The Official OCZ VX Thread - Strictly Test Data Only!!!

[conrad.maranan]

I'll pray that you do at least 265MHz in dual channel.

[Orker]

I'll pray that you do at least 265MHz in dual channel.

lol thanks dude

we now have identical rigs.. main components wise

Ill run memtest shortly i dont remember passing at 260 before..forget how much voltage .. that was about a week ago.. it should have "burned in" more by now... i dont want to go over 3.6

[andyOCZ]

Sweet. And don't forget to test out the dividers for me.

Hey buddy, you live up the road a spell from me I live in Vista, near Oceanside.

So, I tested the dividers for you at 2-2-2-10 1t, 2.85v with the 166 memory buss selection running an actual 187mhz on the K8n neo2 running the 4000+ and it works fine. No memory errors in memtest. I had removed the booster and out the memory in slots 1 and 2 to be sure.

I then tested at 275 HTT and 166 memory buss selection for 225mhz on VX. 2-2-2-10 1t, 3.3v and it errored in memetest.

Next I set 2.5-3-3-10 1t with the same settings as above and there were errors.

So, I grabbed the only thing I have that will take the voltage, good old OCZ BH5 2 x 256 and it worked fine at those last settings.

Now I'm wondering if the errors are due to running 2 x 512mb. So......I load up OCZ TCCD 2 x 512mb at the same settings, but lower voltage of course......and TCCD works fine with the dividers.

Now, truthfully I can't see any benefit running A64 with a divider. The CPU is not FSB limited like an Intel. There are aslo some unusual timing issues that come up when using dividers on the on die memory controller. I believe this is what the issue is here. It may possibly be a BIOS or chipset related issue that shows up when pushing the VX. Any theories? Anyone have other memory that dosen't like dividers?

[Orker]

Looks like im stable at 250 2-2-2-10 3.5v

7000 sandra memory bandwidth!

[andyOCZ]

Looks like im stable at 250 2-2-2-10 3.5v

7000 sandra memory bandwidth!

Very nice!

[Orker]

Very nice!

make that 255 2-2-2-10 3.5v memtest and prime

[conrad.maranan]

Congrats, Orker!

Andy - The reason why I show interest in the memory dividers is because I would like to push my CPU a little bit further. For example, what are my options when I want to run a CPU multiplier of 10 and an HTT of 316MHz, but my RAM maxes out at 263MHz? The 166MHz option in BIOS would be perfect to get my memory down to the target of 263MHz, but selecting this option causes the system to hang. I'll be borrowing my buddy's two-stage cascade for some testing, but I see that with my 3500+, I'll be limited to running 11x263MHz if I must insist on using the VX modules.

So now the question is: BIOS issue or NF3 issue? I can run dividers with my Platinum Deuces, which is what led me to question the VX modules as being the problem.

[Orker]

Thanks conrad

.... btw what is your vcore and booster voltage at 260 2-2-2-10?

[arj]

Anyone have other memory that dosen't like dividers?

I had the VX for a few days in my P4 while waiting to get a new NEO2. On my AI7, it ran perfectly fine at 220Mhz 2-2-2-5 3.2v 1:1 but when I tried to run at at 275Mhz 5:4 which should give me 220Mhz memory, the PC won't even boot. Didn't do any further testing though.

I didn't read the previous page and jumped straight to the last page so forgive me if what I said is totally unrelated or has nothing to do with what you guys are talking about

Back to the VX and NEO2, I picked up a new board yesterday and only had time to install it a few hours ago. The ram is now at 255Mhz 2-2-2-10 3.43v (multimeter) and just crossed the 2 hour mark of memtest test #5

[cantankerous]

awesome! Let's see if 260+ is in store for you.

[conrad.maranan]

VCORE - 1.550V + 10% BIOS
VDIMM - 3.51V Digital Multimeter

[Rabbi_NZ]

VDIMM - 1.51V Digital Multimeter

WOW! that's lower than DDRII hehehehe

Hopefully I will get to test these badboys on a 939 setup one day soon.... and even better, with more than 3.3v... under load I measure vDimm to drop to as low as 3.34v in Sandra when BIOS is set to 3.4v so obviously my 3.3vRail jus isn't high enough, plain and simple... 3.3vDimm BIOS setting actually sits at bout the same level

[conrad.maranan]

My fingers were overclocked and produced errors on that post. I've edited it to reflect 3.51VDIMM.

I said it to you in my e-mail and I'll say it to you again: I think you've got the best set of VX around. Socket 939 and maybe a little bit more voltage should really make those puppies shine.

[Grov]

Yes, quite impressve stuff.

Without a booster it seems to be stable at 240 10-2-2-2 with only 3.1vdimm. Gonna get a booster after me xmas holiday. Shootin for 10x260.

[Orker]

Yes, quite impressve stuff.

Without a booster it seems to be stable at 240 10-2-2-2 with only 3.1vdimm. Gonna get a booster after me xmas holiday. Shootin for 10x260.

10*260 is tough.. Im having alot of errors at that level even with 3.6v... Looks like 250 3.5v is my max for the time being

[andyOCZ]

Congrats, Orker!

