Jump to content


- - - - -

OCZ 4000VX on NF2 Lan Party B


  • Please log in to reply
211 replies to this topic

#13 medianOCer

medianOCer

    Geek

  • Members
  • PipPipPip
  • 714 posts

Posted 16 February 2006 - 08:45 PM

Sounds good, will wait to see what you find.

#14 Guest_Darien_*

Guest_Darien_*
  • Guests

Posted 16 February 2006 - 10:47 PM

after another eventful restore session I finally got my system up and running... here are the alpha's
3-3-5-3-2-2-3

and the latency's
32 32 32

thx

EDIT:
Odd... prime95 failed @ 250 in minute or less... backed it down to 240 lasted 3 minutes then failed... I did it a second time just in case and it's been running for almost an hour and going strong...

#15 kitfit1

kitfit1

    Only Clocking When Awake

  • Members
  • PipPipPipPipPip
  • 3721 posts
  • Location:UK

Posted 17 February 2006 - 01:59 AM

The alpha that makes the most diff is T(W2P), droping it to 4 from 5 increases mem throughput in Memtest from 1882mb/s to 1920mb/s and for some odd reason does seem to make things more stable(in my rig anyway). Something else that might help with Prime stability is to turn Apic back on, it did with me in the early stages of running VX. As for Sata corruption, i'm stumped on that one. It's never happend to me and i've got 2 Raptors raid0 and a 250gig Caviar, all Sata. Also another thing that will defo help with Prime, turn "sync mode memory bypass" to auto.

If you have a good read through my thread "can i do better", you will find loads of usefull stuff in there.
Another thing that will be handy to know, VX dos'nt need 3.3vdimm until you get to 245 or 246fsb, 3.2vdimm will do. When you were Memtesting at 250fsb did you loop test 7 and test 8 for a couple of hours?

Ryzen 2700x @ 4.2Ghz

Asus CH6

32Gb B-die @ 3333 mhz

Asus 1080ti


#16 Guest_Darien_*

Guest_Darien_*
  • Guests

Posted 17 February 2006 - 08:04 AM

Thx for the suggestions I'll try those now... I turned prime95 on torture test before bedtime at 245FSB,.. I woke up and checked it just now and it only ran for 2.5 hours....

no just test looped memtest 5... never did 7 or 8... I'll do it, but why 7 or 8?

EDIT: T(W2P) dropped to 4 memtest86 still reads 1882MB/s @ 250 and memory sync mode was already on auto

EDIT2: just rebooted a second time and now memtest86 reads 1920MB/s @ 250

#17 kitfit1

kitfit1

    Only Clocking When Awake

  • Members
  • PipPipPipPipPip
  • 3721 posts
  • Location:UK

Posted 17 February 2006 - 09:09 AM

Thx for the suggestions I'll try those now... I turned prime95 on torture test before bedtime at 245FSB,.. I woke up and checked it just now and it only ran for 2.5 hours....

no just test looped memtest 5... never did 7 or 8... I'll do it, but why 7 or 8?


Try Prime with Apic enable and see how that goes.
test8 looping and failing after a couple of hrs is a good sign of not enough vdimm,
test7 looping and failing after a couple of hrs is a good sign of to much vdimm.

Ryzen 2700x @ 4.2Ghz

Asus CH6

32Gb B-die @ 3333 mhz

Asus 1080ti


#18 Guest_Darien_*

Guest_Darien_*
  • Guests

Posted 17 February 2006 - 08:34 PM

Try Prime with Apic enable and see how that goes.
test8 looping and failing after a couple of hrs is a good sign of not enough vdimm,
test7 looping and failing after a couple of hrs is a good sign of to much vdimm.

With APIC on it failed after one hour at 245... so it actually did better APIC off...

#19 medianOCer

medianOCer

    Geek

  • Members
  • PipPipPip
  • 714 posts

Posted 17 February 2006 - 09:06 PM

kitfit1,
Interesting feedback on tW2P at 4 compared to running it at 5. 4 has seemed to run
well for me as well, I was unsure if others had found that also or otherwise. Thanks
for the added info on that.

Another thing that has also seemed to help me is having the tDOE and the tRRD set
on an equal footing, so to speak. In other words at 2-2 or 3-3, haven't messed with
lower values yet (such as 0-0 or 1-1, might mess with those later on).

Darien,
If you still have the original thermal tape on your SB heatsink, then it's possible that
the heatsink/t-tape may not be making complete contact with the SB surface. Even
if it is, thermal tape is rather dismal, usually, at transferring heat when compared to
various types of TIM/thermal goop that's available. You might want to condsider
having a go at improving the cooling/heat transfer with the SB, since it looks like
you've mentioned you were thinking in that direction anyway. Them SBs can throw
many a wrench into a smooth running rig scenario, especially when OCing.

You could try 3-3-4-3-2-2-3 on the Alphas if your interested in that, and since it
sounds like changing to 4 was good for kitfit1 also, it might do something for you?
Another angle on the Alphas would be something like this, 3-3-6-3-2-3-3, which is
still keeping the the tW2P and tR2P in a 2:1 ratio (but bumping them up a notch).

After you're most recent session of Memtesting, did you get any errors or are you
running better now? Just curious on how you're progress is coming and if anything
is panning out with solving what's going on there.

