PDA

View Full Version : Memory erroring out when adding voltage



13oost
06-17-2005, 07:48 PM
ive been burning in my OCZ Value Series (BH-5) at the reccomended 3.2v but i can only get 3.14vdimm untill i do a vsense mod, so i go to radio shack and pick up the stuff to do the mod...

i come to my appartment and pull out my Sparkle 460w psu and put in my other 480w psu so i can do the vsens mod on the sparkle...

surprisingly i had forgotten that the old 480w psu's vdimm ran ~ 3.38, so im getting 3.22vdimm max

saddly i completely error out in Memtest86+ at 220mhz on my ram at 3.22vdimm and when im running 3.14 vdimm im stable up to 227mhz

why? :confused:

im on a DFI UT NF3 250gb and my active cooling is a Thermaltake SF2 about .25" above the sticks pumping the full 75cfm across the ram... so heat isnt an issue >.>

at 217mhz its stable, so i tightened down a few timmings so it would spit out 3 or 4 errors a pass, found out what addresses the errors was occuring and set my upper and lower limits to those addresses

any clue to why?

just broke out the DMM and im steady at 3.17v, but what is going on.... this dosent make sense, lol...

ive tried flipping sticks around... dont even mention that... if i flip them around then they error out at anything above 215... reguardless

im also running slot 1 and 3 in my DFI UT NF3

13oost
06-19-2005, 04:33 PM
saddly i can get a good 9 to 10mhz by droping the voltage to 3.1...

all errors are in 1004.X, 1008.X, 1007.X (varring each pass) and consistantly 1023.X every pass

i also noticed that i only get these errors every other pass also

EDIT:: its pretty consistant to get the error at address 1008.1 and 1023.6 every time

EDIT::



i belive i may have made history :wth:

230mhz 3.0-3.3vdimm 2-2-2-5 i get 540 errors in 2 passes
230mhz 2.9vdimm 2-2-2-5 i get 8 errors in 2 passes
230mhz 2.9vdimm 1.5-2-2-5 i get 14 errors in 15 passes


is it me or does my OCZ Value Ram w/ UTT BH-5 absolutely abhore voltage :stick:

may try the 4/15 bios soon, but this is kinda wierding me out.... bh-5 that refuses to take voltage, much less burn-in to accept to take the voltage... i mean heck it just went for a 26 hour wall time in memtest86 test 5 at 220 2-2-2-5 3.3vdimm and i saw a small improvement and more consistant addresses at that frequency

felinusz
06-20-2005, 09:01 PM
Try burning them in extensively, this often seems to make UTT "take voltage" better.

Spanners
06-20-2005, 09:28 PM
Found this in another thread -
"here are the settings I use for 2x512 BH5 @ 260MHz, 3.4v

**Dram Frequency Set(Mhz)**= 200(Mhz)
**Command Per Clock(CPC)**= Enable(1T)
*Cas Latency Control(tCL)**= 2
**RAS# to CAS# delay(tRCD)**= 2
*Min RAS# active timing(tRAS)*= 10
**Row Precharge timing(tRP)**= 2
**Row Cycle Time(tRC)**= 8
**Row Refresh cyc time(tRFC)**= 12
*Row to Row Delay(tRRD)*= 2
*Write Recovery Time(tWR)*= 2
*Write to read Delay(tWTR)*= 1
*Read to Write delay(tRTW)*= 2
**Refresh Period(tREF)**= AUTO or 1168
Write CAS# Latency(tWCL)**= AUTO


*DQS skew Control*= increase skew
*DQS Skew Value*= 128
DRAM Drive Strength**= level 2
*Max Asynce lantency**= 8ns
*Read Preamble time*= 5ns
*Idle Cycle Limit*= 16cycles
*Dynamic Counter*= enable
*R/W Queue Bypass*= 8x
*Bypass Max*= 4x
*32 byte Granulation= disable

of course, all ram is different and the cpu mem controller can play a part, but it should give you an idea if you want to try changing anything. Try 1 stick on it's own and see how far it goes, 2 sticks limits my overclock somewhat."
-Trixxy
I have experienced "burn in" also my 2x512 Mushkin Bh5 where memtest stable @235 3.1v but 235@3.2v would error out .. after time tho 240@3.2v was memtest clean . I fairly sure the settings above are from you board so give them a go a post back :p:

13oost
06-20-2005, 09:55 PM
@Fel :: hmm, burning... over 72 hours of burnin at 2.96vdimm, over 48 hours of 3.1vdimm burnin, 26 hours of 3.3vdimm burn in all at 220 2-2-2-5, and i see VERY little improvement on anything above 3.0vdimm, anything 3.0v and above and it starts adding errors at that

i found my problem, my week on my core is 0431, i was talking to anohter XS member and found out that the revised memory controlers was put on at 0433, so i barely missed it

@spanners:: ive tried that setup except the tref since it changes for frequency... also my ram HATES drive strenght of 2, it likes a weak drive strength

but yea its the memory controler and its evident since the memory seems to bow out at 3.0vdimm

and the older memory controlers was strictly 2.9v max, so yea, its this crap memory controler.... atm i wish i could afford some tccd =/

my memory currently hits its stable limit at 230mhz 1.5-2-2-5 (2-2-2-5/6 is less stable) 2.9vdimm anything over and its like taking a torch to it

Obrom
06-21-2005, 07:35 PM
13oost, I don't have much helpful information but i can tell you i am having a very similar problem on a Nforce2 motherboard.

OCZ Value Winbond Series (BH5) 2x512MB SPD/Sticker:2.5-4-4-8
0516K is printed on the chips.

It won't boot at cas 3 :D
does 200mhz 2-2-2 at 2.7v dual channel no worries :D :D
More voltage past about 3.0v = more errors in memtest :(

Tested each stick seperately, and 222mhz seems to be as far as i can go. 220 dual channel just isn't happening, but close.

Each stick seems to be responding slightly differently to voltage:

Stick"A" Test #5
225mhz 2.9v 286 errors per pass
225mhz 3.0v 20 errors per pass
225mhz 3.1v 42 errors per pass
225mhz 3.2v 116 errors per pass
225mhz 3.3v 216 errors per pass

Stick"B" Test #5
225mhz 2.9v 1750 errors per pass
225mhz 3.0v 12 errors per pass
225mhz 3.1v 2 errors per pass
225mhz 3.2v 6 errors per pass
225mhz 3.3v 14 errors per pass

I have done about 50 hours of burning in + more general gaming etc.

Also, when i am running the sticks in dual channel 220mhz the errors seem to come every "other" pass of test #5. Same as you 13oost.

I have heard a fair bit about Nforce2 having a weak memory controller and trouble with two double sided DIMM's, but i was able to run 260mhz dual channel 2.7v 2.5-3-3-11 with some corsair 3200XL TCCD, 2x512MB.

Is the NF2 memory controller "incompatible" with voltages >3.0v on the ram?

I guess I need to test them with NF4+Venice :P

13oost
06-21-2005, 10:05 PM
mine did 220 2-2-2-5 on my NF2

tccd is a whole diferent ball park

when hitting high frequencies w/ ram on a NF2 you are constantly fighting the memory controler... and no 3v+ is perfectly fine if you have the means of supply it

i was talking to babyelf and told him that these sound like a crossbreeding gone bad Winsung BHCD, lol... all the worst qualities of each, lol...

havent tested seperatly but im stable @ 2.9vdimm 1.5-2-2-5 up to 230mhz, but i think it may be the memory controler on my cpu, but could be wrong

cant complain to horrible much for what was payed, except they are BH-5 and they seem to hate voltage which is extreamly uncharacteristic of a Winbond chip.