Centon results

[sealion]

My centon pc4000 from the blowout deal came home this morning.


Brief testing:10 loops test #5 in memtest

230mhz 2-2-2-5 2.8volts no errors
240mhz 2-2-2-5 2.9volts no errors
250mhz 2-2-2-5 3.3volts no errors


240mhz will most likely be my 24/7 settings unless someone can reassure me that high vdimm wont kill anything.

[Teflon]

I have run 3.4v 24/7 through my OCZ UTT b4 and they didnt suffer for it at all so go nuts

[EnJoY]

3.5v with UTT should be fine, just keep em cool with a fan over em. Btw, why such a huge jump in voltage for 250? You sure it wasn't other settings that could be changed, like drive strengths?

[sealion]

3.5v with UTT should be fine, just keep em cool with a fan over em. Btw, why such a huge jump in voltage for 250? You sure it wasn't other settings that could be changed, like drive strengths?

I dont know what to set drive strength to. I've always ran my memory at 200mhz 2-2-2-5.

And yes that is a huge jump lol. I double checked to see if something was wrong but it actually was 240mhz at 2.9 volts stable

[ozzimark]

mind telling us exactly what's set in the bios for the memory configuration?

[sealion]

1:1 dividier
enable cpc
2tcl (cas)
2trcd
5tras
2trp
7trc
auto trfc
auto trrd
uato twr
auto twtr
auto trwt
auto tref
odd divisor correct: disabled
DrAM bank interleave: enable

Dqs skew control: auto
dqs skew value: 0
dram drive strength: auto
dram data drive strength: auto
max async latency: fast
read preambletime: auto
idlecycle limit: 256
dynamic counter: disable
r/w queue bypass: 16x
bypass max: 07x
32 byte granularity: disable

[LivingL@rGe81]

You really should set all the settings -

Try these
2 trfc
2 trrd
1 twtr
2 trwt

What do the do with 3.6V running through their viens

[sealion]

I give up!! I'm just going to run 3.3 volts through them.


edit: I put 3.3 volts through them and still got errors. The ram FELT cool but I went to open the door and let super cold air (raining outside) in and the errors went away. I love this ram again.

[Sesto Sento]

you just need to put a good 120mm fan on top of them and they should run i think =) utt with 3.3v start to heat a bit without air blowing on top of them

[EnJoY]

These are the best timings you can use for Winbond BH5/BH6 and BH/CH-UTT. These are very tight, so some suggestions for getting 250 stable with less than 3.2v would be to copy these timings exactly (everything!), but change trtw to "2" and read preamble to 4.5ns or 5.0ns if your cpu's mem controller blows.

For drive strengths (the only thing not listed here), try 5/1, 7/1, 8/1, 6/1, and then try them all with 3 (example 5/3, 7/3, etc). Keep whichever gives best results (no errors!). I think with these timings you should be able to do 250mhz no problem with 3.1v assuming everything scales appropriately.

[sealion]

you just need to put a good 120mm fan on top of them and they should run i think =) utt with 3.3v start to heat a bit without air blowing on top of them

lol I have 3 fans blowing directly over the ram. 1 120mm, 1 60mm or whatever the amd stock fan is, and 1 90mm fan.


Thanks Enjoy for the a64 tweak screen shot

[sealion]

ok 250mhz @ 3.1 vdimm stable for 9 hours in memtest. Odd thing is 255mhz (my goal) gets me 8000 errors per loop. It's ony 5 mhz more!!! Any ideas?

p.s. power from the 4vdimm is better than the 3v rail. 3v rail is only good up to 2.9 volts; anything above that gets errors. With the 5v rail I was able to get better results. For example:

using 3v rail:
230mhz: 2.8v
240mhz: 2.9v
250mhz: can not get stable even with 3.2v

using the 5v rail:
230: 2.8v
240: 2.9
250: 3.0 with 30 errors per loop
250: 3.1 no errors

[bachus_anonym]

Stable vdimm sourced from 3.3v is limited to ~0.2v less than what your PSU's 3.3v rail is at. So, e.g. If your PSU's 3.3v rail is at 3.3v under load (best, measured with multimeter) you will be able to go stable not higher than 3.1vdimm. That's why switching to 5v rail is the 1st thing you need to do unless your PSU has adjustable rails and you're able to bump it to 3.4v+.

