Printable View
To test stability, I'd run Torture Test overnight. For burning your RAM in, do whatever you feel is going to push your memory over the edge. There are several of us that suspect the Winchesters have a flawed memory controller. If you fail Prime95, don't be discouraged. Run Super PI 32M and loop a few other benchmark utilities such as 3DMark2001 to test for stability.Quote:
Originally Posted by Orker
Got them at 255 2-2-2-5 with 3.4v priming my temporary 3500 newcastle
at 255 x 10 (chip limit at the moment) on 1.7v
they are memtest stable for 20 loops test 5 at 260 2-2-2-5 with 3.5v
(260 x 10) but cpu isnt stable enough to run superpi 8mb let alone
32mb, strange how its memtest stable tho
So just using 255 x 10 for now
I did try 265 x 10 and got errors with 3.6v, reducing the cpu multi to 9 reduced the number of errors to 2 a test so its showing it could also be mem controller limited
Either way sweet ass memory
Yet prime95 failed yesterday at 240 2-2-2-10 yet memtest has been going for the past 16hrs with no errorsQuote:
Originally Posted by conrad.maranan
I was wondering would the RAM burn in faster if you ran memtest despite any errors, shouldn't in theory after a while the errors just stop as the RAM becomes fully burned it (assuming it can run at the selected speeed)?
DudeMiester,
I've heard a good way to burn in memory is find the MAX error-FREE MHz and then raise it by one or two so you're getting a few errors every pass or every several passes... then run MemTest until you're not getting ANY errors at that MHz anymore... raise a bit more and repeat
I just got 220 errors on memtest86 at 240mhz 2-2-2-10 3.4v :(
Hopefully it gets better as the ram gets worn in
My stick that worked was the best at 3.1v this was stable at 255mhz 2-2-2-10 1t. just keep playing with memtest and the ddr booster. find the setting with the least errors and just keep it looping overnight. :)Quote:
Originally Posted by Orker
I posted this in the Official VX Thread, but some of you may have missed it:
Ok guys I am doing some control testing, I am running 2 x 512mb of OCZ PC3200 Platinum Rev 2 TCCD at the 2-2-2-10 settings shown below on a GB K8NS Ultra-939. On my 754/DFI UT rig, these sticks running together will run P95 for days at 200fsb. Looky see what happens here:
(Of course this is my wonderful Winnie 90nm 3500+)
Andy.
That's one more point scored for the boys who feel that Winchesters just suck. Either that or Prime95 causes serious allergic reactions to these CPUs. Either way, I'm still firm in my belief that our memory controllers are seriously flawed.
EDIT:
RMA for my DDR Booster has been approved. I'll be picking up my new unit at 9:00am tomorrow morning. I have a feeling I'm going to murder this poor Winchester that I borrowed.
http://www.mwave.com/mwave/viewspec....iteria=BA20923
for those of you who still haven't ordered. I got a tracking number today, fedex says december 9th woo woo. too bad i'm not ordering my proc till friday :(
yeh cool... even though ive got errors at 240 should i just increase to 250 and keep burning in at that level? coz it boots at 250Quote:
Originally Posted by trans am
So far i have done 17hrs of memtest... All errors(220) were on test 5.. Today i did 999 runs of sandra bandwidth burn
Try twisting that DDR Booster knob just a wee little tiny bit. If you notice less errors, you're on the right track. :up:
so 3.5v is good... Is 3.4v for 240 2-2-2-10 too high?Quote:
Originally Posted by conrad.maranan
what mhz and voltage can you get at 2-2-2-10?
260MHz 2-2-2-10 at 3.6VDIMM was the last I saw before my CPU died. I'm picking up a new DDR Booster tomorrow morning so I'm hoping to post up slightly higher numbers.
If the memory likes it, don't be afraid to push your VDIMM. Just make sure you have active cooling for the Booster or you'll fry it in no time.
I didnt see your post about your booster... did it cook?Quote:
Originally Posted by conrad.maranan
I have a 120mm fan aimed directly at the ram and booster... Its awfully tight on this mobo.. THeres no space between the dimms
My first Booster fried when I first bought it. I didn't think cooling would be such a big deal. I learned the hard way. My current Booster just conked out, out of the blue. When I removed it, the knob seemed to have a lot of play and wiggled a lot. Tomorrow brings me my third Booster in just over a month. I've got the worst luck, man...hahaha! :D
LOL 3rd booster... well 3rd time lucky :)Quote:
Originally Posted by conrad.maranan
So, Orker. How many errors are you down to now after tweaking the voltage? Let me know pronto. I tend to get excited for other people as well, so if you're doing good, believe me when I say I have a big smile on my face. :D
haha :D .. im the same..Quote:
Originally Posted by conrad.maranan
prime unstable at 3.4 3.5 240 2-2-2-10
ill run memtest86 now for 15min on test 5... the errors last time came within the first 5 min, so that should be long enough.. then ill report back
:)
If you notice that you get the same exact error(s) at the same exact memory address(es), swap your modules out. Plug the module in DIMM 1 into DIMM 2 and vice versa. Run Memtest again and see if the errors are still the same.Quote:
Originally Posted by Orker
Oker, cool that booster dude!! It sounds like you cooked the 2nd one also. I removed my pot and add a Cermet 15 turn job on there. Takes car of that issue. Of course there will be no more replacements cause your warranty will be voided, but you may not need have any more issues. :)
So. When a DDR Booster is fried, it's basically just the pot that's cooked? I doubt my second unit was cooked. I have two 92mm Sunon fans dedicated to cooling the DIMM portion of my mobo. On top of that, I have a very small a/c unit blowing cold air to the open side of my case.
your freaking me out... i thought VX was warranted up to 3.5.. i havent gone above that... btw im using a 120mm fan which is about 2cm away from one end of the ram... is that good enough?Quote:
Originally Posted by andyOCZ
btw conrad... at last some good news.. no errors in memtest86 test 5 at 240 2-2-2-10 3.5v... only ran it for an hour though
Damn you OCz!!! I'm trying to buy a Corvette here and you guys go and find some company that's producing BH5 replicas! :cussing:
Hahaha! Congratulations, bro. 74 loops of Test #5 is gold. Run Super PI 32M and then try 245MHz and 250MHz. If you error out, turn the little knob a little bit more. OCZ will honor their warranty for your RAM as long as you don't exceed 3.5VDIMM + 5%.Quote:
Originally Posted by Orker
:toast: