can't get FSB stable

Angerisagift

Member
Dec 11, 2007
81
0
0
Has anyone had issues with their antivirus or some other from of scheduled maintenance either freezing up or rebooting their PC on an overnight stress test.

I started Prime95 25.5 last night to finally get my 12 hour run out of the way, so far had done, 10min, 2hour, 6hour,

and when I came in this morning I noticed that my PC seemed to have froze, I couldn't see the screen because my LCD was in standby and moving the mouse didn't seem to resume.

and I think the last time I did an overnight run, it froze too, and I wrote it off as instability after several hours under load.

for AV, I run AVG nightly at 3:00am on their task scheduler. and it auto-updates itself before each run around 2AM
 

nullpointerus

Golden Member
Apr 17, 2003
1,326
0
0
An A/V scan should not conflict with Prime95, and this shouldn't hard-lock the machine.

If you open Event Viewer, are there any specific error messages near the crash time?


Off hand, I'm guessing you need a slight voltage bump (e.g. 1.2875v --> 1.3000v).

But, you see, no stability test like overnight Prime95 can prove the system 100% stable -- read this thread for more info -- so, even if you eliminated all background processes etc. and just ran Prime95 by itself, you are not really stressing the whole system 100%, just the CPU/cache/RAM (depending on the test mode).


When overclocking an Intel CPU, the general order of things to verify is:

FSB
-- Put CPU/RAM at stock, and raise FSB. Test with Prime95 Blend overnight.
-- If you get errors here, raise MCH/FSB voltage, and possibly RAM voltage and retest FSB

RAM
-- Raise RAM/FSB to desired settings for final RAM speed. Test with Memtest86+ overnight.
-- If you get errors here, raise timings, lower clock rate, and/or increase RAM voltage and retest RAM.

CPU
-- Raise CPU to desired settings for final CPU speed. Test with Prime95 Blend overnight.
-- If you get errors here, raise vcore or lower the clock rate and retest CPU.

Full System
-- Run F@H in the background 24/7 for one week. For most people, this is redundant, but...
-- If you get errors here... maybe a slight CPU vcore bump... an expert should know...


Many people will say this procedure is overkill, but, with a borderline-stable overclock, I can make my system fail this way. There's 95% "stable", which is what most people settle for, and then there's 100% stable, which is what I wanted.

The reason things ought to be done in the above order is to isolate and eliminate potential problems so that when the o/c-ing does fail at some stage, you can be reasonably certain of exactly what caused the failure. Then you can make whatever small adjustment fixes that exact problem and continue with the o/c-ing process until you get 100% stable.

So... do FSB first if you haven't already. In your case, I doubt it will make a difference. You are only using 2 GB RAM, which is probably just two sticks (instead of four, which can be flaky and require more voltage in rare cases). And your P35 chipset should handle 333 FSB at stock voltages since the chipset is made for 333 FSB processors anyway. Myself, I would do the FSB test anyway (mainly since I'm slightly paranoid of defective hardware).

Next, do RAM testing, if you have not already done so at those exact RAM/FSB settings. Lower the CPU to (or near as possible to) stock clocks -- just lower the multiplier, turn off C1E/EIST, and keep the bus/memory speeds the same -- and run Memtest86+ overnight to make sure your RAM is stable. If it's not stable, then you'll have random errors with everything else, and naturally they'd be very difficult to track down.

Then continue with the CPU testing. Try raising the vcore to 1.3000v and retest w/ Prime95 Blend.

Finally, when your system passes the first three stages (FSB, RAM, CPU), you can move on to regular usage while o/c-ed and run Folding @ Home client in the background 24/7. At that point, any EUE (errors) you get from F@H probably indicate you will need a little more CPU voltage to be 100% stable.
 