Most often, to get above 250MHz fully stable with BH5-UTT (like your Centon) you will need more than 3.3vdimm (sometimes 3.2vdimm). Probably that is why 255MHz is giving you errors.

If I were you I would also run some more demanding stability benches in Windows. Try SuperPi 32M, some SP2004 (Prime95) or Memtest For Windows.

[LivingL@rGe81]

You could just up the vdimm until 255mhz is stable or you could try burning the sticks in with memtest!!!

I generally find if its 100% stable with 3.1V say in memtest I need 3.2V for it to pass prime!!! So keep that in mind...

Are you gonna test higher as I have 4 sets coming here & wanna get a rough idea what to expect

----------------------------------------------------------------------------

What I normally do is find thier max at 3.3V then add 2Mhz & burn in for 24 hours with memtest....

Then I put them in the freezer in an air tight bag for 48hours, let them get to ambient temps again, remove from bag & retest... This approach normally adds 5 - 10Mhz to thier top speed

[sealion]

Been burning them in every night since i got it. They arent improving but oh well =p 3.1-3.2volts 250mhz 2-2-2-5 with every ram tweak default is pretty good to me.

[EnJoY]

Been burning them in every night since i got it. They arent improving but oh well =p 3.1-3.2volts 250mhz 2-2-2-5 with every ram tweak default is pretty good to me.

Use my timings and play with drive strengths!

[sealion]

I tried those timings but got 5 times more errors. I didnt play with the drive strength too much but they didnt seem to help so i quit.

[EnJoY]

I tried those timings but got 5 times more errors. I didnt play with the drive strength too much but they didnt seem to help so i quit.

Wrong attitude man. I've had over 20 different sets of winbond memory, how do you think I figured out what timings were best? Trial and error. If you don't have the desire to put the time in, then settle with a mediocre overclock.

[sealion]

250mhz 2-2-2-5 is not mediocre!! =) 255 would've been nice at 3.2 volts though. When I said "I quit" I meant I stopped doing the drive strength thing and went with higher volts. I'm still burning them in every night

[Repoman]

Well since you say it's pc4000 that's not really an oc at all, maybe you meant ddr400?

Either way I've always found the same thing.. messing with drive strength hasn't ever netted me any more MHz, once you find a value that's good and stable it will be fine all the way up to the max there. I guess it depends on whether your mem controller is limiting you or your memory is limiting you for your max overclock.. say the ram is good to 320 and the mem controller is only good to 300, I bet drive strengths could get something

[sealion]

lol you know what you're right. I cant even oc 5mhz =/

[bachus_anonym]

@sealion,

Did you try 3.3-3.4v yet, off the 5v rail? I'm pretty sure that if the sticks are of decent breed (average UTT-BH5 hits 260Mhz eeeaaaasy) you'll be seeing 260Mhz+ in no time. Get into Windows and run some benches, man

[sealion]

I'm having a difficult time cooling the ram due to the thermaltake big typhoon being so huge. The modules are warm to the touch but seems to still be heat limited.


3.3volts: 255mhz no errors
3.4volts: 255mhz thousands of errors
3.5volts: 255mhz even more errors than 3.4


upon inspecting the modules I've found a tiny gap between the last ram chip and the heatspreader. Both modules (so I assume they were mass produced this way) have noticable gap between them; I know this because when I pinch the heatspreader then release it, it would make the sound of when ... well you peel a sticker off something.

I'm tempted to just rip off the heatspreaders entirely but I've heard mixed reviews about doing that. Centon doesnt seem to have any of these Bh-5 chips left so I guess RMA is out of the question but I'll run it by the centon tech guy on this forum.


I am heat limited for now.

[sealion]

update: I've managed to fix the gap between the heatspreader than the ram chips; I used my musklin blue clip. My mushkin blues are now naked... but they seem to clock better so I'm happy. With the heat problem solved I was able to go to 3.5 volts and boot at 270mhz with 30 errors per loop. Max stable was 3.4volts at 260mhz 2-2-2-5. Good stuff.





260mhz 3.4volts

[friedchips]

nice work champ!! I have 4 of these coming on tuesday and can't wait to get them!!