Darn Instability

[Player0]

Ive introduced some sort of instability in to my system and I cant seem to pinpoint it. One of my thoughts might be using DIMM slots 1 and 4. Can anyone (mdzcpa?) confirm if using 256m modules in 1 and 4 causes stability problems?

One of the problems I get at 200fsb or higher are my windows icons getting all goofy now. If i use the Rebuild Icons utility in TweakUI, it fixes them, until I restart the PC, and the problems come back.

Just have to find the sweet spot again, some bios tweaking should fix this. I cant find any errors with the ram or cpu. Perhaps the video card took some damage when the water leaked on to it...but it runs 3dmark fine image wise, no defects or slowdowns. It will randomly just freeze (screen goes dead/blank or just freezes solid). Prime will usually not detect any problems, everything just freezes solid.

CPU is clocked low (1600), focusing on FSB. Not sure what it is. Id like somoene else to confirm if using DIMM 1 and 4 causes instability that needs to be addressed with different BIOS settings.

[mdzcpa]

I'm at work now...I'll try DIMMS 1 and 4 tonight when I get home.:)

[Player0]

Thanks man. Not sure theres any rush now. I cant get my KX7 to bleeding boot again. I was in BIOS last night, made some changes, and I went to save, pressed F10, pressed enter, and the PC hard froze right at that instant...the Y prompt was blinking, but nothing else.

I turned off the PC, and it no longer boots. *$#@#!!!!

Maybe I just need to order a new bios? I dont know.

[lechumbl]

Hi Tom,

Had the best luck in slots 1 and 3 or 2 and 4, never 1 and 4.

Worth a shot?

Take care.....

[joek71]

I think you need a new BIOS chip, call Abit they have no problems sending a nes BIOS chip, when you flash your BIOS dont press F10 or reset, the best way is just unplug your psu from the computer, and then reset the bios., I tried it with F10 and my BIOS also crapped out so the best way is to just unplug if from the psu. If you order the BIOS chip it takes around 3-4 days to get it. good luck.

[Player0]

Lee,

Im stuck using 1 and 4. I would have prefered to use 1 and 3, but the guys where i bought the ram waterblocks from lied to me.

[mdzcpa]

Well, the test results are in!

Unfortunately my testing indicates that using DIMMs 1 and 4 result in a definite negative increase in RAM stability at real high FSB speeds.

With the DIMMs in 1 and 4 I could not get the machine to run stable above 200mhz. Even using SPD settings. It boots fine, but 3DMark2001SE crashes to desktop after a loop or two. (as compared to my DIMMs 1 and 3 set up which can loop 3DMark2001SE at 205 FSB. Highest stable at SPD is 198.

In addition, the highest FSB I could get with full aggressive timings (CAS 2, 4 way, 2,5,2, 1T ) is only 182mhz, compared to 1 and 3 resulting in 187mhz.

And my day to day settings of CAS 2, 4 way, 3,6,3 2T which is good for a 3DMark2001 looping stable FSB of 202 with DIMMs 1 and 3, now only provides stable speeds up to about 188mhz with 1 and 4. Big difference here.

Bottom line....DIMMs 1 and 4 suck for me using Corsair PC3000. I'm not sure if this is the case for everyone, but it sure makes a difference on my rig.

BTW, also tested using DIMMs 2 and 4 and found them to be just as stable 1 and 3.


Although this doesn't sound good. I hope it helps you. Oh, and BTW, both Memtest86 and Docmem pass with the sticks in any combination of DIMMs. it only seems to be the high overclocking which is effected.

[Player0]

Hi mike, thanks for the info. Your right, MemTest is much more sensetive...no more waiting for Prime95 to run for 2 hours before coming up with a problem, I can see immediately with this program.

MemTest shows immediate problems above 300meg (the RAM in DIMM 4 of course). My Corsair fairs much better in running than yours, maybe because of the extra voltage and cooling.

Setting DCLK to auto certainly helps. Im testing right now at 200mhz, 3-6-3, CL2, 2T and its perfectly stable. 2-5-3 at 200mhz is simply not, again...errors on the RAM in DIMM 4.

Those bastards...I want to hurt the guy who told me 4 ram water blocks would fit in slots 1 and 3.

Well, Im still playing with BIOS timings to see if I cant get DIMM4 more stable.

I may go back to air cooled ram, or watercool my single 512 PC3200 dimm.

[Farabomb]

I love to see how high a 512 stick can get. I've been reading around and they tend not to do so well. Maybe they don't try hard enough :D

[mdzcpa]

P0:

Maybe your voltages is helping...that maybe something I look into later. Based on your advice, i still haven't dome the Vmem mod...maybe later though :)

Perhaps it is possible that if a messed around with some of the other memory settings that I might get the 1 and 4 configuration to work better. I certainly did not take the time to tweak it...I bascially slapped the sticks in 1 and 4 and let things fly with the predetermined settings that worked well with 1 and 3 config.

And a big jolly "boooo" :mad: to those guys who gave you the bad info....shame shame.

Anyway keep us posted on how it goes.:)

[Player0]

Aiight. I couldnt get DIMM4 to work stable at anything above 3-5-3 settings. 2-5-3 gives errors on the DIMM4. I cranked everything way up...no go.

Im testing the standard 512m DIMM right now in DIMM2. It was shakey at first...as a matter of fact, it takes a FULL 3.2vMem and all the DIMM clock settings set to HIGH to make this dimm stable at 200mhz, 2-5-3 2t CL2 4-way. But...it does run stable here. Passing all my tests.

I might even have more stability in DIMM1, but i cant test that yet cause the old ram is in the way. Using such a high Vmem voltage is definately a water-cooled application if ive ever seen one.

So, it looks lik eim going to water cool the single 512 and use that. No, its not as stable as the two 256m, since your able to spread that over the two memory controllers (dimm 1 and 3). But, with enough voltage and signal, you can overcome this limitation, use a single ram waterblock (for more water flow with my 1/4" id system.

So this looks like it will work...woo! :)

Anyone want the other pair of ram blocks? hehe :)

[Player0]

Okay, I take that back. I cant get this 512m DIMM stable at 200mh 2-5-3 C2 T2 either. Its hella closer than the two 256m dimms in 1 and 4...but still gives me a few bugs an hour or two in to testing with MemTester.

I wish I had tested the original 256 PC3000 dimms with this program...i wonder if they were unstable at 2-5-3 either...and i wonder if this minor instability would affect my use?

Well...now that i know its not stable, I have a decision to make. I wont leave it even the least bit instable. Watercooling the PC512 might give it the boost in stability i need to make this rock solid. The other choice is to go back to air cooled 2x256 in slots 1 and 3. But, as i said before, im not sure if this was anymore stable than the current pc512 cause i didnt test it with MemTester. Based on what ive heard, it should be more stable.

The other solutions arent as nice. One is to use 3 ram waterblocks and make them fit in slots 1 and 3. This leaves one of the ram dimms air cooled. But, its better than pure air cooling. Or i could just use 3-5-3 timings, which is perfectly stable in any solution. But...thats not going to happen.