Angerisagift

Member
Dec 11, 2007
81
0
0
[
FSB
-- Put CPU/RAM at stock, and raise FSB. Test with Prime95 Blend overnight.
-- If you get errors here, raise MCH/FSB voltage, and possibly RAM voltage and retest FSB


When you say this are you saying that I should raise the FSB to the speed I plan on using for my OC then set CPU and RAM to stock/normal voltages also with this I'm assuming it means I need to lower my multiplier get as close to stock CPU speed as possible?

because if I'm understanding this right I would need to setup for say 7x333 for 2331Mhz

or 8x333 for 2664Mhz, because I'm assuming we're trying to rule out the FSB of the board as being at fault, so we would want the other components to run at manufacturer rated settings to verify they aren't the cause.

, as a Technician that troubleshoots a variety of technical issues on a daily basis, I understand your school of thought, ruling out each component one by one as a possible point of failure. However I'm having some difficulty understanding how I'd setup the test for myself.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Originally posted by: Angerisagift
Has anyone had issues with their antivirus or some other from of scheduled maintenance either freezing up or rebooting their PC on an overnight stress test.

I started Prime95 25.5 last night to finally get my 12 hour run out of the way, so far had done, 10min, 2hour, 6hour,

and when I came in this morning I noticed that my PC seemed to have froze, I couldn't see the screen because my LCD was in standby and moving the mouse didn't seem to resume.

and I think the last time I did an overnight run, it froze too, and I wrote it off as instability after several hours under load.

for AV, I run AVG nightly at 3:00am on their task scheduler. and it auto-updates itself before each run around 2AM

nullpointerus posted some excellent info and pointers, superb really.

To add to the thread I would point out that Prime95 + AVG is in its own right a form of stress testing...and if Prime95 + AVG is all it takes to "over stress" your system to the point of highlighting an instability in the overclock then it would be prudent to take the datum as such.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
I agree.....I dont see AVG as being a culprit from interfering with prime....especially when you say a hardlock....

Hardlocks are majority of the time power issues....BSOD are usually memory issues....reboots can be power and/or heat, or if you forget to turn off auto restarts in XP after system errors....

 

nullpointerus

Golden Member
Apr 17, 2003
1,326
0
0
Originally posted by: Angerisagift
Originally posted by: nullpointerus
FSB
-- Put CPU/RAM at stock, and raise FSB. Test with Prime95 Blend overnight.
-- If you get errors here, raise MCH/FSB voltage, and possibly RAM voltage and retest FSB
When you say this are you saying that I should raise the FSB to the speed I plan on using for my OC then set CPU and RAM to stock/normal voltages also with this I'm assuming it means I need to lower my multiplier get as close to stock CPU speed as possible?

because if I'm understanding this right I would need to setup for say 7x333 for 2331Mhz

or 8x333 for 2664Mhz, because I'm assuming we're trying to rule out the FSB of the board as being at fault, so we would want the other components to run at manufacturer rated settings to verify they aren't the cause.

, as a Technician that troubleshoots a variety of technical issues on a daily basis, I understand your school of thought, ruling out each component one by one as a possible point of failure. However I'm having some difficulty understanding how I'd setup the test for myself.
Yes, exactly. Set everything to stock except the single component to be tested, and then eliminate them one-by-one.

(Generally speaking, the problem with CPU o/c-ing is that everything is "linked" to the FSB: we do not have perfect flexibility, but we do have some flexibility.)

FSB --> test with CPU/RAM as close to (or under, if need be) stock clocks
RAM --> test with previously-tested FSB, but with CPU as close to stock clocks
CPU --> test with previously-tested FSB/RAM, and CPU at desired settings

So, with your rig, you've already picked a good target FSB. Your CPU should easily hit the desired speed. Your MB/RAM should be able to support the target FSB with little to no excess voltage and at factory recommended settings. That's ideal!

Your Target OC
FSB --> 333 mhz
RAM --> 333 x 2.4 = 799 mhz
CPU --> 333 x 9.0 = 2997 mhz


The only thing to remember when lowering your CPU multiplier is that you must disable C1E and EIST because they'll just automatically set it to its maximum.


Step 1: FSB

OC Test Settings #1
FSB --> 333 mhz ......................... <-- desired
RAM --> 333 x 2.4 = _800 mhz ..... <-- at or below stock
CPU --> 333 x 7.0 = 2331 mhz ..... <-- at or below stock

Now run Prime95 Blend overnight. This test mode will stress CPU, cache, and RAM evenly. Since the motherboard's northbridge (NB) coordinates transfers between cache and RAM, this will also test the NB. So we are verifying that the board operates correctly at 333 mhz.

