Fresh XP Pro+SP2 install, System =100% cpu, DPCs are huge!
I have been asked to repair a Windows XP computer with the general symptom of performing very slowly when doing simple operations. The Task Manager showed the System process consuming a large amount of CPU, eg: >85% sustained - even when the computer is doing no work on behalf of the user. This machine is used for dictating into MS Word via Dragon Naturally Speaking, and was very slow and sluggish (10 - 45 second delay) and inaccurate (because the app seemed to eventually give up on attempting to translate speech into text).
I have low level reformatted the hard drive, reinstalled the OS with Win XP Pro and applied SP2. Sometime after installing Mozilla, I happened to run the Task Manager and saw that it is (still) running at high levels of cpu consumption.
One fyi is that the machine has a 250G disk, but install only showed 137G. Apparently this is true for XP < SP1, because XP followed the letter of the standard rather than the actual bitwidth of the disk cable.
Here's what I've done so far:
I have scratched my head, in true WTF fashion.
I have removed all cards from the backplane, on the outside chance that one of them is a culprit. Doing this did not vary the symptom.
I have applied all of the drivers which came with the mobo, a K7SEM from Elite.
I have not reflashed the BIOS.
I have downloaded and run all of the freely downloadable tools from sysinternals.com. AutoRuns shows 3 reasonable things at startup. FileMon shows occasional bursts, but they are few. ProcessMon shows: system ide process: 3%, Interrupts: 31% (Hardware interrupts), DPCs: 46% (Deferred Procedure Calls), System: 21%. (open question: How can I learn more about what is generating these interrupts and DPCs?). RegMon shows there are occasional flurries of activity in the registry when I move a window, but nothing tied to cpu utilization. TcpView shows a few posted listens, and nothing strange. All others show benign activity.
I have posted an email to the Elite MoBo website describing the problem. No response yet.
The HardwareManager was empty but is now populated. start->control panel->system -> Hardware Tab -> Device Manager. The list is empty. (researched this on EE, refd to an MSFT support article on enabling PnP, which did not vary the symptom). However, when I ran compmgmt.msc from the command prompt, the list in the device manager was *not* empty and looked properly populated. Also, when I just redisplayed the Device Manager via control panel -> system, the list is now populated. This is a variation.
I have researched through the first several pages of results from EE when searching for some combination of high cpu consumption and the system process. Some suspect infections. This is a fresh install + sp2, but I have scanned the computer using PC-Cillin and it found nothing.
WindowsUpdate was run and the dl options chosen are: GDI+ detection tool, .NET framework, NIC update, AUdio Driver update, Video update. (all core/ crucial things, any of which can cause this high cpu consumption). Installing. Done. No change.
Researched into what are DPCs, and what could be causing so many interrupts to cause the percentages to be, shown using ProcessExplorer, 44% and 31%. Btw: PE shows the System process is 20% utilzation.
Is there a way to cause the TaskMgr to show unreliable consumption numbers, or for it to cause high consumption on it's own?
*Any* thoughts on this?
I have low level reformatted the hard drive, reinstalled the OS with Win XP Pro and applied SP2. Sometime after installing Mozilla, I happened to run the Task Manager and saw that it is (still) running at high levels of cpu consumption.
One fyi is that the machine has a 250G disk, but install only showed 137G. Apparently this is true for XP < SP1, because XP followed the letter of the standard rather than the actual bitwidth of the disk cable.
Here's what I've done so far:
I have scratched my head, in true WTF fashion.
I have removed all cards from the backplane, on the outside chance that one of them is a culprit. Doing this did not vary the symptom.
I have applied all of the drivers which came with the mobo, a K7SEM from Elite.
I have not reflashed the BIOS.
I have downloaded and run all of the freely downloadable tools from sysinternals.com. AutoRuns shows 3 reasonable things at startup. FileMon shows occasional bursts, but they are few. ProcessMon shows: system ide process: 3%, Interrupts: 31% (Hardware interrupts), DPCs: 46% (Deferred Procedure Calls), System: 21%. (open question: How can I learn more about what is generating these interrupts and DPCs?). RegMon shows there are occasional flurries of activity in the registry when I move a window, but nothing tied to cpu utilization. TcpView shows a few posted listens, and nothing strange. All others show benign activity.
I have posted an email to the Elite MoBo website describing the problem. No response yet.
The HardwareManager was empty but is now populated. start->control panel->system -> Hardware Tab -> Device Manager. The list is empty. (researched this on EE, refd to an MSFT support article on enabling PnP, which did not vary the symptom). However, when I ran compmgmt.msc from the command prompt, the list in the device manager was *not* empty and looked properly populated. Also, when I just redisplayed the Device Manager via control panel -> system, the list is now populated. This is a variation.
I have researched through the first several pages of results from EE when searching for some combination of high cpu consumption and the system process. Some suspect infections. This is a fresh install + sp2, but I have scanned the computer using PC-Cillin and it found nothing.
WindowsUpdate was run and the dl options chosen are: GDI+ detection tool, .NET framework, NIC update, AUdio Driver update, Video update. (all core/ crucial things, any of which can cause this high cpu consumption). Installing. Done. No change.
Researched into what are DPCs, and what could be causing so many interrupts to cause the percentages to be, shown using ProcessExplorer, 44% and 31%. Btw: PE shows the System process is 20% utilzation.
Is there a way to cause the TaskMgr to show unreliable consumption numbers, or for it to cause high consumption on it's own?
*Any* thoughts on this?
ASKER
Mirthle,
Thank you for responding. Unfortunately, it appears that I was not clear on all of the problem isolation that has been done so far, so here goes ...
The top processes shown by the Task Manager, ordered by CPU consumption is:
System cpu=69% CpuTime=6:19 MemUsed=212K PageFaults=23,061 I/Oread=82 I/Owrite=370 I/Oother=18,521
System Idle Process cpu=14% CpuTime=1:02 MemUsed=16K ... 0
TaskMgr.exe cpu=9% CpuTime=0:00 MemUsed=4,200K PageFaults=1,078
others are low of 0 cpu consumption.
(Mozilla.exe has cpu=0% cputime=1:24 MemUsed=42,460, ..., so that isn't it)
Using the Process Explorer from the bright gnomes at SysInternals:
DPCs cpu=44% Deferred Procedure Calls
Interrupts cpu=31% Hardware Interrupts
System cpu=19%
Idle cpu=4%
Procexp cpu=2% SysInternals Process Explorer
Since this is a totally fresh installation, and the high cpu consumption symptom was discovered 20-30 mins after the installation was complete, the window of opportunity for a nasty to creep in is small, albeit possible.
I have done the msconfig thing and eliminated the startup of all things in the startup path, and the symptom persists.
I have booted in safe mode, and the symptom of high cpu consumption still persists.
Currently, I'm in the process of turning things off on the mobo, to see if I can vary the symptom.
Addl history: I talked w my customer, and he recalls that this problem started when the new (250G maxtor) disk was installed to replace the old (40G Western Digital) disk. This person is definitely nontechnical, and does not know how to observe system consumptions, so it is possible that this high cpu consumption problem occurred well before I arrived on scene. But he tends to have a good grasp on the recollection of history and what people have said.
