned that D9GKX tend to o/c better with CL4, i've even seen some D9GKX kits that do 500MHz 4-4-4-4 on 2.0V or even some tenth of a volt lower.

[zsamz_]

I am aware of that, i just mentioned that D9GKX tend to o/c better with CL4, i've even seen some D9GKX kits that do 500MHz 4-4-4-4 on 2.0V or even some tenth of a volt lower.

my d9gkx wont boot under 2.1 they really like the juice

[mad_skills]

I guess you are referring to the HyperX kit?
Yes, they do need some voltage and are known for that, some ppl here have tested it and it's the case with a lot of those sticks.
However, Cell-Shock, Team Xtreem and Ballistix/B.Tracer that have D9GKX ic have less vdimm need with CL4 overclock.

[Xtreme Addict]

my d9gkx wont boot under 2.1 they really like the juice

Because they are DEAD/BROKEN

[VIctorj]

Another kit died...OCZ SLI 1066.

[zsamz_]

Because they are DEAD/BROKEN

they still do 1200++

[mad_skills]

Another kit died...OCZ SLI 1066.

Wow, dead kits just keep coming up last couple of days.

[Xtreme Addict]

they still do 1200++

There are three symptoms of dead D9:

1) boots with higher voltage than 1.8v and overclocks the same as earlier
2) OC is lowering drastically (degrading) but still boots with 1.8v
3) they are unstable - errors in memtest etc...

*Rarely but from time to time memory with D9 after some coincidence (high voltage & overclocking/after psu broken etc.) doesn't boot at all

Symptom 1 & 2 often appear together but not always. Symptom 1 & 3 are the most frequent. Memory with symptom 1 often may work for years with slightly more degradation effect or even no. Memory with symptom number 2 is going to die definitely. Memory with symptom 3 aren't good for working 24/7 and memory qualifies only to RMA

That's from my experience

[mad_skills]

I really hate when kits degrade, that's the worst possible thig. When they're dead, that's it. This is far worse.

[zsamz_]

I really hate when kits degrade, that's the worst possible thig. When they're dead, that's it. This is far worse.

i use to get mad when i lost a d9
so i bought more n more
after i dumped my gigabut board i got no more dead d9 n 10+ sets just sittin waitin for a bro to die but
not 1 of my sticks have been degraded since i went dfi
gave the gigabut board to my son n he got a dead promo stick
its ok he got my redline now

[Kasparz]

There are three symptoms of dead D9

If it won't do in this case 800MHz CL5 for D9GKX with 1.8Vdimm, then its dead in every way.