Jump to content

NF3 250GB wont work with all 3 slots filled


Recommended Posts

well i got 3 sticks of identical Samsung TCCC 512MB CAS2.5 ram 2 days ago..not looking to oc much if at all anymore, just needs lots of stable ram for video/audio editing...arrived today...popped it into my computer...it defaulted my ram to DDR333 running a 5/6 divider from 200HTT. The highest I can get stable with 3 sticks in is around 180 in memtest...tried loosening timings...raising vdimm to 2.9. 2 sticks works marvelous...same with 1 stick...and i tested...none of the sticks are bad. It generates like 20 thousand errors per one test of #5 memtest86... The ram itself is TCCC, on Brainpower B6U808 PCB.

 

any help is welcome.

Share this post


Link to post
Share on other sites

change your memory from 1T to 2T and underclock (whatever you were overclocking) by a little bit. i could run my system at 250x10 2.5-3-3-10 1T 2.7v with two sticks. with three i had to move down to 245x10 2.5-3-3-10 2T 2.7v. I think i can pull some more out of the cpu with three sticks (maybe 2480-2500) but I haven't been playing around with it a whole lot. the memory i am runnings is pdp xbl. (samsung tccd sticks)

 

three sticks should work fine on the ut250. (albeit at slower timings)

Share this post


Link to post
Share on other sites

Iam running 3 sticks of memory on my setup also and i cant seem to get a good FSB overclock going.I can get 2.6 out of her no problem with the default multiplyer of 11 for my 3400 but i can not get much FSB out of her.Is there a issue with 3 sticks on these boards?

256x2=3500

512x1=3500 corsair BH-5

Share this post


Link to post
Share on other sites

figured i would come back and fix what i said. despite being stable in windows, 3dmark2001, 12hr+ stable prime95, and 12+ hr stable prime 95 blend. my system was not stable in memtest #5 at all at 245x10 with three sticks of tccd. it would give thousands of errors each loop right at the end.(96-98 percent) it was not a memory stick problem as moving sticks and testing them individually indicated.

 

in any case i played with memtest #5 stability.(just for a few loops) and it seems with my three sticks i could only do 218x11 or so stable for this test. (and this is a best case scenario since i was only testing a few loops) my guess is this is a memory controller issue.(perhaps turion mem controller will fix it) anyway i am back to my stable two stick scenario out of fear that the 245x10 memtest problems with three sticks would come back and bite me sometime.(and the fact that i can't deal with 218x11 performance)

 

just figured i would update. anyone else want to comment on memtest throwing thousands of errors and prime 95 blend running for many hours (along with the regular prime 95) without issue? seems weird to me that i would get thousands of errors (right away) out of memtest but be stable in p95 blend and hl2.

Share this post


Link to post
Share on other sites

as i previously stated in my reply to the original poster. it's the limits of the memcontroller of current a64 cpus. this has been known from the first so shouldn't come as a surprise to anyone. perhaps the new gen cpus being produced now will have better memcontrollers.

 

if you can clock your fsb to xxx amount and be windows stable, be happy :) but that is out of the accepted standard limits of the curent processors.

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...