• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

memtest86 test 5 issues

C

Cyberlunacy

Limp Gawd
Joined
Feb 10, 2005
Messages
279
ok eclipse or one of you other technically informed gentlemen.

what is with Test 5 on memtest86 ? and particularly around 98-99% complete.

this test has given me more issues when overclocking than any other of the tests combined.

is this some sort of massively intense test of the memory or something ?

basically,,, i just got my brand new 1 gig dual channel kit of OCZ PC3200 EL Plat Rev 2
TCCD and it DOES have the BrainPower 808 pcb's which from the research i have done is the best of the brainpower pcb's.

ANYTHING over 250mhz just pukes out in memtest. If i try and run 2.5-3-3-10 at 275 or 3-4-4-10 at 275 i get 10's of thousands of errors on memtest test 5. and also very odd errors in test 1.
odd as in one run they will show,, if i reboot,, the next time test one will be clean... but test 5, oh yeah totaly carnage at 99% complete.

i know this TCCD is voltage sensative., but i have read a few forums with OCZ rep's in them that say there are certain kits that were made that actually eat up the volts,, you just gotta have cooling on them or they get insanely hot.

so, given that,,,

i am currently running 230mhz 2-3-2-8 @ 3.0v with memtest coming up clean. i have a fan blowing ON the ram. and so far test 7 is 50% done and ive not had one error.

so,, anyone know about test 5,, and can anyone comment on their experiences with TCCD and voltages at 3.0v and higher ?
 
i don't know exactly what it's doing, but test 5 is also very stressful for my computer.
though normally it'll be pass with flying colors or lock up and i have to use the reset switch to reboot.

also try test 8. sometimes my computer will be just fine on test 5, but 8 kills it.

to get a bit more into technical details... 'block move'
i'm guessing it's taking a chunk of memory and copying it into each address. i honestly don't know the details of how memtest works, so i've always been at a loss to explain why these things happen. :(
 
hhmm

since my DFI has memtest built in, and no its not the hacked bios, its the real DFI beta and DFI included memtest in the bios, i mean holy crap what more can you ask for from a mobo maker,, come on thats awesome!! i hope more people start doing it.

right now im at
3.0v
2-3-2-8 1T
230mhz
and its fine but 235mhz gave me over 100,000 errors at the end of test 5.

eclipse, i was thinking also this morning,,, when raising or lowering CAS timings is there any set guide to go by for stepping when you raise it,,, this is what i mean.

start with
2-2-2-5 then
2-3-2-5 then
2-3-2-8
2.5-3-3-7
2.5-3-3-8
2.5-3-3-10
3-3-3-8
3-3-3-10
3-4-4-10

or have i pretty much hit it on the head. what i just showed you. or can you just mix and match numbers to find the most stable setting.
I understand they all work together to provide clean thruput for data and setting one to high will just create a bottleneck for the others as the latency or idle time or cycles will be too high. but is there any type of "overclockers standard" out there ?
 
i dunno of a standard, but i do something similar to that, but i always leave tras at 10.
and it helps to know about your ram. like i know that cas 2 craps out before trcd and trp
so i can do 2-2-2 up to a certian speed
then i'll need 2.5-2-2
however, my memory controller craps out before i need to loosen timings anymore unless i go to 2t. so really, my actual memory tweaking experience is quite limited.

though at that note, it sounds like you are also being limited by the mem controller, though to 250mhz.
as i forgot to comment on 3v for tccd... well basically, is it just as stable at a lower voltage? to me, there's no reason to bring up the voltage a bunch if it's stable at something safer and cooler.
 
Block move sounds just like what it is, moving large blocks of memory around as fast as possible; it doesnt show errors till its done. I always just start with failsafe (highest) timings and lower one setting to where you want it and see if it introduces errors then after one successful pass I move to prime95.
 
well i dont think its a memory issue at all. because my pc4000 did the same thing. But what strikes me as more confusing is that no matter what mobo i used that pc4000 in i had the same or similar FSB limitation. Which led me to believe it was the ram simply hitting its limit, since that ocz pc4000 wasnt known for good overclocking past ddr500.

now that being said i dont believe i got a stick of ram that is infamous for its cas timings and overclocking, that is bad. The errors are FSB limitation are just almost identicle.
anything over 250 just pukes out on both the tccd and the pc4000.

