• 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.

New Power Supply

J

Jay487

n00b
Joined
Mar 4, 2007
Messages
14
I know system specs will be helpful, so I'll start with those.


Asus A7N8X Deluxe
AMD Barton 2500+ OC'ed @ 2.2GHz (11x200, 1.8vcore)
Antec SL400 (400W) Power Supply
1 GB PC3200 RAM
Nvidia 7600 GS OC 512MB AGP Video Card
(2) 7200 RPM HDs


I've had about 3 random restarts in the past 2 weeks. I've run memtest to double check memory, which turned up 0 errors (as I expected) I know its not a heat issue, temps are fine (no higher than 40C under load). I've played FEAR for 3 hours with no issues, and then when just browsing the web it has restarted, which is another reason I don't think its a heat issue.

I can only think that my power supply is going bad. (got a new video card and hs/fan for xmas, thinking the extra strain of video card and OCing is killing it, its almost 4 years old.)

So anyways, if my analysis is correct, this a replacement I've considered: http://www.newegg.com/Product/Product.asp?Item=N82E16817153028#spec

I dont need anything amazing, as my system is still kind of old, and trying to keep cost as low as possible. (especially if it ends up not solving the issue) but I dont want to buy a cheapo one either.

Are all signs pointing to a dieing power supply?
Any input on the one I've selected?
Any better suggestions for about the same price? ($60 after rebate)

Thanks in advance! :)
 
Well the PSU you're running should be more than capable of running that system, but there is always the chacne it could be going bad.

I can see why you don't think it's a temp issue since playing FEAR would incrase temps, and if it were a temp issue, it would surly restart. However, the same can be said for the PSU. Playing FEAR requires a lot more power than browsing the web, so you would think if it was a PSU problem, it would restart while playing, not merely browsing the web.

As far as memtest is concerned, the fact that it passed says nothing. I have had faulty DIMMs go through several dozen memtest passes without generation a single error, but fail within seconds of doing a Prime95 blend test. That said, I would recommend you run Prime95 (or Orthos) blend test before spending money on a PSU that may not be the problem.
 
Thanks - I'll run Prime95 and see what happens

as for the first response - I don't believe it has a P4 connector, but I'm not really familiar with what that is, anyway I can check? :eek:

20 seconds into a blend test:

Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 0 minutes - 1 errors, 0 warnings.
Execution halted.


What does this error translate into?
 
Jay487 said:
Thanks - I'll run Prime95 and see what happens

as for the first response - I don't believe it has a P4 connector, but I'm not really familiar with what that is, anyway I can check? :eek:

20 seconds into a blend test:

Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 0 minutes - 1 errors, 0 warnings.
Execution halted.


What does this error translate into?
Click to expand...

There isn't any way to tell what caused it to fail in Prime95.

Your machine's symptoms are pretty odd, you'd think if the PSU was dying you'd see problems in high-load situations (like playing F.E.A.R.) but you don't. Nevertheless, I'd try a new power supply just to see if it makes a difference.

Your board doesn't have a P4 connector so go with the one david recommended.

If you still have issues I'd take a hard look at your processor. You're running quite a bit of voltage through that Barton and that takes it's toll after a while.
 
Jay487 said:
Thanks - I'll run Prime95 and see what happens

as for the first response - I don't believe it has a P4 connector, but I'm not really familiar with what that is, anyway I can check? :eek:

20 seconds into a blend test:

Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 0 minutes - 1 errors, 0 warnings.
Execution halted.


What does this error translate into?
Click to expand...


OK, now run the SMALL FFT test, if that one runs with no errors, there is a 90% posibility that memory is your problem. (before anyone asks, that 90% figure I just made up, but more often than not, that is what the problem is, I have had, on rare occassions where the motherboard was the culprit)
 
Rabid Badger said:
There isn't any way to tell what caused it to fail in Prime95.

Your machine's symptoms are pretty odd, you'd think if the PSU was dying you'd see problems in high-load situations (like playing F.E.A.R.) but you don't. Nevertheless, I'd try a new power supply just to see if it makes a difference.

Your board doesn't have a P4 connector so go with the one david recommended.

If you still have issues I'd take a hard look at your processor. You're running quite a bit of voltage through that Barton and that takes it's toll after a while.
Click to expand...

Reason I'm up to that voltage is because when I started to OC, I just kept upping it until it was stable.

i.e. I had it on say 1.7, It restarted before it even got to windows logon screen. 1.75, got into windows but after 30 seconds it restarted. 1.775, seemed stable but restarted when playing a game for only 10 minutes. Bumped it to 1.8 and it had been stable for quite some time, at least 3 weeks, until recently I've gotten a couple random restarts. My understanding was the voltage increases heat, so as long as you kept an eye on the temps then you'd be OK. :confused:
 
Heat is not the only cause of processor melt down. You can very easily supply enough voltage to fry a processor while maintain good enough operating temps.
 
RamonGTP said:
OK, now run the SMALL FFT test, if that one runs with no errors, there is a 90% posibility that memory is your problem. (before anyone asks, that 90% figure I just made up, but more often than not, that is what the problem is, I have had, on rare occassions where the motherboard was the culprit)
Click to expand...

