PDA

View Full Version : Need some help, stuck with 754 DFI and BH-5


twilius_basic
06-02-2005, 12:18 PM
Sorry for the long read :P

I want 275mhz 3D-benchable, but I just can't get it :mad: I've got 2x256 HyperX BH-5 sticks and 1x256 XMS BH-6 to play with (well, also 2x256 TwinMOS BH-5 but it isn't any good). However I cannot get above 270mhz stably. High 260+ range is all that will 3D stably. Sometimes 270 works, sometimes it doesn't.... I've been trying 272 but it keeps kicking to desktop in Lobby High. I've tried all sorts of A64 Tweaker timings, loostening everything up still won't get it to work. CPU is at 8x to rule it out of the equation, LDT is at Auto (but 3x or 4x doesn't make a diff), chipset is at stock 1.6v ATM (tried higher but again, nothing). Currently I'm testing a friends 754 DFI board, running the latest BIOS with 4v mod. With 3.3 rail @ 3.89-3.9v, and vdimm set to 3.7v in BIOS, I get 3.71-3.72v idle, 3.65-3.66v under load :rolleyes: Pretty bad, I know.

But I've also tried this on my own 754 DFI mobo when I had the 3.3-to-wire mod done, and with ~3.68v under load I still couldn't get any higher. I wanted to try my friend's mobo b/c I thought just maybe my old one was stuck at <270mhz.

But my problem remains :( If anyone has any ideas, please post them. A64 Tweaker settings, or whatnot. My PSU is OCZ Powerstream 520W. Like I said, I've tried looser timings > doesn't help, still kicks to desktop. On this mobo I don't think I can get anymore voltage.... PSU is pretty much maxxed at 3.9 on the 3.3.

I can re-vmod my own mobo with the wires (which makes vdimm always ~.05v less than 3.3 rail under load), but I don't think that's going to change anything. I don't think I can pump much more voltage through my dimms (last time I tried 3.7v+, my HDD freaked out).

But I really need 274-275mhz 3D-stable, so I can attempt to use that with a 12x multi next time I play with DI/LN2. So any ideas are welcome to help me squeeze those last ~6mhz.

Thanks guys!

-W

felinusz
06-05-2005, 03:35 PM
Have you burned your sticks in? I'm assuming that you have, but a 24-hour memtes86 period of ~260 MHz at ~3.6V might get your sticks to take voltage a little better, it's worth a shot. 3.7V is also a little "low" for 275 MHz... For 275 Mhz, both of my 2x256 meg sets of BH-5 need 3.7V in my board's BIOS (I have a DFI nForce4), which is ~3.78V in actuality measured with a multimeter.

If you know someone with a DDR Booster you can borrow, it might be worth trying out - both for more voltage, and a clean VDIMM with no droop or fluctuation.

In my past experience with the VDIMM droop/fluctuation issue (various VDIMM voltmods on the nForce2 platform), I have found that VDIMM voltage droop and fluctuation seriously thrashes stable results. At a high ("high" being ~3.5V+) VDIMM overvoltage, my personal philosophy and experience is that a stable and "clean" voltage becomes even more important than it is at lower overvoltages.

So I'd start with that, getting your VDIMM stable under load with a booster, or maybe even a VI/O=VDIMM mod, at which point you can directly control VDIMM by manipulating your PSU's +3.3VDC rail.

I don't use an nForce3 board, so my timings are useless to you, although I'll post them if you want.

Good luck.

13oost
06-05-2005, 04:15 PM
there is a bios to flash on the DFI to allow for up to 4vdimm
http://www.xtremesystems.org/forums/showthread.php?t=52198

on the DFI Lanparty UT NF3 you need to mod a OCZ booster for it to work with 2 sticks
http://www.xtremesystems.org/forums/showthread.php?t=48617&highlight=CHAINTECH+VNF3-250+ocz+booster

Crazykooter
06-05-2005, 04:32 PM
Sounds like your gonna need to break the 4v+ area to get 270's stable bud. Suicide ram run for ya. :stick:

felinusz
06-05-2005, 06:10 PM
Almost all BH-5 does not see gains past ~3.8V without extremely low temperatures. Above 3.6V, gains are typically limited to ~3 to ~5 MHz per 0.1V overvolt.

A lot of BH-5 even loses stability above ~3.7V, some chips need a lengthy burn-in before they'll take voltage, others never take voltage well.

4V+ isn't going to do anything but mess up your sticks. Heck, 3.8V might not do anything.

G H Z
06-05-2005, 11:54 PM
I agree with Felin on going over 3.8v, it always caused problems for me even on well burned in sets. I think that you will find a few more mhz under xtreme cooling however. My 2x512 topped @ 267 air cooled but hit 272 3D bench stable under -48°

You might ask OPB about this though, he has run very close to 4v on BH-5 succesfully.