Open questions:
Any suggs on how to either study what/ where the DPCs are originating from, or in better problem isolation techniques to either learn more about the problem or discover a functional work around.
Any recommendations on tools to use the system to measure itself?
Thank you for responding. Unfortunately, it appears that I was not clear on all of the problem isolation that has been done so far, so here goes ...
The top processes shown by the Task Manager, ordered by CPU consumption is:
System cpu=69% CpuTime=6:19 MemUsed=212K PageFaults=23,061 I/Oread=82 I/Owrite=370 I/Oother=18,521
System Idle Process cpu=14% CpuTime=1:02 MemUsed=16K ... 0
TaskMgr.exe cpu=9% CpuTime=0:00 MemUsed=4,200K PageFaults=1,078
others are low of 0 cpu consumption.
(Mozilla.exe has cpu=0% cputime=1:24 MemUsed=42,460, ..., so that isn't it)
Using the Process Explorer from the bright gnomes at SysInternals:
DPCs cpu=44% Deferred Procedure Calls
Interrupts cpu=31% Hardware Interrupts
System cpu=19%
Idle cpu=4%
Procexp cpu=2% SysInternals Process Explorer
Since this is a totally fresh installation, and the high cpu consumption symptom was discovered 20-30 mins after the installation was complete, the window of opportunity for a nasty to creep in is small, albeit possible.
I have done the msconfig thing and eliminated the startup of all things in the startup path, and the symptom persists.
I have booted in safe mode, and the symptom of high cpu consumption still persists.
Currently, I'm in the process of turning things off on the mobo, to see if I can vary the symptom.
Addl history: I talked w my customer, and he recalls that this problem started when the new (250G maxtor) disk was installed to replace the old (40G Western Digital) disk. This person is definitely nontechnical, and does not know how to observe system consumptions, so it is possible that this high cpu consumption problem occurred well before I arrived on scene. But he tends to have a good grasp on the recollection of history and what people have said.
Open questions:
Any suggs on how to either study what/ where the DPCs are originating from, or in better problem isolation techniques to either learn more about the problem or discover a functional work around.
Any recommendations on tools to use the system to measure itself?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
I agree with Mirthle. It sounds solely like a software related issue, considering you just comleted a clean install of the OS. But i hardly doubt that Mozilla would be the cause of this problem. No doubt nothing is impossible. I personally have never installed a program that caused abnormal high system resource usage. If a software is to blame for this problem more than likely uninstalling it and reinstalling it NOT fix the problem. In my book, you will have to get rid of the software all the together, and find a subsistute for it.Spyware could also be the culprit. In certain sites you visit, they do nothing but install that lovely spyware to search your entire computer for information about you transmit back to there servers where they file it, and then sell it to every party avaiable. Anyways, perhaps performing maintance on your computer could prove benefical to your problem (disk cleanup, disk defragment, scandisk, chkdsk, SFC, registry cleaner, adware and spyware removal). If you wish to do this I have a list of programs ( that wont kill resources ;) and will help you accomplish the maintance every computer needs.
Spyware Blaster- This program will stop adware and spyware from downloading to your computer to begin with, so you dont have to deal with it.
www.javacoolsoftware.com/spywareblaster.html
RegSupreme - This program will help clean up your systems registry
www.macecraft.com
Spybot Search and Destroy - This program will search your computer for installed spyware, etc programs (its one of SpywareBlasters backups for the stuff it misses)
security.kolla.de
Ad-Aware SE Personal - This program performs just like Spybot, except it looks for files it doesn't, for a broader range of what files are being looked for.
www.lavasoftusa.com/software/adaware/
Diskeeper 8.0 - This is a disk defragment program that I just love greatly. I honesly believe its worth the money to but it. But if you want to use the trial, no problem. Just make sure that when you use it, under the settings, that you change the disk defragment priority level to its highest (i think you can do that in the trial).
This is all that I do for computer maintance, and it works great for me.
Give it try, and send me some feedback on whats happening.
Spyware Blaster- This program will stop adware and spyware from downloading to your computer to begin with, so you dont have to deal with it.
www.javacoolsoftware.com/spywareblaster.html
RegSupreme - This program will help clean up your systems registry
www.macecraft.com
Spybot Search and Destroy - This program will search your computer for installed spyware, etc programs (its one of SpywareBlasters backups for the stuff it misses)
security.kolla.de
Ad-Aware SE Personal - This program performs just like Spybot, except it looks for files it doesn't, for a broader range of what files are being looked for.
www.lavasoftusa.com/software/adaware/
Diskeeper 8.0 - This is a disk defragment program that I just love greatly. I honesly believe its worth the money to but it. But if you want to use the trial, no problem. Just make sure that when you use it, under the settings, that you change the disk defragment priority level to its highest (i think you can do that in the trial).
This is all that I do for computer maintance, and it works great for me.
Give it try, and send me some feedback on whats happening.
Nevermind, I've been typing for to long to update my wording. I would still perform the maintance that I listed.
ASKER
Whoa! Two repsonders!! Cool!!!
On the general topic that nothing is ever impossible, I will uninstall Mozilla to try to cary the symptom, and will report back status.
I have not completed the selective turning off of mobo components in an attempt to vary the symptom. More to come on this ... My current operating assumption is that there is a bad element on the mobo someplace (you can't play with mud without getting dirty. ehem...), because the machine does function sans for the wierd high cpu consumption. For personal gratification, I'd love to identify the root problem, and then to turn that thing off so the system can skip merrily along. But from personal experience, this may be one of those problems where it is best to declare victory sooner rather than later, and then move on.
In the tration of the Bus Driver's Holiday, who's idea of a vacation is to take a Winnebago camper across the country, I am in the process of formatting the second half of the 250G hard drive. This second half was invisible to XP at install time because of a failure/ built in limitation by XP prior to SP1 to support LBA. SP2 was applied, so no surprise, it's there now and needs to be formatted in order to be used.
I have already run the Maxtor diagnostic utility on the disk, and it passed both the quick and the long test. I ran a quick low level format, which writes zeroes to the first 300MB of the disk, and that completed successfully, albeit more slowly than I would have expected (about 1H). I also ran a full low level formatting of the disk, starting it last night arounf 730P, it wasn't done when I turned in for the day, but was still running the next morning. My gross recollection is that the reformatting completed about 273/475 of the disk. This is a 7200 rpm 8M cache 250G disk with a 133mbps data path. My quick estimate suggests that this formatting took longer than it should, and casts a bloody finger of suspicion toward the elements in the disk path. Mea Culpa for not being more forthcoming!
I have already run Spybot on the machine before the original posting, on the outside chance that some combination of adware and viri (viruses?) may have opportunistically intected the machine. Son of Mea Culpa!
On doing addl maintenance, eg: registry cleanup and further adware/ spyware scrubbing and maintenance on an otherwise freshly minted machine, I don't think that is a fruitful path at this time. But those are good generalist tools to keep in one's bag of tricks.
On not knowing of processes that just cause consumption to be performed on their behalf, look at the sasser virus. It forks multiple threads which each generate a random IP address, each agressively attempting a connection, regardless of whether the network is up or not. The cpu consumption of the originating process is tiny, and is effectively masked by the high levels of cpu and other resource consumption by services.exe. The machine is so sluggish and tends to want to reboot spontaneously as another diversion, this makes the job of problem isolation difficult. The clear indicator of what is going on is to run "netstat -o" at the command prompt, to display open connections. Bingo!
