So I got the Corsair TwinX1024 PC3200 C2 Pro (black LED heatsink) memory, and at any timings I try setting it to, I get BSOD's in CSS and 3DMark01 won't get past Dragothic HQ test.. It would appear that i'm undervolted at 2.68V (2.55-2.68 as high as this motherboard will go) I even tried switching it to DIMM's 2 & 4 to no avail.. I'm about to reset the CMOS, i'm curious what DDR IO Input Termination atually controls but have been trying to research it for some time now (despite this, all i've gotten is that is has something to do with impediance and termination strength).. It defaults to 170 Ohms, but has options for 300 Ohms, 400 Ohms, and off which i've found decreases bandwidth about 300MB/s.. Anyone have any ideas on if its worth a shot upping this? (I noticed on my old bad ram, changing to 300 caused BSODs, though i'm unsure if it was completely because the ram couldn't handle it..)

Can't answer any of those questions. My Corsair PC3200 XMSXpert with LED runs at the advertised speeds at 2.7 volts. You should be right at the limit's. Maybe you can slow it down just a hair. If not maybe someone will know your volt mod info. Good Luck.

It defaults to 3-4-4-8 (except 1t instead of 2t) like the ram I had in there before, what would you consider a set back from 2-3-3-6? (I've tried a few things with mixed results, none of it ever quite adding up to full stability)..

So im bummed a bit, I've find out now that the issue here is the CAS latency.. Seems I have to run 3-3-3-6 1t where it should be C2 on a decent chipset.. So I guess I might aswell run it at 3-3-3-8 1t?

Update 12/27/05:

3-3-3-8 1t is stable, tried OC to 208FSB and it beacme unstable again.. (seeing as thers hardly ANY performance impact on P4's between CAS2 and CAS3; the important one to keep low being TRCd/ contrary to Tras size, which is better the bigger) 3-3-3-9 1t may actually be faster.. 2.5-3-3-7 I find sketchy as it doesn't really benefit P4 in any way (an overclocking setting too which this ram sucks at)..

After trying a few things, 2.5-3-3-7 is actually a bit stabler than 3-3-3-8 (unfortunately can't OC realistically at all) but its a tad faster (disregarding Sandra) and almost equal performance to 3-3-3-8 @ 208FSB, completes all 3DMarks, so I think i'ma just leave it there..

ED: I think were pretty much resolved, case adjourned until I complete rebuild a secondary gaming system in the future out of these parts (errr, atleast the videocard and ram), 2.5-3-3-7 is excellent ( 110.2ns vs 114.8ns @ 3-3-3-8) and scores very solidly too (infact highest avg out of all timings I've tested so far) without any changes to driver or settings/ no OC's on anything 3DMark01: 17128 3DMark03: 12250 3DMark05: 5501

It sounds like you need a better motherboard. You got a 2x512MB kit, correct? If this is the case, I wonder what the chances are that you got BH5/CH5 memory. That kind really wants voltage to do anything.

Being XMS3032 C2 Pro ver4.1, it's unfortunately most definately Samsung TCCD or later TCC derivative with equivalent voltage requirements.. BH was what I really shouldv'e gotten as it was built for good 2.6v operation to begin with..

So, it should be perfectly stable at 2.5-3-3-7 1t / sync FSB @ stock after having verified memtest86+ overnight/ Prime95, F@H, & rthdribl (dx9 HDR test) all day with no errors at all. Sure enough though, Counter-Strike Source freezes. Only that its always right in the middle of loading a level that it happens. -Really irritating, pretty much ruling out the PSU, I think, -as it still happens at 3-4-4-8 1t (chipset-detected SPD timings). I guess it's an application thing, or is my chipset truely that bad?

Similar exp with a similar product....

When I built my box I figured I'd put HyperX 4000 in it (533) but when I did I set it "factory" in bios and my system became unstable. Contacted Kingston and had it RMA'd and all and finally got ahold of thier tech support (oddly I got one who knew what he was talking about) and he told me the stuff wouldn't run stable at factory because they deliberately made it to only run stable at OC voltages. Yes, you heard right a manufacturer told me to jack it through the roof, so I pumped up the voltage above the factory (2.88 as I recall) and it has ran fine ever since.


So, would suggest you pump it till it smokes then RMA and next time dont go that high.

Quote: Originally Posted by konai_konai666 So, it should be perfectly stable at 2.5-3-3-7 1t / sync FSB @ stock after having verified memtest86+ overnight/ Prime95, F@H, & rthdribl (dx9 HDR test) all day with no errors at all. Sure enough though, Counter-Strike Source freezes. Only that its always right in the middle of loading a level that it happens. -Really irritating, pretty much ruling out the PSU, I think, -as it still happens at 3-4-4-8 1t (chipset-detected SPD timings). I guess it's an application thing, or is my chipset truely that bad?

I've noticed that CS:S doesn't like to work right if it gets even a low level of instability. I can't always run it at my video cards' max clock speeds, so I would investigate that in other games as well.