I'm also curious about, but have yet to look into it further, bumping up the chipset
voltage and if that merely rasises the NB only, or if the SB gets hopped up a touch
from that as well? Anyone know for sure on that one? I may have run across that
info before somewhere, at a quick glance, but I don't seem to remember definitively
what the deal with that is.

Out for now...

#20 kitfit1

kitfit1

    Only Clocking When Awake

  • Members
  • PipPipPipPipPip
  • 3721 posts
  • Location:UK

Posted 18 February 2006 - 03:13 AM

With APIC on it failed after one hour at 245... so it actually did better APIC off...


Some peeps have said they find better results with it on, but you don't know unless you try :)

Ryzen 2700x @ 4.2Ghz

Asus CH6

32Gb B-die @ 3333 mhz

Asus 1080ti


#21 kitfit1

kitfit1

    Only Clocking When Awake

  • Members
  • PipPipPipPipPip
  • 3721 posts
  • Location:UK

Posted 18 February 2006 - 03:21 AM

I'm also curious about, but have yet to look into it further, bumping up the chipset
voltage and if that merely rasises the NB only, or if the SB gets hopped up a touch
from that as well? Anyone know for sure on that one? I may have run across that
info before somewhere, at a quick glance, but I don't seem to remember definitively
what the deal with that is.

Out for now


I'm 99% sure raising the N/B voltage also has an effect on the S/B, but so does raising the fsb. As the two go hand in hand that makes it very hot, much hotter than than the N/B is on it's own. As heat=errors or worse, when your working at the limits the only option is to cool it down :nod:

Ryzen 2700x @ 4.2Ghz

Asus CH6

32Gb B-die @ 3333 mhz

Asus 1080ti


#22 Guest_Darien_*

Guest_Darien_*
  • Guests

Posted 18 February 2006 - 09:21 AM

well here is the thing guys... this is weird and disturbing...

first off medianOcer... don't have any problems with memtest86+.. never did... in fact I ran memtest all night looping test 5 with cpc on 250x10 2-2-2-2-11-13-15-3-3-4-3-2-2-3 Dual Channel, over 8 hours, 390 passes not even 1 single error! :nod:

go to windows at 250 with same or looser setting and it always fails prime95 with in 10 minutes usually quicker... NOW understand thats the RAM and NOT my system becuase my ocz play rev 2 can prime 250 till the cows come home...

S&M will also fail the quick test at 250... test 3 i believe... I had never heard of S&M until last night when I was reading though the "can I do better" thread.... I'm only up to page 9... I ususally try to stay away from 30+ page threads = but anyways some insight into the S&M proggie would be nice...

so as far as I can tell...the RAM is good, but there is some odd problem with it that memtest86+ doesn't show and prime95 and S&M do... I'm hoping there is a way to overcome this...

I know you did right kit... you just said prime95 8+ hours starting at 243 and work them up 1 FSB at a time... I'll give it a try (Major pain in the keester) and hopefully that will work..

as for my southbridge... whatever, for now its fine... heat isn't an issue, only going over 253 in an issue...

thx again for all the help fellas

#23 kitfit1

kitfit1

    Only Clocking When Awake

  • Members
  • PipPipPipPipPip
  • 3721 posts
  • Location:UK

Posted 18 February 2006 - 09:48 AM

Memtest will only tell you if your ram's ok, it won't tell you if it's ok in Windows. Don't forget we are trying to run 2x512 double sided ram at 250fsb with a1t command rate on an NF2. Your tcc5 was at 2t, and it is generally agreed that double sided 512 modules won't run at 1t 250fsb on an NF2. Having said that VX does, it just takes a lot patience and a lot of testing to get there. It took me month's.
As for S+M, if it's showing mem failure then you must back down the fsb until it runs clear on a long test and then Prime again. As i said before, start at around 243fsb not 250.
Just as a side note and i know this is going to sound odd, but drop your vdimm down to 3.2v and memtest and prime again. I say that because with my last lot of VX i got much better prime results at 3.2v than 3.3 and it looks like your overvolting when at 3.3v, ie 3.36v.

Ryzen 2700x @ 4.2Ghz

Asus CH6

32Gb B-die @ 3333 mhz

Asus 1080ti


#24 Guest_Darien_*

Guest_Darien_*
  • Guests

Posted 18 February 2006 - 06:43 PM

Memtest will only tell you if your ram's ok, it won't tell you if it's ok in Windows. Don't forget we are trying to run 2x512 double sided ram at 250fsb with a1t command rate on an NF2. Your tcc5 was at 2t, and it is generally agreed that double sided 512 modules won't run at 1t 250fsb on an NF2. Having said that VX does, it just takes a lot patience and a lot of testing to get there. It took me month's.
As for S+M, if it's showing mem failure then you must back down the fsb until it runs clear on a long test and then Prime again. As i said before, start at around 243fsb not 250.
Just as a side note and i know this is going to sound odd, but drop your vdimm down to 3.2v and memtest and prime again. I say that because with my last lot of VX i got much better prime results at 3.2v than 3.3 and it looks like your overvolting when at 3.3v, ie 3.36v.

1t 2t it don't matter... passes memtest fails prime... no difference... first thing I checked... I actually started it as 2t and was surprised when I found it could do 1t just as stable...

update 243FSB:
failed prime 95 in 58 min (which seriously pissed me off :mad: ) that was odd maybe a fluke... passes SnM long test no errors... just finished S&M... trying prime 95 again... (GRRRR!! AHHH!!! WTF! fails prime in 2 minutes!!!)