• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

0450 Winchester 3500+ question

R

Regulator07

Senior member
not sure if anyone knows or has had issues with this, but i have heard that week 50 3500's have a lazy mem controller and might require more volts or looser mem timings. i have a week 50 3500 and since i got it i have been having issues running prime blend. it either craps out in 4 min, 1 hour, or 2 hour...or it will last for over 12 hours. i have tried a bunch of different voltages from 2.7 to 2.9 on vdimm, and 1.4 to 1.44 on vcore, for over 12 hours i ran it at 1.435 which is a solution but sometimes it wont last even at the identical settings. just wondering if anyone has any insight into this and what i could do?

i have run memtest86+ for hours at 2-2-2-5 2.75v with no errors, and prime 1 and 2 for over 12 hours each no errors.
 
I haven't heard anything about that...i would also want to see some real scientific proof and not a bunch of whiners at extreme crying over bad oc's.....

Real proof has to be in place before I would ever believe such a claim....got any links to this???
 
well i can provide a link to who said they were having a problem and found it to be the mem controller, but i dont know if that is even worth linking to. i cant figure out any reason why blend is failing, mem passes all test, cpu passes all test, just cant do blend. i am running some test tonight and will post my results tomorrow. maybe someone can help or suggest something i havent done.
 
Originally posted by: Regulator07
well i can provide a link to who said they were having a problem and found it to be the mem controller, but i dont know if that is even worth linking to. i cant figure out any reason why blend is failing, mem passes all test, cpu passes all test, just cant do blend. i am running some test tonight and will post my results tomorrow. maybe someone can help or suggest something i havent done.
Click to expand...


When you ran memtest how many passes?? The reason I ask is that the passes are long and most of the times ram that is flaky for me will not error in most of the test except test #5 and Test #6....

trying isolating just that test and run 50 passes on each...the test #5 shoul dbe only 45min but the test #6 will take a few hours.....


Have you tried an app called S&M 1.5?? Also run superpi 32mb and see if it gets any errors there...It appears to be a good test for finding issues on the A64s


The only other thing I can think of is heat an issue either in ram or cpu...different times heat can be a variable....

Also how does the power rails look...the powerstream is adjustable....take a voltmeter and test the +12v line and see if that needs a boost....I dont knwo a way to test the +3.3v rail...
 
memtest passed 14 passes at 2-2-2-5 and 2.75v. i will run test 5 and 6 tonight and see what happenes. also, super pi 32M runs fine, never had a problem. i will try the S&M program, if you have a link that would be great. i have not checked the rails with a voltmeter yet, probe tells me 12.09, 3.32 and 5.01. i am not sure heat is the problem but will test it too, should i just point a house fan at the mem (well whole computer side)?

i dont think it is unstable, i have never had a bsod, reboot or error of any sort, games play fine and fast, and prime 1 and 2 run for over 15 hours each. so i am sorta dumbfounded

these were my tests:
CPU: 1.40v to 1.45v, 11x, 5x or 4x HT, 201 FSB
mem: 2.75v to 2.90v, 2-2-2-5, 2-3-2-5, 2-2-3-5

all these in differen combinations all failed either during the first test of 1024K FFT size or when it got to 869K FFT size. I am sure that would mean something, when the memory is stressed with a bigger size it fails.

I am thinking it might be a bios issue and may test that theory tonight, 1007-002 @ 1.435 vcore and 2.75 vdimm 2-2-2-5 let prime blend run for over 12 hours. i am on 1007-003 now so, i will try flashing back.

any thoughts?
 
I would recommend flashing back to the 1004 bios if you plan on O/C ing. 1006 and both of the 1007's are crap for O/C.

But it sounds like you have a problem with your CPU or perhaps one of your ram sticks if you are failing at basically stock speeds. You might try your ram sticks individually, or test the ram in another system or visa versa.
 
yeah i will try flashing back to 1004. i know what you are saying about failing at stock speeds, but what bothers me is sometimes it goes for over 12 hours, but then it will fail right in the begining other times, it is not consistent. with blend i know i am testing both memory and cpu, but i dont understand why it fails blend but nothing else at all, only blend. i have warrenties on everything so i am not worried about rma'ing anything, it is just i have a feeling i will get the cpu or whatever and they will say there is no errors. i am kinda leaning toward motherboard right now, but cant determine definitely and i dont have any way of testing it on other board unless i buy another one.
 
You should be able to eliminate the board as a problem by testing for your max HTT.

Drop you CPU multi to like 7x, your HTT multi to 3x, and use the DDR200 2:1 memory divider and see how high you can push HTT and maintain stability. If using this method you can get a nice high HTT (275-300, mine will hit 325) then I would say the board isn't the problem and its most likely CPU or RAM. But if using this method you can't raise the HTT and maintain stability then mobo/chipset is the problem

Isolation like in Zebo guide is not only good for finding your overclocking limits, but can also help to identify problem components.

Or if you have a local computer repair shop, they may (for a small fee) test your mobo using there workbench reference CPU and RAM. I know this would be a pain, but it can save you the time and expense of RMA'ing the wrong part.
 
alright, i will def try that tonight. i should be able to run blend with this right, or should i use something else to find out the stability? i know i have overclocked the system to 223x11 no problem, again passed everything except blend, i got the same type of errors with that. I am thinking about just biteing the bullet and rma'ing the cpu and memory just to get this sorted out, perhaps there is something i cant test for and there is a problem. either way, i think i am going to give it till next week and if no go, then just send them back, i have heard corsair and amd are both good at getting parts back relatively quick.
 
problem solved: i found out the problem was that prime blend was using too much memory. i have 1024, and prime blend is set to use 923. apparently this was using too much memory, as some is used in windows, and tryen to acess the harddrive which it cant access as fast therefore causing errors. so far i have lowered it to 800 and it works fine. apparently, according to the prime forum, this is a known problem and just means you have to adjust. thank you for the help and i hope this helps other people who might have this error.
 
Originally posted by: Regulator07
problem solved: i found out the problem was that prime blend was using too much memory. i have 1024, and prime blend is set to use 923. apparently this was using too much memory, as some is used in windows, and tryen to acess the harddrive which it cant access as fast therefore causing errors. so far i have lowered it to 800 and it works fine. apparently, according to the prime forum, this is a known problem and just means you have to adjust. thank you for the help and i hope this helps other people who might have this error.
Click to expand...



Good job...this becomes a very good thread....I think this possibly could solve some of the issues back a few months ago of ppl having sisues like this and blaming the earlier then week 40 cpus....

I guess I was partially right...I thought ram had to be the main factor...
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top