There are 2 0605 to be tested so :)
the beta and te regular one:D lots of work and pleasure ;)
Printable View
There are 2 0605 to be tested so :)
the beta and te regular one:D lots of work and pleasure ;)
http://img394.imageshack.us/img394/2...tstablepm7.jpg
http://img380.imageshack.us/img380/7950/bios1nb2.jpg
http://img380.imageshack.us/img380/6210/bios2yu0.jpg
http://img363.imageshack.us/img363/715/bios3dg1.jpg
http://img394.imageshack.us/img394/619/bios4mu5.jpg
I added a little more vcore dunno if it'll fix the 3h prime error that i had but since it didn't happen in the same test twice i think it's vcore.
I dunno what timmings should i set in the rams and i don't want to push too much the vdimm... any suggestion?
the CRC32 hashes of the two 0605 BIOS's match, so they are the same :up:
@ nunor: NB of 1.33 may be a bit low given you are going FSB 425 MHz. I see that you are also playing a bit with your CPU and NB clock skews. I haven't had to adjust these yet so I don't know how these will factor in.
Re: 0605, I haven't noticed much of a benefit yet but haven't had a chance to push it that much yet either.
A 0.02 increment on the vNB will make prime stop in less than 10 minutes so i had to set it low.
Anyway i tried to run prime again and i got an error in 57 mins this time ffs. I think it's random errors because when i get the error i do the specific test again and i don't have the error anymore :(
Need help to fix the random errors :\
I would raise your vFSB to 1.3v as I have found vFSB has alot to do getting stable. I would also change your Strap to NB to 333 and lower the clock twister one setting at a time testing with memTest and/or P95 each time you lower. Disable PCIE Spread Spectrum and put your PL to auto. once you get stable you can try tweaking these settings tighter. Try putting your ram timings to 7-7-7-18-2T-60 and raise vDimm to 1.96-1.98v.Quote:
Originally Posted by nunor
The problem is that when i increase the NB or vFSB voltage the system becomes more instable :s and i have to find the good values for everything again.
Tried some changes but i got another error after 8h29m :(
UPDATE: Using strap 333 i got an error in 56 mins against 8h29m with strap 400.
AI Clock Twister: Light = error in 50 mins using strap 400.
AI Clock Twister: Lighter = error booting on windows.
Quote:
Originally Posted by nunor
Some of these settings I am giving you are not going to be exact because you are running the Samsung IC's not the D9GTS ic's.
I would consider the 8h49m stable but that depends on the person and what they are using their computer for. If your main use is gaming and can game COD4 for 4-6-8 hours straight with no problems or crashs then I would consider it stable even though you may get an error in p95 at 8h49m.
Just my 2 cents
Timings are not the problem imo. Atm i'm running at 8-8-8-21 2T just to be sure i don't have errors because of tight timings. Tight timings is the next step when i get a stable oc. I just don't know which value to set on tRAS... do you think 21 is ok?
I use the pc mainly for gaming and university works and i don't think i play UT3 more than 4h in a row. I'm just scared because the errors aren't always at the same time and if they are random it can happen at any time.
Atm i'm running prime with AI Clock Twister set to Stronger. It's performing better than with a lower setting, 2h and no errors til now...
another strange board :)
cant either get it right even after months ... still priming ...right now, trying a notch of stuff every time...
naah, soon try another board, and if its the same I can admit myself beaten, that it's not the board if the other one is the same :)
its so damn picky.... and prime is SO random in crashes, im starting to wonder whats up?
btw, this Clock overcharge thing? googled it, but didnt find anything..
Failed the same test... 2048K :(
Can anyone tell me why when i increase the vNB prime fails in some seconds?
couse this board is strange :)
i tried yesterday to do as "everybody" says, that VTT should be slightly over vcc , so i have 1.40vcore and i did put 1.42 vtt , and it failed immedietly in prime, ofcourse i set the GTL to match the vtt, my cpu likes 1.16vtt nothing else wont work well.. so this HIGH VTT thing i dont get why people are using?
and the vNB are also strange as you experienced, dont get it why its so picky? why wasnt it like that before with older boards?
are you priming blend?
nunor, you have done some tweaking that is beyond what many of us have needed to do (NB clock skew and CPU clock skew) How have you come to find that your current settings are not part of the problem? Perhaps the latest BIOS 0605 will allow you to run on auto. With my memory, I have found that a DRAM Clock Skew delay of 275ps for both channels A and B (black slots) has worked well. As deachus has mentioned, increasing FSB voltage has been helpful for increasing the FSB frequency. You may also want to up the CPU pll a bit as well (1.56 would be a good place to start).
Just a thought: with auto settings, voltages are adjusted according to BIOS. For example with FSB setting of 450 MHz, the NB voltage of 1.6v is uncovered which may or may not work. So if you manually set some voltages at the lowest level just to do so, you may in fact be doing yourself a disservice.
