How's 3.6 for Vdimm?

[dobbz]

im just wondering if anyone has run their memory at 3.6v 24/7. ive been at 3.5v with mine for two days and cant pass 250 tight, or 255 CL2.5, so i was gonna try 3.6v. but i want a voltage that i can run all the time, and hopefully keep it goin about a year, so is 3.6v too much for that?

[pkrew]

I've run a little over 3.5 for about 2 months, no problems. I tried 3.6 and it really didn't help much. I think it gave me 1 more fsb so I back it down to 3.5.

[dobbz]

oh whoops, i forgot to say: 3.6 gets me 12+ hrs stable at 255 tight. i got a lil worried after that long though (heh prolly no reason to be).

[TheWeaseL]

that sounds a bit high for 24/7 use...for benching no sweat, keeping it up there for upwards of 72 hours, but that seems a bit much for every day...

[enzoR]

winbond themselves allow upto 3.6v. should be OK if your cool em a bit, and even higher....

[FUGGER]

I run 3.37v 24/7 for a few months now on this machine. Hasnt been a problem. It can be a problem if you have a lot of heat built up inside your case due to poor ventalation.

[Endre]

It may be allright, but I wouldn't risk such nice sticks for a few extra MHz.

[The Stilt]

BH-5 and CH-5 maximum voltage is rated to 3.6V, so they should be capable to take it, if chip temperature stays under 72c.

[dobbz]

wow, you're right. Windbond even says the recommended BH-5 voltage is 3.3v.

ill see what 3.65v gets me.

[OPPAINTER]

I heard Kingmax says you can run they're ram at 3.4V, thats from the manufacturor:)

OPP

[ricjax99]

Originally posted by dobbz
wow, you're right. Windbond even says the recommended BH-5 voltage is 3.3v.

Where did you get that from? :confused:

[a5h]

does the PCB not come into play ...i heard something that KHX BH-5 PCB can handel more volts than say ...Corsair 3500 PCB ...(same winbond BH-5 chips, keep in mind) dunno whether this is true tho.

[Zeus]

Originally posted by The Stilt
BH-5 and CH-5 maximum voltage is rated to 3.6V, so they should be capable to take it, if chip temperature stays under 72c.

Wow, that's some good news, i was always scared to go over 3.3V.
;)

[megahurtz-oc]

anyone got any info on moddin a ps to get 3.5v from the 3.3 rail?

[sjohnson]

Another vote for 3.3v, been that way for over 6 months on my year and a half old TwinMOS BH-5, 3.2v before that (highest my unmodded kd7-e would go).

[oc-rookie]

Originally posted by megahurtz-oc
anyone got any info on moddin a ps to get 3.5v from the 3.3 rail?

In the hardware mods section. http://www.xtremesystems.org/showthread.php?s=&threadid=28023 Worked for me!!!

[niaboc79]

3.75v 24/7 with corsair BH5 for 290 2 2 2 5 1/1

[solo]

Originally posted by niaboc79
3.75v 24/7 with corsair BH5 for 290 2 2 2 5 1/1

do you have any active cooling on those :eek:

[Formann]

3.5vdimm on Twinmos /w BH5 for 277 2.5-2-2-5 stable. The sticks dont have any cooling as I have Vapo on cpu, but they dont get varm. Not even with 3.7 they get warm. Using the "finger on chips" method i guess the temp is below 40c.

Im not at all concerned about the sticks when running 3.5+ vdimm.. its the CPU im worried about.

[dobbz]

Originally posted by niaboc79
3.75v 24/7 with corsair BH5 for 290 2 2 2 5 1/1

and how long has it been?

i cant past 255 with 3.6v. the mosfets get HOT as hell, like literally burned my finger, but my OCZ copper ramsinks came the other day and theyre barely warm with a 40mm laying on em. only got me 5MHz though. ill try 3.7.

[R.Rabbit]

5mhz huh? not bad, maybe i'll chop up my 1u sink to make some pretty little mosfets sinks

[dobbz]

3.7v turns out to be unstable for me. i can barely load Windows past 3.65.

and for some reason my mem bandwidth has gone down 100Mbps. i hope i didnt damage the RAM.

[Zeus]

Originally posted by niaboc79
3.75v 24/7 with corsair BH5 for 290 2 2 2 5 1/1


Man! :eek: Can you show a Sanrdra bandwidthtest screeny @290MHz 1:1? :slobber:

[macci]

3.75v 24/7 with corsair BH5 for 290 2 2 2 5 1/1
What CPU multiplier?

[bachus_anonym]

i'm not that familiar with A64 setups, but i belive this is HyperTransport thing...
i may be wrong but 290 11-2-2 cas2 1:1 on A64 doesn't exactly translate to 290 11-2-2 cas2 1:1 on NF7-S...

correct me if i'm wrong... BTW... just take a look at ORB links...

[Liquid3D]

all current high-end DDR chips are using internal voltage regulators to reduce the core voltage to 1.8V which is necessary to run at high frequency. The only parts of the chips that actually see the increased voltage are input/ output buffers and, granted that those will wiggle a bit faster at higher voltages, the net effect is counteracted by the fact that the internal voltage regulators produce a lot of heat that will overall slow down the chips. In other words, a sensitive approach would be to leave VDD alone but increase VDDQ. Admittlely, there are some older DIMMs that will run at 2.5V internally but those will hardly be used in PC2700 mode of operation which is the only scenario where the massive overvoltage would be of benefit.
http://www.lostcircuits.com/motherboard/asus_a7v333/5.shtml

I would put my resources into water-cooling the NB, or if it's an Athlon64 the CPU. Once you get some extreme cooling on the CPU, your going to experience greater bandwidth anyway. Since the MCH is now on-die in the A64 your actually at an advantage in many respects to those of us trying to cool our NB-MCH's. Yes the memory will run at 3.7V and so will any car engine run off Nitrous for a year, but is that the way you want to ride? Fish-tailing from stop light to stoplight, slamming on the breaks?

The anaology holds true because the added heat, and stress is bound to create many erros. Every RAS and CAS miss is an error which mus then be corrected, or worse left un-corrected.