Printable View
loading windows is usually harder on the system as true dual channel comes into effect not to mention the loading of drivers etc. Memtest is good to be sure your memory is stable but cpu, cache and mem controller are totally different things that come into play during the loading of an OS. You can either raise voltage like mentioned above to give the memory a bit more juice when loading the OS or back of a few mhz at the current voltage.
Try raising your vcore, maybe its the proc and not the ram.
4000+ uses clawhammer, which has CG mem controller(prob the best one right now). At 9x 260mhz I am at 2340mhz which is below stock for 4000+. Even when I use 1 stick (single channel) I have problem with window based memtest at 255. I can run memtest at 260mhz 3.33v but can't even run 3dmark01 at 250mhz 3.5v.
Does VX have the same problem or just twinmos SP?
Good idea, or better yet lower muti far below your CPUs limits for more RAM FSB headroom.
I have the same problem,I'm memtest stable at 250 but windows stable at 243 your sticks may be faulty never know.
lol, but isn't memtest suppose to test to see if your ram is faulty or not?
yes,but like cantankerous said, it tests only ram :) not ur controller/cpu or any other factor(that I know of) that affects ram overclocks.
ok, I just ran 25 loops of memtest #5 at 5x 260 2-2-2-8 at 3.4v on DFI SLI-D with all the other timing set to auto. Not one error. Boot into windows and try to run 3dmark01 and it crashed on the second game. Should I up the chipset voltage or the LDT voltage?
Yes I know, but I know the CPU/mem controller is good, test it on TCCP and runs 275mhz 2.5-3-3-7 at 2750mhz just fine.
How is it going now?
What type of error did it give did you check?
Memory is being cooled?
Maybe you video card can't handle those timing being throw at it? Is that possible?
Thought I read something about that on this forum.
i follow the instruction of giving it more voltage. So I throw 4.0v at it and now it is completely stable at 265mhz in windows. 3dmark and everything. I just hope it doesn't die. someone really need to come up with water cooling block for memory, even 1/4 tubing is ok, just make it a seperate loop.
4 volts may be getting a bit hot, better cool those suckers to be safe. they aren't hot when your just surfing the web or writing a paper, but if you start playing games or somthing, they'll start getting warm > warmer > hot
haha, I have fat ass copper ram sink on them and a 92mm fan blowing at it. Even memtesting the ramsink is kinda cool. But one of them died.....
Hmm...funny. I plug in my back up stick and now I can run at 260mhz and 3dmark01 no problem. I bet you anything one stick killed the other one cuz they don't get alone. Oh well survival of the fittest.
Quote:
Originally Posted by mcbarnet007
Are you suggesting maybe something like this?
http://www.kaltmacher.de/viewtopic.p...er=asc&start=0
im gettim some speed premium AA4T soon :WOOT:
hello, one shop give me this number of the twinmos memory:
P/N: M2G9J16A8ATT9F081AA4T
MODEL: M2G9J16A - TT
the chips are winbond UTT???????
yes, they are.
yeahhhh :banana4: :banana4:, in spain is very difficult find a twinmos memory, thanks
no problem. go and buy the sticks as soon as you can. and as many as you can afford. ;)
tell us what the top 5 letters/numbers are on the chips themselves. should be like 04414, then, on the 2nd line what the last 3 things are. should be something like 44D, or 50D.
price? :P
150€=191.76 $
not bad for europe. only about $15 (~11€) over our prices.
spain in technology is very bad for buy components, today is arrive the dfi nf4 sli-d, sli-dr and ultra-d, the price of ultra is 162€ around 200$
http://www.newegg.com/app/ViewProduc...218-062&depa=0
Just received this dual pack. AA4T, 50D. production date is 05034 (translate please). Looks to be a better value than the SP.
its winbond utt...
made in 2005, 3rd week. 50D... "supposedly" slightly worse then the 44D, but i dont really know how true that is.