Self-Test 8K passed so far on SMALL FFT test.... I have 2 sticks of 512, so I will try taking one out and running the blend test and see if the error goes away, and vice versa.
 
RamonGTP said:
Heat is not the only cause of processor melt down. You can very easily supply enough voltage to fry a processor while maintain good enough operating temps.
Click to expand...

what *should* it need to run at 2.2ghz? I can back it down to 1.775 again and see if it will be stable :confused:
 
Jay487 said:
what *should* it need to run at 2.2ghz? I can back it down to 1.775 again and see if it will be stable :confused:
Click to expand...

That isn't really a question that anyone can answer. All processors are different. I've had a Barton 2500+ that wouldn't be stable at 2.2GHz no matter what I did and I've had others that could get there with relative ease.
 
Ramon is right, there is no guaranteed overclock. If you never stress tested that overclock then there is no guarantee that it really is stable.

It seems likely to me that your processor has been introducing errors into your system over the past few weeks and that is resulting in the system instability you're seeing now.
 
OK, tried swapping out RAM sticks, still failing the blend test right away....

turned my CPU settings back to default (11x166, 1.65vcore)

and the test has been running for ~5 minutes without an error yet.

What would cause this error to show up only when OC'ed?
I guess the bettter question is how do I obtain a stable OC? 3 hours of fear with no problems SEEMED stable to me, but fails the blend test right away :(
 
If you're failing BLEND but passing SMALL FFT, that suggests that the CPU is doing ok, but either the memory OR the northbridge on the motherboard is causing errors.

First thing I would do is relax your memory timings and possibly bump up your dram voltage by .1 volts, and then run Blend again.

Keep in mind, that if it is the memory causing the problem, it may actually be bad. I have a stick here that will not run stable at all at it's rated speed of 200MHz (DDR400) but will work at DDR333, so I have religated it to the "guest computer" which doesn't need DDR400 speeds.
 
RamonGTP said:
If you're failing BLEND but passing SMALL FFT, that suggests that the CPU is doing ok, but either the memory OR the northbridge on the motherboard is causing errors.

First thing I would do is relax your memory timings and possibly bump up your dram voltage by .1 volts, and then run Blend again.

Keep in mind, that if it is the memory causing the problem, it may actually be bad. I have a stick here that will not run stable at all at it's rated speed of 200MHz (DDR400) but will work at DDR333, so I have religated it to the "guest computer" which doesn't need DDR400 speeds.
Click to expand...

I can't really touch memory timings. I mean, I have the ability to in BIOS, but for example:
Right now I'm at 11x200, with timings of 7-3-3-2T. If i even change the option to "Optimal" - it changes them to 9-4-4-3T. If I save and exit, the system will not post. I cannot say that I have a solid enough understanding of memory timings to tinker around with them indivudually.

When I had thought it was the OC that was causing the issue, I had it up to 1.95GHz, timings of 7-3-3-2, and it passed the blend test. (was increasing FSB by 5MHZ, and running the test, and repeat)

UPDATE: I changed timings to 7-3-3-2.5T (just changed 2T to 2.5T)

Ran Blend test. Failed, but with a little different error. This Time I got
FATAL ERROR: Rounding was 0.4676833153, expected less than 0.4

as opposed to the previous error of
FATAL ERROR: Rounding was 0.5, expected less than 0.4

does this mean maybe memory timings is the issue, and I'm getting "warmer?"
 
Not necessarily, and error is still an error. If you can't adjust timings your options are limited to one of two things...

1) Overclock less
2) Increase voltage to the memory (may or may not solve the problem)
 
my computer just restarted during a game again.... so I un-overclocked it back to default speeds... seeing if it stays stable. So far, so good.

If it does NOT reboot for a couple days, then for whatever reason it only reboots when OCed. Im going to test RAM again to see if both sticks are stable @ 200mhz (they are supposed to be)

assuming they both test out OK, how could I possibly fix this? The 400mhz decrease is definatly a noticable performance drop, especially bottlenecking my videocard.
 
I thought you said you already tested them individually and they failed at 200MHz???

Here are your options.

If the ram doesn't work at DDR400
-Run them at a speed they are stable at
-Get new ram
-Increase voltage to RAM

There is also the possibility that if the ram does not work at DDR400 that the chipset isn't capable of running at 400MHz without a volrage incrase to the chipset.
 
Do you think that the error Im getting in the blend test is directly related to my random restarts?

i.e. - if I can get rid of that, I can assume it should stop restarting?
 
Jay487 said:
Do you think that the error Im getting in the blend test is directly related to my random restarts?

i.e. - if I can get rid of that, I can assume it should stop restarting?
Click to expand...

Yes, I believe they are directly related. You need to run those tests error free for several hours to really know if it is stable.
 
Ok, I've taken out a stick of memory and ran the blend test with FSB at 200mhz. Failed. (although it did run for 6 minutes- longer than usual)

Swapped sticks. Failed again.

I find it hard to believe that neither stick is stable @ 200mhz, when they are both advertised to be so. They are not the same type of memory either. Ones OCSystems, one is OCZ. I tried upping the dram voltage to from 2.6 to 2.7, failed as well.

