since my previous mobo was the exact make/model as you, i feel strongly compelled to lend my assistance... hopefully it'll have a postive effect for you ^_~
RichardKemp said:
I've just recieved my second one of these as I blew the vdimm controller on the first.
I was hoping to run my sig rig at 200mhz fsb with an 11 multi out of the box.. but it doesnt seem to be happening.
It can more or less do it but seems to require overly large amounts of vcore, so.. a few questions!
Actually, i think a quick question at this point will help to clarify and will help us to help you more: did you have no issues hitting 200x11 with your previous 8rda+? And is it correct for us to assume all of the other hardware in your sig is identical to that previous setup?
- Does cpu-z have a voltage detection bug with athlon XPs? As it always shows 1.584..
hmm, just checked with my cpu-z copy... as you can see I also am running an athlon (barton), and the voltage i'm seeing is 1.632V at the moment. spamming the refresh button i see that value jump to 1.648 and back... If the value you are seeing isn't varying at all after hitting refresh a few times, that may just mean the voltage supplied to your CPU is very stable... I'm curious to know what your PSU make and model are...
at any rate, i wouldn't worry about this if our main goal is to hit 11x200 and you were able to hit that speed with your previous PSU...
- Is VDD meant to increase in line with VCore or stay quite constant?
To my knowledge, the value you set for VDD and Vcore are independent... raising or lowering one shouldn't affect the value of the other. Are you seeing values (like in CPU-Z's) that would lead you think otherwise?
- What sort of temperature can the southbridge tolerate? I have the northy cooler from by old 8rda+ (which was actually living on the southy of that) slapped on top of the southy of my current board, but do to an unfortunate lack of thermal paste, i dont think its doing much!
(My ceramique has dried up!!
)
I'm not much of a gambler, but if I had to bet on the root of your overclocking woes... I would put a chip on this one.
Undercooling my southbridge on my 8rda+ was a serious bottleneck for me. I did wind up using the stock NB on the SB, just as you are, but I did use thermal paste as well. If you can't snag any off of your friends... don't use nothing! As you may have read/heard, there are many materials out there that will do almost, if not as good a job as thermal paste... as in toothpaste. Though personally I would recommend spending a few bucks for a tiny tube of thermal paste+shipping (which should last you a good while if you remember that you never need more than a film applied to any surface), I would defnintely advise NOT overclocking without some kind of thermal transfer medium... This board, as you probably know, really CAN'T do too much before some sort of cooling is applied to the SB, and I think some kind of thermal transfer material between the SB and it's HS will help a lot.
To answer your question directly... I dunno what temperature the SB is rated for... I never cared enough to set up thermal probes around the SB... and I doubt many would want to heat it to the point of failure anyway just to find out... I can say that the NBHS on your SB should, after a while, get pretty uncomfortably hot if you leave your finger on it...
- I am currently running blend stress test on prime 95 with the following settings:
8 multi
200 mhz fsb
memory at 83% to give 166 mhz. Stock timings, it is rated for 200mhz.
Is this a worthwhile way to test the fsb limits? Because I know there is very little benefit from raising ram past fsb.. so does it apply the other way? Slow ram impedes fsb in this sense? I want to be able to test each aspect (Ram, fsb, cpu) separately..
A golden rule... always leave your memory at 100%.
If you want to test the fsb limits of the mobo, up to 200, start your ram out at 166 (100%) - the ram won't mind. Then, incrementally knock it up say 5~10 fsb at a time until you see an issue. You may find running the RAM slower than it was intended to run (i.e. 166) may actually cause some instability.
Another golden rule... unless the primary role of your computer is to calculate huge prime numbers, prime95 is not the best stability tester. The way I see it is this - no computer is 100% stable, over a long long period of time. The question is, how much money/time/stress will you spend getting as close to 100% as possible, and how much of your overclocking potential (speed) will you sacrifice to have more stability? My
personal recommendation (italicized because I know there are people who are on the opposite camp in these forums who will get stressed if I state this as fact), is this:
if the most intense thing your computer does is play games... bust out the most system-intensive game you play (say, i dunno, Quake 4), and play the game in a way that taxes your system as much as is feasible (run a multiplayer game with max bots on a huge level). If your system can take that level of abuse, then realistically you WON'T see your system crash due to an overclocking issue. Keeping in mind that things like the season (ambient room temps) will affect your OC'ing potential, I'd reccommend doing something like this to test OC stability.
The real detriment to using something like prime95 to test for OC stability is
1) only your CPU is really being stress-tested... passing p95 for hours on end may mean nothing if it doesn't show your memory is on the brink of causing lockups.. remember that the CPU is only one part of the system, and each part of the system must be put under realistic levels of stress... which brings me to:
2) like i mentioned above, unless your computer really is just a number-cruncher day in and day out, your CPU won't undergo that level of stress doing something like counterstrike. Some may say "Ah! but
what if during an intense match, the evil terrorists plant a bomb that uses a mersenne prime for the defusal code? then my hours of p95ing will really pay off! (j/k
)
Any tips appreciated. FYI my best SPI 1M is an impressive 48 secs or so...woot! Up from 55 or so on my old board, which was reluctant to even run all the components at stock speeds.
PS! The old board had ramsinks on the mosfets, and an old pentium cooler on the northbridge.. should I watch out for the heat of these components on this relatively unmodified board? I dont plan to go past 200 fsb..
hmm... this is the second major eyebrow raiser for me... you say you have the SB heatsink from your previous board on your current SB, and the way you've structured this sentence, it would seem you are using the stock NB heatsink on your current NB... is that correct? If that's the case, and the NB heatsink you are using is anything like the stock passive NB HS from my previoius 8rda+, you may be in serious trouble trying to OC at all... there are many cheap solutions availiable for things like a NB that will give it active cooling, and your NB should be high on the list of things you need to worry about as far as cooling!
regarding the mosfet heatsinkds... I wouldn't worry too much... if you can get around to it, it probably wouldn't hurt to take the mosfet HS's off the old board and pimp out your new ride... those really matter when you up the vcore a lot, but if you aren't doing that it shouldn't be a huge issue...
hope this helps
~chixo