This to me indicates its either something in the clawhammer core, or the nforce3 chipset.

I could with the pc4000 run a 9/10 divider (180mhz) and crank the shit out of the pc4000 all the way up to 320mhz. which according to Eclipses math equated to be something around 280mhz fsb. which was far over what i got at 1:1 .

this again leads me to believe its an issue with either the CPU or the nforce 3 chipset.

The only motherboard i havent tried this ram in is my MSI NEO Platinum. and frankly i dont want to go thru all the trouble if all im going to have is the same issues.

I have a guy at wotk with a skt478 pent 4 and an Abit IC7 mobo, im going to bring it into work tommorrow and let him take it home and see what his results are.
voltages, timings, fsb, etc.

lastly right now i changed the LDT upstream and downstream from auto to 8bit with no effect,,, also interleaving on and off,, no effect.

Im also still confused on how this dfi handles the LDT multiplier. i can almost never boot at 2x when all my other AMD64 boards will run 2x without a hitch. and infact will not boot at 3x and 4x if i go to high with the FSB.
This DFI, well, hoinestly i think they have the LDT in the bios backwards. i can actually run 4x LDT at 250mhz and up and not have any issues at all, but dropping to 2x is almost guarenteed to cause a no post, or incomplete post.
Performance wise,, there is absolutely no difference running 2x 3x 4x.. they are all the same. 5x LDT forget it,, it wont even post with default settings, not even 200mhz.

Im just a little lost on how the clawhammer and nforce 3 handles everything.
i HIGHLY HIGHLY doubt this ram is bad,, but ill know tommorrow. i dont think the board is bad either.. i just think that anything over 250mhz 1:1 the chipset cant handle it.
 
its about finding out if its the ram or mobo or cpu. not true overclocking bragging rights.. just WHY it wont go as far as it should.
Unfortunately my 2nd system is using the K8N-E deluxe, and thanks to the retards at asus they only have 2.7v vdimm so i cant test anything very good. but it will run 220 2-3-2-5 in that with that particular 3200+ CG.

cuz it can be any one of those or all 3. (cpu mobo, or ram)
or maybe my psu has enough droop in it that anything over 250mhz calls for enough power that the psu wont give it what it needs.
the 1 and only 1 common factor with this system and every other system ive used is the psu.. but that is changing soon. pc pnc 510 deluxe is coming

also Mamma the audigy has 3 AD8620 OPAMPS chips for clearer better audio .
 
hope u dont mind this small hijack buut did you do it yourself?
 
nah i got it from one of the guys here. let me check my heat to see who it was...

tick tock tick tock....

it was ufokillerz and i swear he said he did it himself.
personally compared to my other audigy,, i didnt hear any difference. but there are much better sound cards than the audigy that are true multi channel. where as the audigy cards aren't multi channel meaning one song from the front and another song from the rear mostly used for DJ'ing or mixing. fade one song to the next by fading front to rear.

just tested some other things a while ago... 1 run thru of Test 5 and getting over 2 million errors at 260mhz 3.0v 2.5-3-3-10 and also 3-3-3-10. but they dont happen until right at the end 98 and 99%
this is the EXACT same thing the pc4000 did,, got to 98 and 99% then just puked out oodles of errors.

im gunna start swapping slots and sticks around to see if maybe Slot 0 is bad.
 
i find that the closer the slot is to my cpu, the better my ram clocks at a low voltage.
however, it doesn't eliminate what i'm pretty sure is a limitation of my memory controller limiting me.
though i'm in the same situation. i have ram that should be able to go fast. and it doesn't. most of the time i feel that it's the memory controller on the cpu, but there's always that nagging doubt that it could be the board itself...
someday i'll figure it out :D
 
yup i agree with the onboard mem controller.

but i may have struck gold here..