Anyways, if its not the memory, the only thing left to cause this issue would be the northbridge, no?

I'm going to lower the FSB some and run the blend test again.

*- lowered FSB to 180MHz (so Im at 1.98GHz) - on test 3 of the blend test so far, and still going

Thanks for all the help - so far you've at least saved me from wasting money on a component I didnt need! :)
 
At this point, you either have two bad sticks, or a stubborn chipset IMO. Many nForce2 boards have an option in the bios to up the voltage to the chpset.

I had an old EPoX board that did NOT have this option and I had to do a volt mod to it which requried some very careful soldering.
 
I looked but unless I missed it, I did not see an option to increase voltage to the chipset.

I ran the blend test for about 12 hours without an error on 185MHz FSB. Bumped it up to 190MHz and letting it run again for another 12 hours. Hopefully I can get it up to 195 or so, and then increase my multiplier to 11.5 so that I can still be around 2.2GHz. The board (Asus A7N8X Deluxe) is SUPPOSED to support 400MHz FSB, guess I just got a stubborn one. :confused:

Although the RAM seems to be fine, one thing I did notice is that after I did a MEMTEST (and was error free for multiple passes) I stopped the test and rebooted. Upon reboot, I got the message saying "Windows has recovered from a serious error" although there was no error. This is the same message I recieved on another computer when it used to reboot on its own (ended up being bad ram, failed memtest). However I do NOT get this message after MY computer reboots on its own. I also got this message one other time after I had changed my FSB back to 200mhz and rebooted.

Thanks for all your help thus far, I appreciate it immensely. :cool:


Here's a random question that has me stumped:
Fixed the mouse issue = had to disable the ability to disable usb devices for power saving
 
alright now im just confused

blend test ran for 12.5 hours on 190x11, 1.75 vcore - no errors

bumped it up to 195, had an error within 5 minutes. Bumped vcore up to 1.775, took SIX HOURS to get an error.

Back down to 190, tried upping the multi to 11.5 (fsb is the same, therefore the blend test should pass with no problem I thought) and it had an error right away again.

my question is, I thought the beldn test was testing my RAM modules and or chipset, but the vcore setting on my CPU seems to have an effect on it as well.

This is from STRESS.TXT in the Prime95 program:
WHAT TO DO IF A PROBLEM IS FOUND?
---------------------------------

The exact cause of a hardware problem can be very hard to find.

If you are not overclocking, the most likely cause is an overheating CPU
or memory DIMMs that are not quite up to spec. Another possibility is
you might need a better power supply.

If my power supply is actually bad (and thats what is causing the failing test, and rebooting) is this still a possibility with such a drastic change in test results? (sometimes it runs for 12+ hours without error, other times it runs for 2 minutes and finds an error)

also is vcore at 1.8 really that dangerous? I mean it has more than enough cooling, but am I killing my cpu by the minute?
Im thinking that I need to play with my RAM timings a bit, but it seems whenever I alter them from 7-3-3-2 my system wont even post.
I are lost :eek:
 
Well the Blend test tests a lot of ram, which also naturally tests the chipset since it is the connecting link between your RAM and CPU. The CPU is doing all the processing so it is also testing the CPU... Hence, the "Blend" test... Since you passed SMALL FFT, which tests the CPU almost exclusively, it is usually safe to assume the CPU is ok. However, like I said previously, to really know if it is stable, these tests all need to be run for several hours to make any concrete conclutions.

If you have narrowed it down to the CPU, there might not be much else you can do. Some 2500+ just won't reach 3200+ (2.2GHz). Others may reach it, but not without a lot of voltage, which then introduces another problem, which is a MUCH hotter running CPU that could lead to instability.

FWIW, I've fried 2 Barton chips in my time, both were overclocked to 2.2GHz or more with incrased vcore. Not right away, but over time I started seeing more and more instability, followed by times when it would not boot properly, then it wouldn't boot into windows at all, and finally failed to POST all together.
 
RamonGTP said:
Well the Blend test tests a lot of ram, which also naturally tests the chipset since it is the connecting link between your RAM and CPU. The CPU is doing all the processing so it is also testing the CPU... Hence, the "Blend" test... Since you passed SMALL FFT, which tests the CPU almost exclusively, it is usually safe to assume the CPU is ok. However, like I said previously, to really know if it is stable, these tests all need to be run for several hours to make any concrete conclutions.

If you have narrowed it down to the CPU, there might not be much else you can do. Some 2500+ just won't reach 3200+ (2.2GHz). Others may reach it, but not without a lot of voltage, which then introduces another problem, which is a MUCH hotter running CPU that could lead to instability.

FWIW, I've fried 2 Barton chips in my time, both were overclocked to 2.2GHz or more with incrased vcore. Not right away, but over time I started seeing more and more instability, followed by times when it would not boot properly, then it wouldn't boot into windows at all, and finally failed to POST all together.
Click to expand...


yikes, thanks for the info

It ran for 12.5 hours on 190x11, maybe I'll leave it at that for now
 
You must log in or register to reply here.
Back
Top