Andy - The reason why I show interest in the memory dividers is because I would like to push my CPU a little bit further. For example, what are my options when I want to run a CPU multiplier of 10 and an HTT of 316MHz, but my RAM maxes out at 263MHz? The 166MHz option in BIOS would be perfect to get my memory down to the target of 263MHz, but selecting this option causes the system to hang. I'll be borrowing my buddy's two-stage cascade for some testing, but I see that with my 3500+, I'll be limited to running 11x263MHz if I must insist on using the VX modules.

So now the question is: BIOS issue or NF3 issue? I can run dividers with my Platinum Deuces, which is what led me to question the VX modules as being the problem.

Honestly I dunno what the cause is. I see your point about the super cooling. Looks like you need an FX55 or to use the 3200 II until we figure it out. Just kidding about the FX55, but the 4000+ I have now is really nice. It appears it share the same cache and core size as the FX55, but runs at FX53 speeds stock.

[jinu117]

Am I like the most accursed OCZ customer in world or is there QA issue again... (which every OCZ employee I talked to there gurantees me there shouldn't be any QA issues).
I bought 2 pairs of VX. (so 4 sticks of 512mb)
Now, they do come matched in good way or bad way in this case...
From best to worst. All test done @ 3.5v. Will test the bottom 2 duds at 3.2v and 2.6v to see if they even work @ guranteed timing. Memtested on #5 for 30 loops each as I wanted to get preliminary feeling quick.
267 2-2-2-10 (!!!!)
255 2-2-2-10 -> Not bad at all.
225 2-2-2-10 -> errr.... @ 3.5v?
-200 -> couldn't do 200... this one will get through testing done and probably will be RMA'd as well as 225 one that is... just... (I bet it will fail stock volt/timing test considering how poor it did with 3.5v)
Can we say.... I've just covered about whole spectrum of below 200 to top end of VX with only 4 sticks?... I really think there should be some more consistent result...

[andyOCZ]

jinu117,

Sorry you seem to always have bad luck. We do appreciate your loyalty and will look after you. RMA those 2 sticks and the guys will test them. I am interested in knowing what the "duds" will do on out test rig.

Thanks

[jinu117]

Who knows... Just tested one of the duds on 2.6v 2-3-3-8 and failed.... running memtest 7... (hmmm not 5... strange) in the mean time the better 2 are running in pair in dual channel @ 250 no sweat right now. Wonder if I will actually get 254 dual channel

Mem runs warm (probably around 30c-32c) @ 3.2v but at 3.5v it is hot as in 45-50c with 2 80mm fan blowing on it. How hot are yours for those who run 3.5v?

Yeah it must be my curse... either i get something exceptional from OCZ or something that doesn't even do stock well. Well, the silver lining is that I do get few exceptional things here and there I guess :P

[cantankerous]

Don't feel bad Jinu... I have had horrible luck the past year. Very weak chipsets on my motherboard, a mobile Athlon that can't even do 2.5ghz, an X800 that has flashing textures all the time and ram that won't clock anywhere near what others are getting with the same stuff. It sucks I know. To top that off I had 2 psu's blow on me and a dvd burner that decided it didn't want to burn anymore. bah.

[arj]

With 3.43v, my VX is 10 hours memtest Test#5 stable at 255Mhz 2-2-2-10 but failed Prime95 Blend Test and SuperPi 32M no matter what I do. I already tried increasing the vocre and use lower multiplier and the result is still the same. I can only complete 32M SuperPi at 240Mhz. Is there anything else I should change?

[blinky]

edit: woops my bad

[conrad.maranan]

blinky - That looks more like 2x256MB to me. My questions always revolve around stability. Some people can post some sweet looking suicides, but what is the max stable frequency they can run at.

Rabbi_NZ is at 268MHz using two 512MB modules - Super PI 32M stable! I'd like to see BH-5 do that.

[bachus_anonym]

has anyone matched BH5 yet?
heres a speed to shoot for...
macci hit 275 2-2-2-5 with 2x512, on p4c

yep, looks like 2x256MB to me.... i have a screenie like that too.... also SuperPi1M @ 272x10 6-2-2-2.0 Dual

EDIT: also, with all due respect for macci, if CPU-Z is all this DDR550 2x256MB Dual is capable of then this is certainly NOT something to shoot for

EDIT2: nevermind... i've seen better screenshots on macci's website but not with 2x512MB

[aCidbAbY]

Who knows... Just tested one of the duds on 2.6v 2-3-3-8 and failed.... running memtest 7... (hmmm not 5... strange) in the mean time the better 2 are running in pair in dual channel @ 250 no sweat right now. Wonder if I will actually get 254 dual channel

Mem runs warm (probably around 30c-32c) @ 3.2v but at 3.5v it is hot as in 45-50c with 2 80mm fan blowing on it. How hot are yours for those who run 3.5v?

Yeah it must be my curse... either i get something exceptional from OCZ or something that doesn't even do stock well. Well, the silver lining is that I do get few exceptional things here and there I guess :P

mine was really hot then i figured out to use slots 1-3 then the fan can better cool them

EDIT::>>>whats up with all the chatter guys this is the "strickly test data only thread"