i swapped the sticks around and so far... 250mhz 2.5-3-3--5 3.0v no errors 4 passes/
the other way around i was getting tons of errors. in fact after 30 min it looked something like this. 700 errors 625 errors, 700+ 600+ 700+ 600+ back and forth for each test.

ive seen people complain about 1 stick working in slot one while the other doesnt and even though i believe it, basic chaos dictates this.

so,, not sure exactly what to call it,, but so far so good.

ive got a long way to go so im in for a long night,, thank god for 2 computers.
nearly identicle i might add lol.
 
haha, you're lucky. i have to kill my computer to do any work on it.
though this talk of memory controllers makes me want to fiddle with my own ram again. :D
 
1 512 stick

290mhz memtest error free 2.8v 3-4-4-10 1T

2 sticks 512

250mhz memtest error free 2.8v 3v whatever i choose 2.5-3-3-8

no cas setting above 250mhz works to make it stable..

VERY VERY disappointing guys VERY disappointing...

please can i have my PC4000 back >:|

anyone want to trade me this ram for some g.skill ?? or OCZ VX ? or a 1 gig stick of TCCD ?

eclipse i cant tell if its this mobo this chip or the ram... but 2 sticks do NOT play well together,,, and here i though OCZ hand tested all their platinum and gold stuff..

yup,,, looks like an RMA is in order

***EDIT***
yup i spoke to soon windows is by no means stable. nope,, openmed IE and got a BSOD with IRQ Not Less Than Or Equal Too error..... ggrrr..

down to 275mhz 3-3-3-10 2.7v lets see what happens now. still one stick.

----------------------------------------------------------------------------------------------------------------------------------
***EDIT #2***

Maybe its just me, although i know the speed of my system but here is where i am now, and i'll swear up and down my system is screaming!!

3-3-3-10 2.7v 1.5x LDT
9x multi - 2475mhz
Sandra 4049/4059
Super Pi 1meg - 36sec

I cant tell if its the LDT or i just hit a sweet spot in the ram but everything is running MUCH faster.
I broke 4000 on sandra FINALLY!!! around 3700 was the old score thats a gain of 300 points.
much more testing to go so...
 
i'll be very upset if you rma something because it doesn't oc as well as you want.
and can you get further with 2t? it could be your memory controller holding you back.

also, did you do at least 2-3 passes in memtest? or just make it past test 5 and decided to go into windows?
 
no actually i been letting run for 5-10 min.

i went to work this morning at around 10:30, and got home around 5:15
at 290mhz both sticks 760 some tests and had something like 1200+ errors.. i didnt stick around to watch it do more.

i will rma it if i feel like it eclipse hehe just kidding... but both sticks not working together just bugs the hell out of me,, although this is pretty normal and not much i can do about it.

need to look for a single 1 gig stick of something and quit messing around with this 2 stick stuff.

check my Edit #2
 
i still havent figured out what the deal is with this DFI and its whacky LDT.
Ive used the chaintech VNF3-250 and the ASUS K8N-E Deluxe and they all ran 2x LDT with no issues, and the more you overclocked, naturally you had to drop the LDT... but this DFI...
The more you overclock it seems as if you have to raise the LDT....
2x is NEVER stable, 4x,,, ive never NOT been able to use it... even now 4x will run,,, but 1.5 seems like its rockin... ive never used it before,, never thought i needed to.

Is there an application that will actually show what the Hyper Transport bux is running at ?

Ya know i dont even wanna reboot or touch the bios the way this girl is running lol.
but i need 1 gig of ram,,, ggrrr 512 is just too much swapping on the HD.

**EDIT***

Ok i moved to 2.5-3-3-10 still 2.7v

gained 2100+ points in 3dmark 2001 from 23017 to 25138


***EDIT*** #2

After installing the 2nd dimm im back to 250mhz. yup no matter what voltage no matter what cas timings no matter what i do, im stuck at 250mhz with 2 sticks.
Even running 2T dont help.
Running a divider helps but only up to around 275mhz,, which puts my ram truly back down around 235 or so,, so its not worth it..

very sad for me i guess....oh well guess ill stay at 250
1.5xHTT didnt run as fast with 2 sticks..
oh well...
 
You must log in or register to reply here.
Back
Top