I'm still really curious how to study the DPC's more better, (sp = ;-) ) I'm hoping that a thing exists that will quickly and easily provide a finer granularity of insight into the source that is originating the DPC's, without getting into something as deep as SoftIce.
I'm still really interested in experiences people have had using different hw/ sw diagnostic solutions, both good and bad. I'm not aaverse to buying something, but I do have a preference to a) hear about people's experiences, and b) to try it out. I buy a huge number of (hw, sw, hand) tools for the well meaning intention of reusing them, but somehow it just doesn't work that way.
On the general topic that nothing is ever impossible, I will uninstall Mozilla to try to cary the symptom, and will report back status.
I have not completed the selective turning off of mobo components in an attempt to vary the symptom. More to come on this ... My current operating assumption is that there is a bad element on the mobo someplace (you can't play with mud without getting dirty. ehem...), because the machine does function sans for the wierd high cpu consumption. For personal gratification, I'd love to identify the root problem, and then to turn that thing off so the system can skip merrily along. But from personal experience, this may be one of those problems where it is best to declare victory sooner rather than later, and then move on.
In the tration of the Bus Driver's Holiday, who's idea of a vacation is to take a Winnebago camper across the country, I am in the process of formatting the second half of the 250G hard drive. This second half was invisible to XP at install time because of a failure/ built in limitation by XP prior to SP1 to support LBA. SP2 was applied, so no surprise, it's there now and needs to be formatted in order to be used.
I have already run the Maxtor diagnostic utility on the disk, and it passed both the quick and the long test. I ran a quick low level format, which writes zeroes to the first 300MB of the disk, and that completed successfully, albeit more slowly than I would have expected (about 1H). I also ran a full low level formatting of the disk, starting it last night arounf 730P, it wasn't done when I turned in for the day, but was still running the next morning. My gross recollection is that the reformatting completed about 273/475 of the disk. This is a 7200 rpm 8M cache 250G disk with a 133mbps data path. My quick estimate suggests that this formatting took longer than it should, and casts a bloody finger of suspicion toward the elements in the disk path. Mea Culpa for not being more forthcoming!
I have already run Spybot on the machine before the original posting, on the outside chance that some combination of adware and viri (viruses?) may have opportunistically intected the machine. Son of Mea Culpa!
On doing addl maintenance, eg: registry cleanup and further adware/ spyware scrubbing and maintenance on an otherwise freshly minted machine, I don't think that is a fruitful path at this time. But those are good generalist tools to keep in one's bag of tricks.
On not knowing of processes that just cause consumption to be performed on their behalf, look at the sasser virus. It forks multiple threads which each generate a random IP address, each agressively attempting a connection, regardless of whether the network is up or not. The cpu consumption of the originating process is tiny, and is effectively masked by the high levels of cpu and other resource consumption by services.exe. The machine is so sluggish and tends to want to reboot spontaneously as another diversion, this makes the job of problem isolation difficult. The clear indicator of what is going on is to run "netstat -o" at the command prompt, to display open connections. Bingo!
I'm still really curious how to study the DPC's more better, (sp = ;-) ) I'm hoping that a thing exists that will quickly and easily provide a finer granularity of insight into the source that is originating the DPC's, without getting into something as deep as SoftIce.
I'm still really interested in experiences people have had using different hw/ sw diagnostic solutions, both good and bad. I'm not aaverse to buying something, but I do have a preference to a) hear about people's experiences, and b) to try it out. I buy a huge number of (hw, sw, hand) tools for the well meaning intention of reusing them, but somehow it just doesn't work that way.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Guys,
This *does* happen while in Safe Mode... It's reliable, unavoidable, and inescapable.
Any suggs on tools of insight into problems that may be HW related?
cpe
This *does* happen while in Safe Mode... It's reliable, unavoidable, and inescapable.
Any suggs on tools of insight into problems that may be HW related?
cpe
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
I agree.. if it is the hard drive itself, then pulling hardware out will certainly be a moot point.. :)
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Gentlemen,
First, I want to tell you that it feels really good being with smart people! Thank you!! The "Back to Basics" is an excellent reminder, similar to one of my favorite troubleshooting mantras: "When you hear the sounds of hooves but don't see what's running by, think Horses, not Zebra."
Here are some findings:
1) I swapped the "old" 40G western digital harddrive with the "new" maxtor 6y250P0. The "old" drive had XP Home on it originally but was transferred over to the "new" hard drive via Ghost, ant it still (surprise, surpeise) bootable. The high cpu consumption by the System process exists when using the "old" drive. This rules out the different/ bigger hard drive as the culprit. One down.
fyi: the 250G drive was quite warm to the touch when I removed it. The machine was on over night, but the machine was not responsive when I moved the rodent or pressed the shift key, suggesting that it had slunk into a non-operational hibernate mode, and required graceless shutdown A (hold the power button for 8 seconds. B would be yanking the power via cord removal).
minutae: The temperature was about as warm as freshly brewed coffee poured into a standard (unwashed) ceramic mug, and touched about 1-2 mins later. Not hot, but not something you really want to hold for more than a minute or so. The drive had 1 inch of air space top and bottom, and was mounted into the cage using 3 screws, both providing some thermal transfer. This heat will be an aging agent to the drive, but this isn't a problem now.
2) There is no Mozilla et. al. installed on the "old" 40G drive, so this rules that out. Two down.
3) Within the Device manager, I have disabled all nonessential devices. These include: sound, parallel, serial, gameport, the 2 cd drives, usb. I did not disable things like: the system timer, keyboard, rodent. Note: the Task Manager did not show any change in cpu consumption while the devices were being turned off.
Restarted, logged in, and ran the Task Manager, the Processes tab showing that the System process is still consuming beaucoup cpu, but it's cruising at 75% rather than 95%. System Idle Process time is cruising at 22%, the Explorer process is cruising at 2-6%. However, CPU Usage reported at the bottom cruises at 96%. Note: the machine isn't being used, other than for me running the Task Manager.
Inspected the Hardware Manager after the restart, to see if the disabled things are indeed still disabled. They are still disabled. Before I rule this one out, I want to reenable these things and remeasure which process is consuming what.
I'll need to report further results later. My wife wants to go out on date night. So even though there is this persistent and intransigent problem, at least some things are going right ...
First, I want to tell you that it feels really good being with smart people! Thank you!! The "Back to Basics" is an excellent reminder, similar to one of my favorite troubleshooting mantras: "When you hear the sounds of hooves but don't see what's running by, think Horses, not Zebra."
Here are some findings:
1) I swapped the "old" 40G western digital harddrive with the "new" maxtor 6y250P0. The "old" drive had XP Home on it originally but was transferred over to the "new" hard drive via Ghost, ant it still (surprise, surpeise) bootable. The high cpu consumption by the System process exists when using the "old" drive. This rules out the different/ bigger hard drive as the culprit. One down.
fyi: the 250G drive was quite warm to the touch when I removed it. The machine was on over night, but the machine was not responsive when I moved the rodent or pressed the shift key, suggesting that it had slunk into a non-operational hibernate mode, and required graceless shutdown A (hold the power button for 8 seconds. B would be yanking the power via cord removal).
minutae: The temperature was about as warm as freshly brewed coffee poured into a standard (unwashed) ceramic mug, and touched about 1-2 mins later. Not hot, but not something you really want to hold for more than a minute or so. The drive had 1 inch of air space top and bottom, and was mounted into the cage using 3 screws, both providing some thermal transfer. This heat will be an aging agent to the drive, but this isn't a problem now.