If you get errors/lockups of any kind, raise MCH and FSB by 0.1v and retest at step 1.

When this testing passes without errors, proceed to step 2 w/ same settings, except we now raise RAM to target speed.


Step 2: RAM

OC Test Settings #2
FSB --> 333 mhz ......................... <-- desired, tested OK
RAM --> 333 x 2.4 = _800 mhz ..... <-- desired
CPU --> 333 x 7.0 = 2331 mhz ..... <-- at or below stock

(The numbers happen to be the same as step #1, but that's just a coincidence.)

Now run Memtest86+ 2.00 overnight. This test mode will place most of the stress on RAM. it will catch most (but not all) errors in memory, due to either defective RAM or RAM being run at unsupported timings/clocks and/or with insufficient voltage.

If you get errors/lockups of any kind, raise VDIMM by 0.1v and retest at step 2.

When this testing passes without errors, proceed to step 3 w/ same settings, except we know raise CPU to target speed.


Step 3: CPU

OC Test Settings #3
FSB --> 333 mhz ......................... <-- desired, tested OK
RAM --> 333 x 2.4 = _800 mhz ..... <-- desired, tested OK
CPU --> 333 x 9.0 = 2997 mhz ..... <-- desired

Now run Prime95 Blend overnight. This test mode is a mixture of Small FFT (which tests heat) and Large FFT (which tests voltage), so it's perfect for overnight CPU testing of all fundamentally important CPU functions.

If you get errors/lockups of any kind, raise VCORE by a small amount (varies by board) and retest at step 3.

When this testing passes without errors, proceed to step 4 w/ same settings.


Step 4: Full System

Target OC Settings
FSB --> 333 mhz ......................... <-- desired, tested OK
RAM --> 333 x 2.4 = _800 mhz ..... <-- desired, tested OK
CPU --> 333 x 9.0 = 2997 mhz ..... <-- desired, tested OK

(By now, all components have tested fine individually, and stressing all of them through multi-tasking should (in theory) (via step 4) produce no errors. But in practice, you can pass steps 1-3 with a given target OC and still have problems in step 4.)

Now run the Folding @ Home client constantly in the background for one week. If you get no errors or EUEs, you can be sure your o/c is *very* stable.

If not, then you must get expert advice diagnosing the errors. Once a particular problem is pinpointed, you must go back and retest at step 1, 2, or 3. You wouldn't have to do them all in a series *by this point*, so just try to fix FSB, RAM, or CPU settings individually and then retry step 4 to verify that your fix worked.

(For your particular OC and parts, a failure at step 4 almost certainly means you need a slight bump in VCORE -- nothing more.)
 

Pederv

Golden Member
May 13, 2000
1,903
0
0
Originally posted by: Duvie
I agree.....I dont see AVG as being a culprit from interfering with prime....especially when you say a hardlock....

Hardlocks are majority of the time power issues....BSOD are usually memory issues....reboots can be power and/or heat, or if you forget to turn off auto restarts in XP after system errors....

Duvie, hardlocks may be caused by power issues, most of the time, but I had an issue last night where my system went into a hardlock. I was converting an ipod slideshow into a DVD and my system kept locking up. Turned out my CPU fan had died and the over heating was causing the lockup.

I have no idea how long my fan has been dead, the most work my system gets anymore is as a print server.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: Pederv
Originally posted by: Duvie
I agree.....I dont see AVG as being a culprit from interfering with prime....especially when you say a hardlock....

Hardlocks are majority of the time power issues....BSOD are usually memory issues....reboots can be power and/or heat, or if you forget to turn off auto restarts in XP after system errors....

Duvie, hardlocks may be caused by power issues, most of the time, but I had an issue last night where my system went into a hardlock. I was converting an ipod slideshow into a DVD and my system kept locking up. Turned out my CPU fan had died and the over heating was causing the lockup.

