Need help isolate Obscure Glitch on new rig

January 18, 2010 at 13:29:34
Specs: Windows 7 x64
I'm having the most annoying glitch with my recent build, and i can't even conceive where it begins.

Every now and again, sometimes as often as every 10 minutes, all the application first start slowing down, then stop responding completely. Lasts for a couple minutes, then snaps back, like nothing happened.

At first this sounded like a problem with the OS, so i replaced the old XP64 with win7 64, but the problem persists. Now my guess is the hardware, but what? Any ideas would be appreciated

My rig: Phenom II x4 965, Asus m4a79 deluxe, 16(4x4gb) DDR2 800 by GSkill), XFX ATI HD 4850, 3x500gb green hd's, on 700W of power.


See More: Need help isolate Obscure Glitch on new rig

Report •

#1
January 18, 2010 at 13:38:24
Did u test ram with memtest? Http://www.memtest86.com download .Iso v3.5 file 4 windows and burn it.And make sure ram voltage are set according to manufacture don't accept defaults.

Report •

#2
January 18, 2010 at 13:42:27
If you run a performance monitor log from a remote machine you can usually find a place to start looking. The main issue is trying to find the correct metric and how to interpret that metric.

I had a server that had a cache on the scsi drives that seems to conflict with 2003's cache. Took me a day or so to find it.

Perfmon has every tool you might need to get to this area. Might have to use other tools later to get exact but start there.

Memtest can't hurt but some versions are faulty. You'd have to run a few to be somewhat sure. Can't hurt to run ultimate boot cd's list of tools if you want to go that route.

Playing to the angels
Les Paul (1915-2009)


Report •

#3
January 18, 2010 at 14:05:12
@jefro
According to wikipedia memtest stable release, memtest86 v3.5 &memtest86+ v4.00

Report •

Related Solutions

#4
January 18, 2010 at 14:29:31
I am saying just because it is "stable" it may or may not correctly detect an issue. I say that because V4 didn't find a recent fault in some ram. I downleveled to 3.3 and 1.70 and they both found the errors same as W7 disk.

Vista and Windows 7 dvd's also offer a memtest feature.

While the OP is testing might as well try the latest Ultimate boot cd.

http://www.ultimatebootcd.com/forum...

Playing to the angels
Les Paul (1915-2009)


Report •

#5
January 18, 2010 at 16:39:43
I would scan for nasties. Disconnect the computer from the internet and see if the same thing still happens.

Report •

#6
January 18, 2010 at 17:28:47
Can't hurt.

Playing to the angels
Les Paul (1915-2009)


Report •

#7
January 19, 2010 at 06:08:44
Yes, i have tested to see if it was a virus in the past.

Good call on checking the voltage, i'll look into that.

And Jefro, i'm not quite sure on what u'r saying i should try doing? Thanks guys


Report •

#8
January 19, 2010 at 06:20:58
Did you try running while not connected to the internet. If your computer has been comprised the purpose is usually to use the internet for some purpose. If that is the case then the occasional slow downs should stop.

Report •

#9
January 19, 2010 at 12:42:18
I have, i was disconnected from the internet for several weeks, same thing...

Also, note that it's not even a slowdown - it's a complete halt. The whole machine stops responding for some time...


Report •

#10
January 19, 2010 at 12:46:13
OK, that sounds like overheating. Install speedfan and monitor the temperatures. Get speedfan at the link below.

http://download.cnet.com/SpeedFan/3...


Report •

#11
January 19, 2010 at 13:18:01
Speedfan gives some rather obscure results. While nothing is over 60, idle even lower, one of the temps registers as -128 degrees. Another odd thing is that it reports a lot of the voltages completely wrong, even though they look right in bios.Though i'm still not sure if the RAM might be overheating, but my best guess is no, since glitches happen under no heavy load sometimes.

Report •

#12
January 19, 2010 at 13:41:30
The version I linked to is good for Vista so I assumed it would be good for Windows 7.

If your motherboard doesn't have a certain sensor then you can get goofy readings.

Try speedfan watching speedfan constantly by having it set in the tray. See what the temps do when under stress.

What kind of voltage differences are you seeing? If the +12V is dropping under heavy loading that could indicate a faulty power supply. You can't place a load on the computer when in the BIOS screens.


Report •

#13
January 19, 2010 at 14:29:11
no, its wierder than that.
vcore1: 0
vcore2:1.89
+3.3V:1.17
+5v:5.08
+12V:4.8
-12V:3.89
-5v:-1.13
+5v:4.97
Vbat:3.31

Which soundstotally off... And i had the exact same numbers under winxp 64


Report •

#14
January 19, 2010 at 14:54:19
I don't use CPUz it but I think it will monitor the voltages. Get it at the link below.

http://www.cpuid.com/cpuz.php

I just tried speedfan 4.40 on my Windows 7 system and got goofy readings too. Sorry, I thought it should work in 7.


Report •

#15
January 19, 2010 at 17:57:37
Almost can't believe you have correct numbers to even show. Pretty odd.I guess it could report. Use a multimeter maybe to prove.

Performance monitor is a windows application that can record and show almost any OS related metric. That and event viewer and dump logs are places to look for normally software related issues.

Some issues are difficult. Memtest and Ultimate boot cd are testing tools that should half split the issue between software and hardware.

I would look at both software issues and hardware issues. Software may be an easy first step as much of the data might be there to be viewed by event viewer or perfmon.

Secondly I'd consider looking at hardware. Run some of the cpu tools on ubcd and see if it locks up. Run a few memtest versions. See the ultimate boot cd web page for the latest stable or the beta V5 tool list. It is extensive.

Playing to the angels
Les Paul (1915-2009)


Report •

#16
January 21, 2010 at 05:59:16
One thing i haven't not noticed before: during the glitch, the harddrive activity light lights up solid, doesn't flash even a bit.

Could this suggest that something is wrong with hd's?


Report •

#17
January 21, 2010 at 06:02:09
What is the capacity of your hard drive and the Windows partition and how much free space on the Windows partition?

Report •

#18
January 21, 2010 at 06:29:58
i got 3x500gb drives. windows has one hd devoted to it, with most of it unused space. all of my data is on the other two.

Report •

#19
January 21, 2010 at 06:39:34
it really is the only external sign of the glitch. Everything else - the cpu usage, the ram usage, the temps, everything else so far stays normal during the glitch, nothing spikes. but that hd activity light... can't believe i haven't noticed it before...

Report •

#20
January 21, 2010 at 06:48:20
To eliminate the possibility that the drive has a problem I suggest you download a drive fitness utility from the drive manufacturers site and run it.

Report •

#21
January 22, 2010 at 11:19:59
I ran the fitness utility, and all the reports came out clean.

However...

During the testing of two of the drives, the hd activity light was looking healthy, flashing only now and again, but during the testing of the third drive it was once again lit up solid... So it still seems a little fishy to me


Report •

#22
January 22, 2010 at 11:53:23
Are the drives SATA or IDE? If SATA check the data cable for sharp bends and straighten it out if they exist. If IDE change the cable.

Are all three the same brand? Sounds like WD brand?

Did you look at and compare the reports for each one. Also, If SATA there should be a jumper to force a SATA II drive to run in SATA I mode. Be sure that isn't set.


Report •

Ask Question