2) There is no Mozilla et. al. installed on the "old" 40G drive, so this rules that out. Two down.
3) Within the Device manager, I have disabled all nonessential devices. These include: sound, parallel, serial, gameport, the 2 cd drives, usb. I did not disable things like: the system timer, keyboard, rodent. Note: the Task Manager did not show any change in cpu consumption while the devices were being turned off.
Restarted, logged in, and ran the Task Manager, the Processes tab showing that the System process is still consuming beaucoup cpu, but it's cruising at 75% rather than 95%. System Idle Process time is cruising at 22%, the Explorer process is cruising at 2-6%. However, CPU Usage reported at the bottom cruises at 96%. Note: the machine isn't being used, other than for me running the Task Manager.
Inspected the Hardware Manager after the restart, to see if the disabled things are indeed still disabled. They are still disabled. Before I rule this one out, I want to reenable these things and remeasure which process is consuming what.
I'll need to report further results later. My wife wants to go out on date night. So even though there is this persistent and intransigent problem, at least some things are going right ...
Date night, eh? Glad to see you are doing well with that, at least..!! :)
Hard drives get hot. SATA/SCSI drives can run 60+ C and IDE's 45-50 C.. So I would not worry about those temps.
Let us know how the other you find when you get back on it..
FE
Hard drives get hot. SATA/SCSI drives can run 60+ C and IDE's 45-50 C.. So I would not worry about those temps.
Let us know how the other you find when you get back on it..
FE
ASKER
Gentlemen,
Date night was cut short with a phone call from Thing 2 less than a minute after posting my last message, saying that their ETA was in about an hour. So we went on a drive to no where through the mountains, may have seen some Elk, and chatted a bit.
My take away from the SI article on DPC's was that it was interesting, but I'm still ISO some method of pointing a bloody finger of accusation at a culprit that is based on data.
3) Continued. Turn off (practically) everything in the BIOS. (btw: I have already physically removed all (both) cards installed in the PCI slots (usb, modem)). Done that. Booting up. BIOS display presents. Screen goes black and stays black, but the display active light is still green rather than amber. HD activity light is flashing off regularly (sugg: something is wrong). I think I went too far. After about 2 mins of letting the machine timeout and come up, rebooting into the Bios and resetting to the failsafe defaults. Btw: pressing the power button once and quickly took the machine down, sugg: the OS wasn't up yet. Done. Rebooting.
Coming up in safemode. btw: cpu usage = 100%, System cruising at 93%, System idle process = 0%. However, the Task Manager update rate was set to "normal"; setting it to "low" gives different results: System = 74%, Idle = 20%, Task Mgr = 3%, and CSRSS.EXE = 3% (which, according to My Friend Leo, the mnemonics stand for: Client Server Runtime SubSystem, so I'm not too worried about this process being a rogue. But I kinda wish I had a CRC32 or MD5 to compare my csrss.exe against something "known good").
Btw: in each of the following runs, it takes about 2 mins for the processes launched at *each* startup+login time to quiesce.
Mobo Manuf = Elite.
Bios = Award, k7SEM v1.2b 1/27/2003
Restarting to get into the Bios again.
Leaving Std CMOS features alone.
Gently modify the Advanced Bios features. Save, exit, restart. OS=up. Does high CPU consumption persist? Yes. Get into the bios.
Leaving Advanced Chipset Features alone.
Integrated Peripherals. Disable USB, Onboard PCI LAN. Save, exit, restart. OS=up. Does it? Yes. Back into bios.
Power Mgmt setup. Disable: ACPI function, Video Off. Save, exit, restart. OS=Dark. Too far. Undo this and continue. Done.
PnP/PCI Configurations. Nothing interesting to change.
PC Health status: Enable shutdown temp to 60C/140F. Save, exit, restart. OS=Safe Mode. Does it? Yes. More Bios.
Frequency/ Voltage control. Not interesting.
No further make-able changes.
In the Device Manager, reenabled all of the devices I disabled before, and kept the Task Manager up and running to show whether reenabling anything varies anything. It didn't. With the TM update freq = low, System = 78%, Idle = 18%, Explorer = 3%, Taskmgr = 1%.
I think I've ruled out whether the cause of high CPU consumption is caused by something that can be turned off on the Bios *or* the OS. 3 down.
Some thoughts:
4) Since the "new" drive was duped from the "old" drive via Ghost, which implicitely includes complete fidelity, and the high cpu consumption exists when either disk is installed, I have not yet ruled out the possiblity that the problem is SW/OS related and not HW related. If there was a SW/OS issue before, it makes sense that they symptom would follow to the new disk. Note: The new HD was wiped and reformatted and reinstalled fresh with the OS and SP2, so there is 0 opportunity for a nasty to carry forward on the disk. Tomorrow is going to be "take the computer to the HW Guru" day, unless this problem is divided in half better and suggests (implicates) that (some) SW is the root cause.
Unless the HP-like enlightenment of the "Ahah!" in the shower happens, like, now, this box is going into the shop first thing tomorrow morning.
5) An alternative is to temporarily return the box to my customer, and let him catch up on his work for few days. Since the box has had this problem before, and he was able to get some meaningful work done with is while it was pegged, returning isn't necessarily a bad thing to do. However, returning the box to him without solving both of the major motivators for rebuilding the box (dictation via dragon naturally speaking is so slow and inaccurate as to be nonresponsive, and cpu=pegged), may be recieved poorly.
Open question: Any recommendations or experiences for tools of troubleshooting enlightenment, free or $.
Date night was cut short with a phone call from Thing 2 less than a minute after posting my last message, saying that their ETA was in about an hour. So we went on a drive to no where through the mountains, may have seen some Elk, and chatted a bit.
My take away from the SI article on DPC's was that it was interesting, but I'm still ISO some method of pointing a bloody finger of accusation at a culprit that is based on data.
3) Continued. Turn off (practically) everything in the BIOS. (btw: I have already physically removed all (both) cards installed in the PCI slots (usb, modem)). Done that. Booting up. BIOS display presents. Screen goes black and stays black, but the display active light is still green rather than amber. HD activity light is flashing off regularly (sugg: something is wrong). I think I went too far. After about 2 mins of letting the machine timeout and come up, rebooting into the Bios and resetting to the failsafe defaults. Btw: pressing the power button once and quickly took the machine down, sugg: the OS wasn't up yet. Done. Rebooting.
Coming up in safemode. btw: cpu usage = 100%, System cruising at 93%, System idle process = 0%. However, the Task Manager update rate was set to "normal"; setting it to "low" gives different results: System = 74%, Idle = 20%, Task Mgr = 3%, and CSRSS.EXE = 3% (which, according to My Friend Leo, the mnemonics stand for: Client Server Runtime SubSystem, so I'm not too worried about this process being a rogue. But I kinda wish I had a CRC32 or MD5 to compare my csrss.exe against something "known good").
Btw: in each of the following runs, it takes about 2 mins for the processes launched at *each* startup+login time to quiesce.
Mobo Manuf = Elite.
Bios = Award, k7SEM v1.2b 1/27/2003
Restarting to get into the Bios again.
Leaving Std CMOS features alone.
Gently modify the Advanced Bios features. Save, exit, restart. OS=up. Does high CPU consumption persist? Yes. Get into the bios.
Leaving Advanced Chipset Features alone.
Integrated Peripherals. Disable USB, Onboard PCI LAN. Save, exit, restart. OS=up. Does it? Yes. Back into bios.
Power Mgmt setup. Disable: ACPI function, Video Off. Save, exit, restart. OS=Dark. Too far. Undo this and continue. Done.
PnP/PCI Configurations. Nothing interesting to change.
PC Health status: Enable shutdown temp to 60C/140F. Save, exit, restart. OS=Safe Mode. Does it? Yes. More Bios.
Frequency/ Voltage control. Not interesting.
No further make-able changes.
In the Device Manager, reenabled all of the devices I disabled before, and kept the Task Manager up and running to show whether reenabling anything varies anything. It didn't. With the TM update freq = low, System = 78%, Idle = 18%, Explorer = 3%, Taskmgr = 1%.
I think I've ruled out whether the cause of high CPU consumption is caused by something that can be turned off on the Bios *or* the OS. 3 down.
Some thoughts:
4) Since the "new" drive was duped from the "old" drive via Ghost, which implicitely includes complete fidelity, and the high cpu consumption exists when either disk is installed, I have not yet ruled out the possiblity that the problem is SW/OS related and not HW related. If there was a SW/OS issue before, it makes sense that they symptom would follow to the new disk. Note: The new HD was wiped and reformatted and reinstalled fresh with the OS and SP2, so there is 0 opportunity for a nasty to carry forward on the disk. Tomorrow is going to be "take the computer to the HW Guru" day, unless this problem is divided in half better and suggests (implicates) that (some) SW is the root cause.
Unless the HP-like enlightenment of the "Ahah!" in the shower happens, like, now, this box is going into the shop first thing tomorrow morning.
5) An alternative is to temporarily return the box to my customer, and let him catch up on his work for few days. Since the box has had this problem before, and he was able to get some meaningful work done with is while it was pegged, returning isn't necessarily a bad thing to do. However, returning the box to him without solving both of the major motivators for rebuilding the box (dictation via dragon naturally speaking is so slow and inaccurate as to be nonresponsive, and cpu=pegged), may be recieved poorly.
Open question: Any recommendations or experiences for tools of troubleshooting enlightenment, free or $.
Other than starting to swap out the primary components now, I am at a loss. With a new OS, new hard drive, and disconnecting all unnecessary hardware, it is becoming evident that the problem lies in the System board, memory, or one of the attached devices (keyboard/mouse)... Tough call, this one..
FE
FE
Any chance you can put a fresh w98 on the 40gb drive? Run the resource meter to see if the same problem exists. That would definately tell you whether its an OS/software problem.
ASKER
Gentlemen,
The keyboard and rodent have been changed, using mine rather than the customers, since I only need to take the box and nothing else.
Sometimes sleep is indeed restful, as the clarity of a couple of different avenues are now apparent.
First is to complete the SW installation on this box. Even if the mobo is changed, this is something that needs to be done before returning it to the customer anyway. So I'm in the process of doing that now.
Another is something which I've discussed with my customer, which is to loan him a computer, but it turns out that I need to have one to loan him first. Problem is that the notebook I wanted to loan him is my wifes, she'll be away for several days, and she needs to take it with her. Another problem is the desktop I was going to loan him has been my linux box, so the XP hd has been removed for a long time. Installing that disk results in the task bar or start menu is not presented after logging into with a normal bootup, and there are lots of game icons on the backdrop which someone obviously installed. Starting up in safe mode and logging in does present a task bar. Since it's been a while since I last updated the machine, I decided to apply SP2 in an intended panacea solution to all problems. However, during the installation, the install process reported that the Cryptographic service needed to be running, and exited when the Bummer button was clicked. (segue: my usage convention when I need to present bad news to a user, but I don't want to give them a one and only bad choice, so I like to label the button Bummer rather than OK.) Now, the display goes black right after the XP splash screen, the monitor light stays green and not than amber, and there is zero disk activity. I've lost the original XP Home install disk, so I can't do a healing installation recovery. I do have an unused copy of XP Pro that I could use to reinstall with, but I'd really rather not use that just yet. (any suggs are welcome). (open question: is there a method of creating an XP OS install disk from the installed upon computer?)
Another is to offer my customer his reinstalled machine as a temporary solution for him to get over the hump of being backlogged with work while his computer has been in repair (with me). This has the advantage of quickly getting to him what is essentially a loaner, even though it is his machine, because this computer does still need to go into the shop for HW work.
The keyboard and rodent have been changed, using mine rather than the customers, since I only need to take the box and nothing else.
Sometimes sleep is indeed restful, as the clarity of a couple of different avenues are now apparent.
First is to complete the SW installation on this box. Even if the mobo is changed, this is something that needs to be done before returning it to the customer anyway. So I'm in the process of doing that now.
Another is something which I've discussed with my customer, which is to loan him a computer, but it turns out that I need to have one to loan him first. Problem is that the notebook I wanted to loan him is my wifes, she'll be away for several days, and she needs to take it with her. Another problem is the desktop I was going to loan him has been my linux box, so the XP hd has been removed for a long time. Installing that disk results in the task bar or start menu is not presented after logging into with a normal bootup, and there are lots of game icons on the backdrop which someone obviously installed. Starting up in safe mode and logging in does present a task bar. Since it's been a while since I last updated the machine, I decided to apply SP2 in an intended panacea solution to all problems. However, during the installation, the install process reported that the Cryptographic service needed to be running, and exited when the Bummer button was clicked. (segue: my usage convention when I need to present bad news to a user, but I don't want to give them a one and only bad choice, so I like to label the button Bummer rather than OK.) Now, the display goes black right after the XP splash screen, the monitor light stays green and not than amber, and there is zero disk activity. I've lost the original XP Home install disk, so I can't do a healing installation recovery. I do have an unused copy of XP Pro that I could use to reinstall with, but I'd really rather not use that just yet. (any suggs are welcome). (open question: is there a method of creating an XP OS install disk from the installed upon computer?)
Another is to offer my customer his reinstalled machine as a temporary solution for him to get over the hump of being backlogged with work while his computer has been in repair (with me). This has the advantage of quickly getting to him what is essentially a loaner, even though it is his machine, because this computer does still need to go into the shop for HW work.
ASKER
JC,
Resource Meter in Win98??? Say more ...
And on putting win98 on that box, I'd rather use one of my spare HD's for the testing, and keep his "old" drive pristine.
cpe
Resource Meter in Win98??? Say more ...
And on putting win98 on that box, I'd rather use one of my spare HD's for the testing, and keep his "old" drive pristine.
cpe
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
JC,
Interesting tips! Number two looks like it will be a real bacon saver *when* the mobo is swapped, since my strong preference is to avoid an OS/SW reinstall. Not a bad Res, either ...