I have no idea how long my fan has been dead, the most work my system gets anymore is as a print server.



Good point.....that can happen...I usually have the feature set on my boards to shutdown if it hits a certain temp.....

I have seen it happen with a video card that was overheating as well....
 

Angerisagift

Member
Dec 11, 2007
81
0
0
everything seemed to be blowing when I found it, it's an Antec Nine Hundred, with 2 front fans mounted in the top and bottom positions, top to provide air flow to the CPU heatsink and mobo components. bottom to cool the power supply.

side window intake to help cool VGA, 200mm top fan as exhaust for CPU and 120mm rear fan. It's by no means a loud system but you can hear it when it's running. the 5 tricool fans are all set to medium, I got used to hearing fans with the 7 80mm fans in my P4 system which is much louder that this, I don't care for silent PC's the sound of fans dissipating heat, helps me sleep at night.
 

Angerisagift

Member
Dec 11, 2007
81
0
0
Ok so I ran the overnight FSB testing using the prescribed settings, HARDLOCK, after approx 40 minutes, went to bed around 12:45am, came in this morning and the clock was stuck at 1:20am

so this morning I tried to up the MCH and FSB voltage as suggested and now when I run blend it reboots within a minute

was able to run a small FFT test for something like 8 hours yesterday which confuses me,
 

nullpointerus

Golden Member
Apr 17, 2003
1,326
0
0
Originally posted by: Angerisagift
Ok so I ran the overnight FSB testing using the prescribed settings, HARDLOCK, after approx 40 minutes, went to bed around 12:45am, came in this morning and the clock was stuck at 1:20am
That sounds odd, given the settings...

Are you sure C1E and EIST are both disabled in the BIOS?

Can you open CPU-Z during testing and verify CPU/RAM settings are as expected?

so this morning I tried to up the MCH and FSB voltage as suggested and now when I run blend it reboots within a minute
Try increasing the vcore as jaredpace said.

It's odd, though... Your CPU ought to be able to easily handle a 333 MHz FSB when the multiplier is turned down and the RAM is at stock settings. Even my CPU, which is an o/c dud that craps out at 3.0 GHz w/ 1.5v vcore, can still do 333 MHz FSB w/ lower multiplier.

was able to run a small FFT test for something like 8 hours yesterday which confuses me,
Yep, different tests will expose different weaknesses. It's normal to be able to run one test for nine hours and then fail a different kind of test in minutes. That's why it's important to test each aspect of the o/c individually.

As Idontcare suggested a while ago, (reproducible) failures are actually a good thing during o/c-ing. Once you find an instability, try changing one setting at a time -- usually some kind of voltage or an inappropriate BIOS setting -- until the system no longer gives you that particular error.

The more issues you work through, the more stable the system becomes.
 

Conroy9

Senior member
Jan 28, 2000
611
0
0
Originally posted by: nullpointerus
When overclocking an Intel CPU, the general order of things to verify is:

FSB
-- Put CPU/RAM at stock, and raise FSB. Test with Prime95 Blend overnight.
-- If you get errors here, raise MCH/FSB voltage, and possibly RAM voltage and retest FSB

RAM
-- Raise RAM/FSB to desired settings for final RAM speed. Test with Memtest86+ overnight.
-- If you get errors here, raise timings, lower clock rate, and/or increase RAM voltage and retest RAM.

CPU
-- Raise CPU to desired settings for final CPU speed. Test with Prime95 Blend overnight.
-- If you get errors here, raise vcore or lower the clock rate and retest CPU.

Full System
-- Run F@H in the background 24/7 for one week. For most people, this is redundant, but...
-- If you get errors here... maybe a slight CPU vcore bump... an expert should know...

I think I'll try going through this process. I've been just setting everything higher together and hit a barrier at around 425mhz where I'm not sure which of my components is causing problems.