I fixed the two 2048K FTT errors. The first one i had to set the twister to Strong/Stronger and the 2nd i needed PL 9. Tested 1 hour of 2048K tests and no more errors. Anyway, is there a way to reduce PL and keep the stability?
I'm running prime blend ofc. Sometimes high voltages don't mean stability. My system runs prime with no errors 12h @ 3.2 and rams @ 1600 with all stock voltages... 1.25vNB, 1.50vPLL, 1.10vFSB and 1.20vcore(bios!). Why would i need more than 1.6vNB to run more 25FSB? I think it just needs the right settings and no excessive voltages.
As i said before tried higher voltages and resulted in 3 second prime errors. About the skews i saw deachus oc profiles, tried something like that and after lots of hours spent the 3 minutes errors are now in 8h.
btw, i'm already using bios 605 ;)
no im sick of this board, one setting works, then it doesnt.
been trying to get my cpu 3hr stable in prime, small or blend, IT DONT WORK nomather what freakin setting.....tried low, tried high tried good GTL... nothing, it still errors or BSOD.
first i change one thing, get back in priming, and like, wow, 1hr gone, 1..20 ...1...30. .. BAM , bsod...
and then i get back in bios, think a bit...what to change, and i decide to lower some nb, and get back then it works for like 2hrs before it BSOD again..
WTF? then i lower it some more, now it crashes earlier.... and then i get it back and it f-upped there to? it works pretty good until it crashes, then the given setting wont work anymore im so sick of this. specially when you are trying trying and trying.... and it just wont work. tried skews nb skews cpu skews, good combination of stuff, not crossvolteging, setting GTL right, upped and lowes GTL a bit, up with vtt, etc etc, ALL , and still it crashes.
I got this:
http://images0.hiboox.com/images/420...cf0b4de44b.jpg
and this is how it looks: http://images0.hiboox.com/images/420...b0ed967494.jpg
For those that would like to ask, the mosfets aren't hot at all without anything on it :D :D
BSOD's are weird... run memtest to see if you got errors. Set the refresh cycle time to 60, it's important. Using auto the board sets it to 88 and i was getting memtest errors but when i defined it to 60 was ok.
Atm i'm trying to fight with the 8h error but it's something stupid. Sometimes i can run the test with success but after a reboot i'll get an error on that test.
About the MB temperature, where is the MB sensor located? I bought a OCZ XTC ram cooler to push a little bit my rams and it lowered my MB temp 10ºC.
BSOD i dont know why, to low voltages usually are becouse of BSOD, and I've done memtest 2.01 many times, but i got it right now i think, passed 3hr 10 mins yesterday, and i stopped it myself, stable enough for me.
but it required 1.42v for 4,5ghz, i think i was to proud to admit that it required that much volt :) so i did EVERYTHING, fine tuned it to the top until i admitted that it needed more vcore :) atleast its tuned now hehe. still getting the rampage extreme i think, now im sick of this board anyway, and hoping for lower voltage and a more tuned board, as I've seen folks getting stable at lower vcores when changing boards.
and yes im using either trfc 72 or 60 , i figured that out a while ago, when it was set to 82 or 86 or what it was as auto, and i was getting crashes and all kinds of stuff, couldnt get why?!?! so after many trial and errors i changed it down just for a lucky shot, and it was rock stable, howcome the spd not putting right value there? sure i have ooold spd in my ram thats from january when the RAM was brand new released.
I noticed something important with this board .
With FSB at 445 mhz and Vram = 1.7 volt , i start a test with prime95 and immediately , i have got an error . But not with superpi .
So , i decide logically to increase Vram at 1.8 volt .
I do the test again and prime95 work well without any error but when i do a superpi test , i notice an freeze after 2 minutes !!
Incredible !! :eek:
The solution to solve this problem was to put CPU clock skew delay at 300 ps and NB clock skew delay at 200 ps like recommended here :
http://www.ocztechnologyforum.com/fo...d.php?p=287371 :up:
So what's with the lower trfc giving a more stable system? I would have thought that the opposite would be true but I don't really know what this setting is actually doing anyway.
HiQuote:
Originally Posted by Razmatazz
About stability I don't know. It helps the bandwith & super Pi for sure ;)
I can't pass the 425FSB barrier. After a lot of tweaking i'm able to run hours of prime with 425fsb but more than this will result in instant prime errors :(
If your ram is not in question ,try to play with cpu clock skew : increase it at 500 ps and test again .Quote:
Originally Posted by nunor
I had the same problem to pass the 400fsb until i found the right skews. The problem is that i don't know how to pick the best GTL's, nb skew and cpu skew :( I'm just trying lucky shots.
Atm i'm using GTL67/69, cpu 400ps, nb 200ps.
I can run prime 8h with no errors but sometimes i get 1 error after 8h30m. I tried to run that test and i can pass the test now and if i run it again i get an error dunno why. Anyway i think 8h is stable enough.
OCCT says stable...