This morning, I brought the box down to the Guru for further analysis. Unfortunately, he/ they are busy and I suspect that his ~4H diagnosis time estimate may slip a bit.
In the mean time, I have an XP box to build as plan C (A=take back computer, B=replace mobo and respond with adjustments, C=provide a loaner box).
Open question: any suggs on making a multiboot machine using grub. The box I will be using already has Redhat Enterprise Linux version 3, and I'd prefer not have to completely repave over that, especially since it has a live subscription for updates. From past research, I know that grub can be configured to do the disk partition fake outs that I understand XP needs to see (eg: it wants or needs to be the first partition, and grub can present that), but I haven't actually configured grub for multibooting of different OS's on different partitions. Any suggs on grub configuration?
Interesting tips! Number two looks like it will be a real bacon saver *when* the mobo is swapped, since my strong preference is to avoid an OS/SW reinstall. Not a bad Res, either ...
This morning, I brought the box down to the Guru for further analysis. Unfortunately, he/ they are busy and I suspect that his ~4H diagnosis time estimate may slip a bit.
In the mean time, I have an XP box to build as plan C (A=take back computer, B=replace mobo and respond with adjustments, C=provide a loaner box).
Open question: any suggs on making a multiboot machine using grub. The box I will be using already has Redhat Enterprise Linux version 3, and I'd prefer not have to completely repave over that, especially since it has a live subscription for updates. From past research, I know that grub can be configured to do the disk partition fake outs that I understand XP needs to see (eg: it wants or needs to be the first partition, and grub can present that), but I haven't actually configured grub for multibooting of different OS's on different partitions. Any suggs on grub configuration?
ASKER
Umm... Replace "JC" with "FE" on the last comment. Maxima Mea Culpa!!!
:) NP, Chris..
No hands-on experience with Grub.. but we have discussed it a lot here in the threads.. do a search and you will find plenty of info on this..
FE
No hands-on experience with Grub.. but we have discussed it a lot here in the threads.. do a search and you will find plenty of info on this..
FE
In win98 the resource meter should be in your c:\windows folder. Its called RSRCMTR.EXE. Double click to run.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Gentlemen,
There is news. The HW Guru has confirmed my findings - it's a mobo problem. But he was able to give further insight into the problem: it's the keyboard controller. Unfortunately the keyboard cant be turned off in the Bios, and using a USB keyboard made no difference. He also confimed my findings about getting a replacement mobo, Elite not only doesn't make the board anymore, but added that they don't have an upgrade to replace it. I asked him to look further into possibly getting one from the used market, in the hope that he has superior searching skills and equipment sources than me.
The solutions I see for this computer and customer are: A) Live with the high cpu consumption, B) find an exact replacement mobo on the used market, C) replace the mobo with something different, and deal with the device difference issues when they arise, D) buy a Dell.
Open question: Anyone know how or where to get an Elite K7SEM mobo?
cpe
There is news. The HW Guru has confirmed my findings - it's a mobo problem. But he was able to give further insight into the problem: it's the keyboard controller. Unfortunately the keyboard cant be turned off in the Bios, and using a USB keyboard made no difference. He also confimed my findings about getting a replacement mobo, Elite not only doesn't make the board anymore, but added that they don't have an upgrade to replace it. I asked him to look further into possibly getting one from the used market, in the hope that he has superior searching skills and equipment sources than me.
The solutions I see for this computer and customer are: A) Live with the high cpu consumption, B) find an exact replacement mobo on the used market, C) replace the mobo with something different, and deal with the device difference issues when they arise, D) buy a Dell.
Open question: Anyone know how or where to get an Elite K7SEM mobo?
cpe
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
FE,
Opinions are not intrinsically offensive, especially in a society that values and champions Free Speech, and they are especially valuable when those shared opinions describe first hand experiences.
I realize that I am now safely off the original topic, and I have selfishly enjoyed getting here, but what are your thoughts on Asus? I built a computer about 2Y ago using a dual athlon mobo from Asus, and it has been about as trouble free as turning on water from the tap.
cpe
Opinions are not intrinsically offensive, especially in a society that values and champions Free Speech, and they are especially valuable when those shared opinions describe first hand experiences.
I realize that I am now safely off the original topic, and I have selfishly enjoyed getting here, but what are your thoughts on Asus? I built a computer about 2Y ago using a dual athlon mobo from Asus, and it has been about as trouble free as turning on water from the tap.
cpe
Asus also is a good MB.. I had a problem a few years ago with their website and the support they provided there (hard to locate specific drivers, etc...) but it has improved dramatically since. So, I would place their MB's in the good column.. :)
Good luck, and thanks..
FE
FE
Asus have a good reputation, which is why I have bought mine. Its an Asus A7N8X-Deluxe revision 2. Its has had to be replaced twice in the last year, each time taking several weeks for Asus to get a replacement to me.
I think I would follow the advice to get an MSI board. I have one of those and its been great. Gigabyte are pretty good too and maybe Abit.
I think I would follow the advice to get an MSI board. I have one of those and its been great. Gigabyte are pretty good too and maybe Abit.
Ooooh - totally on an aside. How did the hardware guru manage to find out that it was the keyboard controller? I'd love to know how to diagnose a mobo that precisely.
ASKER
Mirthle,
Thank you also for the mobo experiences!
Btw: I checked on ebay for a replacement mobo, and there are 3, one of which has a processor (not needed, but I might infer that the board might be a higher quality because it hasn't been stripped yet), and two others. The Guru advises biasing towards local and personal sources, on the grounds of his experience in quality (he may also just prefer to shop locally, too).
Open question: Anyone know how MSFT handles the case where a fresh install has been activated on a machine that needs it's mobo has had to be changed (small change, radical change) because it failed?
On fine resolution of the problem, he used a software tool called QuickTech Pro from Ultra-X.com. The symptom he got suggested a possible bad keyboard, but he called the manuf for clarification, and they said that the diagnostics are written in such a way that this keyboard related symptom is indicative of a bad keyboard controller.
The tool contains it's own OS, so it doesn't depend on the already installed and potentially suspect OS, which may cause or allow the underlying to wiggle while under test. The tool has an interesting and large collection of tests that it can run.
As a geek for hire, I'm always interested in tools that make me better and more self sufficient, and I will be looking into this further. However, the Guru has a chink in his armor from a past service with a computer that wouldn't start after it was struck by lightning. As he removed boards, the computer booted progrssively better, until 3 were identified as toast. He replaced the three and returned he computer as Fixed. Unfortunately, he missed a malfunctioning parallel connector, revealed to my customer when gobbledygook was printed consistently, even when copying straight from the keyboard to the parallell port directly. If this was known, then the need and approval for the 3 replacement boards would have been different. I'm not sure what the mix is between a design limitation and age of the tool versus human oversight.
I/ we all try to be as precise and methodical and thoughtful as possible, but hindsight sometimes does show a missed opportunity in achieving greatness. Learning, what a concept...
Open question: Any recommendations or experiences for tools of troubleshooting enlightenment, free or $.
Thank you also for the mobo experiences!
Btw: I checked on ebay for a replacement mobo, and there are 3, one of which has a processor (not needed, but I might infer that the board might be a higher quality because it hasn't been stripped yet), and two others. The Guru advises biasing towards local and personal sources, on the grounds of his experience in quality (he may also just prefer to shop locally, too).
Open question: Anyone know how MSFT handles the case where a fresh install has been activated on a machine that needs it's mobo has had to be changed (small change, radical change) because it failed?
On fine resolution of the problem, he used a software tool called QuickTech Pro from Ultra-X.com. The symptom he got suggested a possible bad keyboard, but he called the manuf for clarification, and they said that the diagnostics are written in such a way that this keyboard related symptom is indicative of a bad keyboard controller.
The tool contains it's own OS, so it doesn't depend on the already installed and potentially suspect OS, which may cause or allow the underlying to wiggle while under test. The tool has an interesting and large collection of tests that it can run.
As a geek for hire, I'm always interested in tools that make me better and more self sufficient, and I will be looking into this further. However, the Guru has a chink in his armor from a past service with a computer that wouldn't start after it was struck by lightning. As he removed boards, the computer booted progrssively better, until 3 were identified as toast. He replaced the three and returned he computer as Fixed. Unfortunately, he missed a malfunctioning parallel connector, revealed to my customer when gobbledygook was printed consistently, even when copying straight from the keyboard to the parallell port directly. If this was known, then the need and approval for the 3 replacement boards would have been different. I'm not sure what the mix is between a design limitation and age of the tool versus human oversight.
I/ we all try to be as precise and methodical and thoughtful as possible, but hindsight sometimes does show a missed opportunity in achieving greatness. Learning, what a concept...
Open question: Any recommendations or experiences for tools of troubleshooting enlightenment, free or $.
Guess I missed the part about what CPU, and memory were in this thing
Sorry for missing your comments ChrisEddy, for some reason I did not get notified about your post.
I like your style, nothing wrong with trying to learn how to be better at diagnosing computer hardware :) Thanks for the info on the tool used to identify the error, I too will look into that to see if it will be of use to me.
I like your style, nothing wrong with trying to learn how to be better at diagnosing computer hardware :) Thanks for the info on the tool used to identify the error, I too will look into that to see if it will be of use to me.
Switch Off ACPI in the Bios and rebuild the workstation.
Problem should be sorted!
Problem should be sorted!
I have just experienced a problem with CPU spiking to 100% every 6 or 7 seconds. I am running XP Pro and installed SP2 about a month ago. This machine has two network cards, one for cable connection and one to connect to a hub. Since we are moving I disconnected my little network and am just using the XP machine as a standalone. That's when the problem started, but not immediately. It first occurred on saturday morning and I rebooted and problem went away. When I shut down and then started up again later that day the problem was consistent, every 6 or 7 seconds the CPU spiked to 100%. This is quite annoying! After some research I saw others were having this type of problem. Some suspected causes were Sasser virus or keyborad controller. I did due diligence on any viruses on my machine and there were none. Actually I was stumped. And then I thought, you dummy, check the event viewer. Here is the error I was getting: The netcard driver failed the query for OID_GEN_LINK_SPEED. This was the problem. I disabled that network card that I was not using anyway and, like magic, the problem disappeared.
I haven't tried to figure out if SP2 is the culprit and I need new drivers for that network card but maybe. Anyway, the point of my story is if you are having CPU 100% issues, make sure and check your network card. It may be the problem.
Cheers!
I haven't tried to figure out if SP2 is the culprit and I need new drivers for that network card but maybe. Anyway, the point of my story is if you are having CPU 100% issues, make sure and check your network card. It may be the problem.
Cheers!
I would be curious about the processor type, speed and whether this thing ever worked correctly to start with, Dragon is notorious for bogging down machines. It is extremely processor intensive. That was why I asked earlier about the system specs.
Having installed Dragon in dozens of machines, it appears to me that accuracy of this program is greatly related to the speed and amount of memory in the box. A machine that is really not up to the task will display text in larger blocks and have an annoying number of misses.
Having installed Dragon in dozens of machines, it appears to me that accuracy of this program is greatly related to the speed and amount of memory in the box. A machine that is really not up to the task will display text in larger blocks and have an annoying number of misses.
ASKER
Gentlemen,
On the topic of the network card being the culprit for spiking the CPU to 100%, the problem turned out to be the keyboard controller on the mobo. But you're right, hardware can play a funky role in troubleshooting, and should not be overlooked just because we're software guys.
On the topic of Dragon and CPU horsepower and accuracy, processor consumption reported by the task manager during speech to text conversion, the conversion process stays "contemporaneous" with my customers speech, and the graph shows average consumption cruising around 25% and memory about 20% consumed.
Since this is a fresh installation, speaker dependent retraining of Dragon was required. Although the accuracy was pretty good, there were certain words that Dragon got caught up on. But there is a feature where you can specifically train for a specific word, and incrementally improve the accuracy rate from quite good to downright impressive. So if the translated word comes out as "flock", but that isn't what you said, you can add that word to the speech dictionary.
On the topic of the network card being the culprit for spiking the CPU to 100%, the problem turned out to be the keyboard controller on the mobo. But you're right, hardware can play a funky role in troubleshooting, and should not be overlooked just because we're software guys.
On the topic of Dragon and CPU horsepower and accuracy, processor consumption reported by the task manager during speech to text conversion, the conversion process stays "contemporaneous" with my customers speech, and the graph shows average consumption cruising around 25% and memory about 20% consumed.
Since this is a fresh installation, speaker dependent retraining of Dragon was required. Although the accuracy was pretty good, there were certain words that Dragon got caught up on. But there is a feature where you can specifically train for a specific word, and incrementally improve the accuracy rate from quite good to downright impressive. So if the translated word comes out as "flock", but that isn't what you said, you can add that word to the speech dictionary.
I have run into this problem once as well. When i saw this the problem was that SP2 had been download from the internet over a dial up connection and that the download never finished, so when it tried to install it didnt install properly. The only way that i know to fix this is to do a fresh load on the machine and go from there. I know that you have done this once before, but once the load is done then go get your updates for Windows and then restart the computer. I dont know of any other thing that would cause this problem.
Greetings All!
I have read through this whole posting with great interest. In fact I just bought my membership to EE specifically to be able to read this thread.
In my case I have a client who's AMD based system using an Athlon 1800+ on an ASUS A7N266-VM motherboard was exhibiting severe slow downs (to the point of this tech doing a "WTF is this system doing?!!") but as the system was running Windows ME at the time I attributed the issue to spyware and other corruptions.
The client indicated they wanted me to upgrade the operating system to Windows XP Pro as many times when they call software manufacturers for support the support person groans when they hear they were running Windows ME.
I performed the upgrade only to find the system process (not system idle process) again was pegged at 100%. With their permission (they wanted none of their old data back) I reformatted the drive and did a clean install of Windows XP Pro. This system is not connected to the net at any time during this process and as the drive went from FAT32 to NTFS in the reformat process their is no possibility that this issue is related to any previous software.
I was disheartened to come to the conclusion it was a hardware issue but I started troubleshooting down this path. I upgraded the BIOS to the latest release, disabled anything un-needed, installed updated NForce drivers from ASUS site (downloaded via my computer... this machine was never connected to the Internet) all to the same end. The system always crawls along because of this 100% (sometimes dropping to 96%) CPU going to the system process.
I have had this system for almost 3 weeks and I'm really not looking forward to telling her that she needs a new motherboard when I told her at the outset it looked like a software issue.
I will try Matt Ward's suggestion of switching off ACPI in the BIOS and rebuilding this system but I wonder why that would be an issue. (Or more specifically why ASUS would let a motherboard get out with this potential issue.)
I felt it important to mention that this problem occurs with a non-"consumer level" motherboard as well. (One spec I left out is this system has one 512M DIMM and one 128M DIMM for a total of 640Mb which become 608M of memory available to the O/S once 32M is stolen for the NVidia graphics controller).
I will post again after the results of my non-ACPI rebuild. (Assuming anyone is still reading this thread. :)
I have read through this whole posting with great interest. In fact I just bought my membership to EE specifically to be able to read this thread.
In my case I have a client who's AMD based system using an Athlon 1800+ on an ASUS A7N266-VM motherboard was exhibiting severe slow downs (to the point of this tech doing a "WTF is this system doing?!!") but as the system was running Windows ME at the time I attributed the issue to spyware and other corruptions.
The client indicated they wanted me to upgrade the operating system to Windows XP Pro as many times when they call software manufacturers for support the support person groans when they hear they were running Windows ME.
I performed the upgrade only to find the system process (not system idle process) again was pegged at 100%. With their permission (they wanted none of their old data back) I reformatted the drive and did a clean install of Windows XP Pro. This system is not connected to the net at any time during this process and as the drive went from FAT32 to NTFS in the reformat process their is no possibility that this issue is related to any previous software.
I was disheartened to come to the conclusion it was a hardware issue but I started troubleshooting down this path. I upgraded the BIOS to the latest release, disabled anything un-needed, installed updated NForce drivers from ASUS site (downloaded via my computer... this machine was never connected to the Internet) all to the same end. The system always crawls along because of this 100% (sometimes dropping to 96%) CPU going to the system process.
I have had this system for almost 3 weeks and I'm really not looking forward to telling her that she needs a new motherboard when I told her at the outset it looked like a software issue.
I will try Matt Ward's suggestion of switching off ACPI in the BIOS and rebuilding this system but I wonder why that would be an issue. (Or more specifically why ASUS would let a motherboard get out with this potential issue.)
I felt it important to mention that this problem occurs with a non-"consumer level" motherboard as well. (One spec I left out is this system has one 512M DIMM and one 128M DIMM for a total of 640Mb which become 608M of memory available to the O/S once 32M is stolen for the NVidia graphics controller).
I will post again after the results of my non-ACPI rebuild. (Assuming anyone is still reading this thread. :)
Anyone that is still subscribed to this thread will read you comment, but to really get help with your TSing, you should open a new thread in the appropriate area. Since I believe you should start with the OS, I would place it in the XP section. (You might want to have a look at those DIMM's though. Pull out the 128, or make sure that they are compatible with each other. Meaning, I would only use 133 Mhz Dimms, if it were me..)
Also, use Process Explorer from Sysinternals to find out exactly what process is eating up those CPU cycles. This will tell you a lot... If you post your own thread, let us know so we can follow you over there..
FE
Also, use Process Explorer from Sysinternals to find out exactly what process is eating up those CPU cycles. This will tell you a lot... If you post your own thread, let us know so we can follow you over there..
FE
I wanted to post the final cause of the grief I was having and I'd like to add that had Fatal Exception not put me onto Process Explorer I'd probably have just bought the client a new motherboard out of frustration! (Sorry this took a while to get posted.)
This one was a little bit interesting because I hadn't seen disabled hardware continue to cause issues in the operating system software after the system was rebooted.
Here's the weird anomaly I ran into:
- System process is running at 100% CPU and therefore I disable all un-necessary system hardware components on the motherboard as well as pulling any unessential cards.
- Upon system reboot the BIOS reports "Updated ECSD successfully" and yet the operating system still shows the system process running at 100%.
- I run Process Explorer as per Fatal_Exception's suggestion and low and behold the system process drops immediately to 1% and system idle is now 99%! At this point I incorrectly assume that PE did something to fix Windows XP's woes and proceed to re-enable everything that had been shut down.
- When I restart the system again the system process shoots up to 100%.
- This time when I ran PE it should the root of the problem being that number of context switches for USBPORT.SYS was astronomical.
So it turned out that disabling the motherboard USB Ports and buying a USB 2.0 PCI card for the client fixed everything up.
The strange thing is that even when I had disabled them many times in the past and the POST reported that the "ECSD Updated Successfully" XP was still jumping up to 100% CPU utilization presumably from trying to load USBPORT.SYS even though there were no USB ports present in the system!
It wasn't until I ran PE and had the CPU utilization on the system process drop to where it should be that I realized that there was a solution to be had.
Curiouser and curiouser said the Cheshire Cat.
This one was a little bit interesting because I hadn't seen disabled hardware continue to cause issues in the operating system software after the system was rebooted.
Here's the weird anomaly I ran into:
- System process is running at 100% CPU and therefore I disable all un-necessary system hardware components on the motherboard as well as pulling any unessential cards.
- Upon system reboot the BIOS reports "Updated ECSD successfully" and yet the operating system still shows the system process running at 100%.
- I run Process Explorer as per Fatal_Exception's suggestion and low and behold the system process drops immediately to 1% and system idle is now 99%! At this point I incorrectly assume that PE did something to fix Windows XP's woes and proceed to re-enable everything that had been shut down.
- When I restart the system again the system process shoots up to 100%.
- This time when I ran PE it should the root of the problem being that number of context switches for USBPORT.SYS was astronomical.
So it turned out that disabling the motherboard USB Ports and buying a USB 2.0 PCI card for the client fixed everything up.
The strange thing is that even when I had disabled them many times in the past and the POST reported that the "ECSD Updated Successfully" XP was still jumping up to 100% CPU utilization presumably from trying to load USBPORT.SYS even though there were no USB ports present in the system!
It wasn't until I ran PE and had the CPU utilization on the system process drop to where it should be that I realized that there was a solution to be had.
Curiouser and curiouser said the Cheshire Cat.
*grin* Glad you got it though!
My first suggestion is to start Task Manager and then click on the Process tab. Once you have done this click on the CPU column to sort the entries by the most active process. Watch this for a little while and note what is consuming all of your CPU.
If you recognise the process, eg Mozilla, then you have found the culprit. If not, its time to shutdown all of your applications one at a time to see if your machine can recover. If this does not identify the application that is causing the problem, it is time to do a restricted startup.
If the disfunctional process was Mozilla, totally uninstall all of it. All the add-ons etc etc, and then re-install. After doing this everythhing should be OK.
If you have not been able to identify what application is causing the problem in order to perform the restricted startup you need to do the following :
Start > Run > then type "msconsfig"
Select Diagnostic Startup
Re-start the machine.
After doing this, your machine should run normally. Verify this. Then start the applications that you like to use one at a time. Some may not work as you have selective startup enabled. No matter, record which applicatons work without causing a problem. Using a process of elimination you should be able to identify the cause of the problem.
Report back with your initial findings so that we can advise you further. Hopefully it will be something really simple, whereby you just need to un-install and then re-install a piece of software.
Simple things like making sure you have the latest version of the software you want to use is important. Updated your software can often solve problems like this...once you know which piece to update of course :)