I have a question though - how do you keep the RAM at stock speed while raising FSB? Would it be easier to test at high RAM speed with stock FSB first?
 

nullpointerus

Golden Member
Apr 17, 2003
1,326
0
0
Originally posted by: Conroy9
Originally posted by: nullpointerus
When overclocking an Intel CPU, the general order of things to verify is:

FSB
-- Put CPU/RAM at stock, and raise FSB. Test with Prime95 Blend overnight.
-- If you get errors here, raise MCH/FSB voltage, and possibly RAM voltage and retest FSB

RAM
-- Raise RAM/FSB to desired settings for final RAM speed. Test with Memtest86+ overnight.
-- If you get errors here, raise timings, lower clock rate, and/or increase RAM voltage and retest RAM.

CPU
-- Raise CPU to desired settings for final CPU speed. Test with Prime95 Blend overnight.
-- If you get errors here, raise vcore or lower the clock rate and retest CPU.

Full System
-- Run F@H in the background 24/7 for one week. For most people, this is redundant, but...
-- If you get errors here... maybe a slight CPU vcore bump... an expert should know...

I think I'll try going through this process. I've been just setting everything higher together and hit a barrier at around 425mhz where I'm not sure which of my components is causing problems.
Sorry, I haven't been checking the forums much in the past few days.

You could try:
-- Loosen RAM timings.
-- Raise RAM voltage.
-- Raise FSB and MCH voltage.

If you can't get it to work on your own, then please post a new thread w/ rig specs.

I have a question though - how do you keep the RAM at stock speed while raising FSB?
To do that, you must have (a) RAM that's rated higher than the FSB at 2.0 multiplier, or (b) a board that lets you go lower than the 2.0 multiplier. However, (b) is quite rare on these Intel platforms, IIRC. For your situation, you would need DDR2-1066 RAM to follow my method to the letter.

Would it be easier to test at high RAM speed with stock FSB first?
You can, but it won't push the RAM as hard as it would be if the FSB was higher.

IOW, my DDR2-667 value memory passes @ 275 x 3.0 (815) but fails @ 350 x 2.0 (700).

Put yet another way, RAM multipliers above 2.0 are useless in most real-world situations; without the higher FSB, the CPU can't take advantage of the extra memory bandwidth.

This theory was thoroughly tested here:
http://forums.anandtech.com/me...112855&highlight_key=y
 

Conroy9

Senior member
Jan 28, 2000
611
0
0
Originally posted by: nullpointerus
You can, but it won't push the RAM as hard as it would be if the FSB was higher.

IOW, my DDR2-667 value memory passes @ 275 x 3.0 (815) but fails @ 350 x 2.0 (700).

Put yet another way, RAM multipliers above 2.0 are useless in most real-world situations; without the higher FSB, the CPU can't take advantage of the extra memory bandwidth.

This theory was thoroughly tested here:
http://forums.anandtech.com/me...112855&highlight_key=y
Ah... I understood that a higher RAM multiplier isn't worth the trouble; the only reason I suggested doing it was to try to isolate components, since I saw no way to OC the FSB and leave the RAM stock. I see that there's not even a point to that, from your experience.
That would explain why my RAM passed tests at 400*2.4 (960) but required extra DDR2 voltage to work at 450*2.0 (900). I was very confused about that.

Thanks for the clarification!

Anyway, since things seem stable at 450 fsb, x2 ram, x6 cpu (a day of prime95 blend), I guess my problem is probably the processor. I'm probably done pushing it because I don't want to increase CPU voltage any further.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,570
10,202
126
Originally posted by: nullpointerus
Would it be easier to test at high RAM speed with stock FSB first?
You can, but it won't push the RAM as hard as it would be if the FSB was higher.

IOW, my DDR2-667 value memory passes @ 275 x 3.0 (815) but fails @ 350 x 2.0 (700).

Put yet another way, RAM multipliers above 2.0 are useless in most real-world situations; without the higher FSB, the CPU can't take advantage of the extra memory bandwidth.

This theory was thoroughly tested here:
http://forums.anandtech.com/me...112855&highlight_key=y
That's a very interesting point. Since you are testing the RAM using the CPU, it makes sense that even with the memory multiplier at 3.0, the CPU's FSB is only running at 275, so the memory only gets tested at 275, really, whereas at 350 x 2.0, it gets tested at 350.

 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |