I7 860 == Unpredictable Blue Screen of Death == Windows 7 64-bit

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
That is an understandable explanation. Besides running Crysis, Call of Duty 6 and Fallout 3 (the only three games I have installed at the moment), is there any way that I can stress this machine to the point where a BSoD might have occurred at 1333Mhz or X.M.P. 2000MHz? The only things that I can think of is to encode/convert a number of high quality videos all at once, whilst defragging or burning something (possibly both), but I'm sure that there's software out there that would do a much better and cleaner job.

I don't suppose Googling "stress machine software" would help?
 
There are lots of tools and many people have their favorite. I've used Prime95 to stress my CPU, check temperatures, launch 8 threads to make sure hyperthreading is still working and, my favorite, launch just 1 thread to make sure Turbo is working. You can find it here: http://www.mersenne.org/freesoft/

Two of the tests, Blend and Large FFTs, are often used to test memory. You can read more about it here, or just google around: http://www.overclock.net/faqs/18731-info-prime95-primes-stability-testing.html

Here's a discussion with more stress / benchmark / torture testing utilities: http://forums.tweaktown.com/f69/latest-overclocking-programs-system-info-benchmarking-stability-tools-30530/
 
Well my machine seems to pass them pretty well.

But I'm still getting BSoDs. Not complex like when I had F3 though. It's simple and straightforward. I haven't got a screenshot of it, but I get it whether I'm running my RAM at 1333MHz or 1600MHz or 2000MHz (X.M.P.). Could it possibly be a software issue now? I'm running Mozilla Firefox with Ad-block Plus and the Web of Trust, and I know that WoT wasn't originally 64-bit compatible. I'm also running a copy of Revo Uninstaller Free, which also is not 64-bit compatible.
 
Just so I'm clear, there was a point where you had eliminated the BSoDs, right? Seemed like after you updated drivers and while running RAM on auto (1333) they went away. Are the programs you mentioned newly installed since then?

If the software is causing the problem, there's a better chance you'll find something in the Event Logs. Open Computer Management. Go to System Tools - Event Viewer - Customer Views - Administrative Events. Or look under System Tools - Event Viewer - Windows Logs and in there look at Application and System.

Is there any common thread between the BSoDs? Anything that you're doing at the time or software that is running at the time?
 
Hi,

This may or may not be useful to you but may be worth a go.

Early on in the thread I think you mentioned you were using zone-alarm, if this is the case, try uninstalling it, and see if the BSOD problems stop.

I only suggest this as I'm running Windows 7 and was getting BSOD almost every 5 - 10 minutes, tried everything I could think of, updated drivers etc, etc. Still got BSOD's.

Uninstalled Zone Alarm and since then no more BSOD's.

Although this does kind of suck as I've used ZoneAlarm forever. Googled the problem since and have found a few people having same or similar problems.

Hope this helps.

 
Just so I'm clear, there was a point where you had eliminated the BSoDs, right? Seemed like after you updated drivers and while running RAM on auto (1333) they went away. Are the programs you mentioned newly installed since then?

If the software is causing the problem, there's a better chance you'll find something in the Event Logs. Open Computer Management. Go to System Tools - Event Viewer - Customer Views - Administrative Events. Or look under System Tools - Event Viewer - Windows Logs and in there look at Application and System.

Is there any common thread between the BSoDs? Anything that you're doing at the time or software that is running at the time?
Click to expand...

To be honest, I'm not sure if such a point existed anymore. It could just have been the fact that I wasn't using this machine for long enough to get the next problem. I've used WoT and Ad-block Plus with Firefox since practically day one. Revo came in a little while later, and I have since uninstalled it (so I can't give a date).

I'll take a look in the Event Logs now and post my findings.

I only seem to get these BSoDs when I leave the computer for an extended period of time. I might go for lunch, or a quick toilet break, etc. The usual configuration is Firefox, Windows Live Messenger, Zone Alarm, Skype, Razer Mouse Control Centre and Avira AntiVir Personal Edition, but it has happened with Modern Warfare 2 running as well (once).

Hi,

This may or may not be useful to you but may be worth a go.

Early on in the thread I think you mentioned you were using zone-alarm, if this is the case, try uninstalling it, and see if the BSOD problems stop.

I only suggest this as I'm running Windows 7 and was getting BSOD almost every 5 - 10 minutes, tried everything I could think of, updated drivers etc, etc. Still got BSOD's.

Uninstalled Zone Alarm and since then no more BSOD's.

Although this does kind of suck as I've used ZoneAlarm forever. Googled the problem since and have found a few people having same or similar problems.

Hope this helps.
Click to expand...

I've updated ZoneAlarm to the latest version, which Zone Labs claims is Windows 7 64-bit compatible (after reading this post). I've also set ZA to Windows XP SP3 Compatibility Mode. Would this change the situation at all? Because I'd rather not ditch Zone Alarm.
 



Honestly not really sure if setting the compatibilility mode will make a difference. I didn't want to ditch ZoneAlarm either. I'd suggest uninstalling it for a few days, see if the BSOD's stop. If they do, raise it as an issue with Zone-Alarm and see what they suggest.

Good Luck!
 
Alright, I've got good news 😀.

Zone Alarm seems to be the new BSoD culprit - no BSoDs since uninstalling it, and I have been stressing it to the max the whole time ZA has been gone.

X.M.P. is still a culprit however, as I got a (different) BSoD when I used the IntelBurnTest with X.M.P. on (all other settings reverted to normal), but when I used X.M.P. = off and SPD = 12 with the same test (and more), I got no BSoDs.

I can't really say that my system is completely stable until tomorrow or the day after, but I think I can now look at some light-to-mild overclocking and start pushing my RAM frequencies closer to 2000MHz (as close as possible). Note that I also manually entered in my timings.

Before I get to work here, I do have three questions:

1. Is it safe (and beneficial) to change Memory Performance Enhance from Turbo to Extreme? What does this do exactly?

2. Is C.I.A.2 a safe way to overclock, and if so, is it safe to have Memory Performance Enhance set to Turbo/Extreme?

EDIT: I'm only interested in C.I.A.2 if it doesn't tamper with the voltage. I'm quite happy with my CPU having the same voltage.
 

Glad you got that solved! Thanks to whoever the anonymous tipster was. I checked ZoneAlarm's support forums and there were a number of other people in the same situation.

X.M.P. is still a culprit however, as I got a (different) BSoD when I used the IntelBurnTest with X.M.P. on (all other settings reverted to normal), but when I used X.M.P. = off and SPD = 12 with the same test (and more), I got no BSoDs.
Click to expand...
I'm sure this is b/c XMP is OCing/changing your CPU. With SPD 12 you are running at 1600, so at least you're close. I think you're going about this the right way. That is, now start manually OCing to get your CPU a little faster and get your RAM up.

1. Is it safe (and beneficial) to change Memory Performance Enhance from Turbo to Extreme? What does this do exactly?
Click to expand...
The consensus seems to be if you're going to OC, then set it to standard, get your OC where you want it, then if you can increase it to Turbo or Extreme and the system stays stable leave it there otherwise back it down.
From everything I've been able to find, the Memory Performance Enhance affects a Memory subtiming called tRd, or Static tREAD. Here's a good discussion about it: http://forums.anandtech.com/showthread.php?t=135224
Also, another piece worth quoting is from a guide which will tell you more than you ever wanted to know about memory and timings: http://forums.tweaktown.com/f69/memory-timings-explained-w-suggested-timings-memset-vs-bios-27283/
Static tREAD Value: (tRD) This setting is most commonly known as Performance Level or tRD this is the most effective chipset performance register available to adjustment. As FSB and memory speed are scaled/increased, tRD and Northbridge voltage will have to be increased to accommodate the additional data throughput. The idea when tweaking your system for the BEST performance is to run the tightest (Lowest) tRD possible at any given FSB or memory speed.
Click to expand...
Of course this was written pre-P55. Here's part of an article that mentions tRD on X58/P55: http://www.anandtech.com/showdoc.aspx?i=3671&p=6
Interestingly, it suggests the tRD is the RTL (Round Trip Latency) which is exposed in the BIOS. If I'm understanding this right, then you ought to be able to change the Memory Performance Enhance setting and observe a change to the Round Trip Latency. It may take a BIOS save and reboot. This level of detail is new to me so if you find any more information or test any of the settings please share your results.

2. Is C.I.A.2 a safe way to overclock, and if so, is it safe to have Memory Performance Enhance set to Turbo/Extreme?

EDIT: I'm only interested in C.I.A.2 if it doesn't tamper with the voltage. I'm quite happy with my CPU having the same voltage.
Click to expand...
I don't know a lot about C.I.A.2. Personally I consider it kind of a gimmicky overclock tool, like they now have software and buttons for. Something useful for someone who may want a quick bump without really understanding their settings. You're already past the beginner stage and know more than most so I would avoid this setting. In my mind the biggest benefits of changing settings by hand is you learn what the board can do and when you're done you know exactly what it is doing. Of course, as I said, I don't know much about C.I.A.2.
 
Rider,
I'm hoping you can help me test a situation with XMP and Turbo. I've been helping a number of people who got more RAM than their CPU can handle at default. And I know what happens in most situations. But there's one I'm not clear on. If you turn on XMP and Turbo is enabled, what happens? I can't test it because my RAM is only 1600 MHz so when I turn on XMP, my SPD jumps to x12 and my BLCK does not change. Your system is a good test because we know XMP changes your BCLK to ~167. But I don't know what it does to Turbo.

What I'd like to see are what your system is doing at idle, at load, and finally with only 1 core running at max. I think there are 4 CPU-Z screenshots that would provide all the information I need. It will require running Prime95 which I think you downloaded earlier during your stability testing.

If you're willing, here's the test:
In BIOS:
■Turn on XMP
In BIOS on the Advanced CPU Core Features set the following:
■Intel Turbo Boost Tech. - Enabled
■CPU Cores Enabled - left as All
■CPU Multi-Threading - Enabled
■CPU Enhanced Halt (C1E) - Enabled
■C3/C6/C7 State Support - Enabled
■CPU Thermal Monitor - Enabled
■CPU EIST Function - Enabled
■Bi-Directional PROCHOT - Enabled

Then when you boot your CPU, open CPU-Z.
1. Take a CPU-Z screenshot of the Memory tab to confirm that the RAM is running at ~2000. (Wouldn't hurt to include the Task Manager - Performance screen either.)

2. Let your machine idle for a bit. If you see the multiplier drop to x9 or something low that is what I'm looking for. Take a CPU-Z screenshot. If it does not drop, go to Control Panel - Hardware and Sound - Power Options and make sure you have it set to Balanced. You could go into Advanced Power Settings - Processor power management - minimum processor state and make sure this is set to 5%. But if you haven't been in there before that should be the default value. If you are running at Balanced and the multiplier never drops that's fine, please take an Idle screenshot anyway. (Wouldn't hurt to include the Task Manager - Performance screen either.)

3. Run Prime95. The goal with this test is to run all 4 cores. You can run all 8 threads, or you can run just threads 1, 3, 5, and 7. If you have not done this before, you do that by opening Prime95. Usually it starts all threads by default when you open it, if so Stop it. Then go to Continue, but when it pops a window uncheck the resume all checkbox and enter a 1 in the thread to start field. Then Repeat, Continue just on thread 3, then 5, then 7. If you check Task Manager at this point you should see 4 of the 8 CPUs running at 100% and it should be every other CPU. Take a CPU-Z screenshot. Wouldn't hurt to include the Task Manager - Performance screen either.

4. Stop Prime95. Now restart it just on thread 1. Take a CPU-Z screenshot. Wouldn't hurt to include the Task Manager - Performance screen either.
 
Before I agree to doing this, I would like to know if there are any potential or apparent risks. Unlike when I changed my RAM to 1600MHz, I know for a fact that X.M.P. causes BSoDs, even with Turbo set to enabled, etc. because I did it yesterday and crashed the system. Obviously I'm going to take a cautious step here.

While I can't give you much of a report from that, but I did deliberately leave Turbo enabled when X.M.P. was on, simply hoping to get the additional frequencies to help my memory at 2004MHz. The CPU frequency was still 2.84GHz, which leads me to believe that all of the rest of the CPU-Z shots I took with X.M.P. enabled will be the same.

Don't get me wrong - I know that you wouldn't intentionally want to damage my machine, but I need to know all the risks (please list them in the following order: Certain, Almost Certain, HIGHLY likely, likely, probable possible, unlikely, HIGHLY unlikely, 1 in 1,000,000 chance, etc.), so I can be aware of what might happen.
 
You're right. That's a good point. Honestly I didn't think to much about risk, primarily because I knew the settings are at least similar to what you ran before. Obviously any of them have a risk of BSoD. The single thread on P95 w/ Turbo enabled could carry a risk of overvolting if it did run at Turbo. Probably the least likely but most potentially damaging risk involved in the tests, assuming Turbo mode actually did kick in. It sounds like it didn't in the past.

The risk of a serious problem is probably very unlikely. The boards are constructed with protection built in so they should shut themselves down before anything bad happened. But sometimes the failsafes fail. Even if it's a one in a million chance that you board would overvolt I'd hate for it to happen on a test I asked you to run. So while I'm sure it's small, I don't want to ask you to do anything that will put your system at risk or that you're not comfortable with. I had just helped somebody less savy with a similar problem and then got your update and responded without considering risk.

Based on what you saw, I believe you're conclusion is probably right: "Which leads me to believe that all of the rest of the CPU-Z shots I took with X.M.P. enabled will [show the CPU frequency at 2.84GHz]" It sounds doubtful EIST or Turbo is actually working in this scenario.

I'm going to think about it some more. I really should be able to come up with a safer way to test these scenarios, or a way to simulate it on my own.
 


That's perfectly fine :). Again, I know you wouldn't want to intentionally damage my machine, but I needed to know all of the risks beforehand. Seeing as there's an overvolting risk, I'm not sure that I'd like to do this, regardless of the potential for that risk to occur. If there's a more risk free way, I'll be happy to try that (again, so long as I know all of the potential risks at hand).

It's a pity that virtual machines are yet to carry the latest and greatest CPUs on the market for this kind of stuff (if I knew how to make such a device, I'd try to make it as universal as possible, and allow people to pick the components they want, so long as it doesn't exceed the physical computer's abilities). Perhaps VMWare should get in contact with companies such as Intel, AMD, Gigabyte, etc. and make some kind of ultimate Virtual Machine Box.

Also, I just read your post before you asked me to run these tests (didn't see it before, sorry). Thanks for posting all of those links. Although I haven't gotten round to reading them just yet, the last link that you sent (about Turbo) was very interesting, so I'm pretty sure you've given me some more interesting reading material over lunch.

After I found out that C.I.A.2 was in both the BIOS and EasyTune, I ditched the idea right away. EasyTune just doesn't sit with me; especially after everything I've read about it, both in this thread and elsewhere, it seems that it is really bad to use.

Thanks for having the confidence in me to overclock through the BIOS - I just leave SPD at 12 and adjust the BCLK in +10 increments, right? Courtesy of this thread.

If I get a chance today, I might do a minor overclock to make my BCLK = 153.
 

That sounds right. Just remember you'll want to make sure you SPD x BCLK stays under your RAM rated frequency. The other thing that most guides don't seem to mention is the initial settings for voltages. This is one are where I wish they were more clear, but it seems the exact number can differ on each individual board. The thing to do may be to note what they run at for each BCLK. If they change you can set them to the lowest value and see if the system stays stable. Key Vs I've read about are CPU Vcore (and it's corresponding DVID setting), QPI/Vtt, PCH Core and CPU PLL.
 
Did you put in your corsair 750w PSU yet ? if so wiggle the cables around make sure none are loose.

zip up your minidumps and post them on http://www.sevenforums.com/ so someone can give you a read on them.

uninstall all AV software for testing and only use a AV that is Windows 7 compatible ( you can look here ) http://www.microsoft.com/windows/compatibility/windows-7/en-us/Search.aspx?type=Software&s=zonealarm

Also make sure your hardware is compatible http://www.microsoft.com/windows/compatibility/windows-7/en-us/Search.aspx?type=Hardware&s=Razer%20Lachesis
 
Did you put in your corsair 750w PSU yet ? if so wiggle the cables around make sure none are loose.

zip up your minidumps and post them on http://www.sevenforums.com/ so someone can give you a read on them.

uninstall all AV software for testing and only use a AV that is Windows 7 compatible ( you can look here ) http://www.microsoft.com/windows/c [...] =zonealarm

Also make sure your hardware is compatible http://www.microsoft.com/windows/c [...] 20Lachesis
Click to expand...

I think that the Corsair was installed the day before I flashed the BIOS, and therefore probably wasn't a problem at all (as you suggested).

I'm not sure where my minidumps are stored, and I haven't seen an error report request for nearly four weeks, even though I've had a number of BSoDs since. This could be a problem.

Funnily enough, all of my BSoDs stopped when I removed Zone Alarm, even though it is "Windows 7 64-bit compatible". I'm not keen on installing it again, at least until version 91.008.005 or later is released.

Finally, all of my hardware is Windows 7 64-bit compatible, as are my drivers are up to date (as suggested earlier).

===

I decided that a 4-600MHz increase is fine for the time being, but I'd like to know before I overclock:

1. By how much will my CPU temperature increase without the voltage being amped up?
2. The "instability" of a machine means experiencing BSoDs, right? Are there any other "unstable" symptoms (perhaps lag, or something similar)?
3. This IS the only way to get my RAM to work at ~2004MHz besides X.M.P., right?
 
1. If you don't increase voltages (which you will probably have to) you won't see a big jump in temps. Generally speaking, CPU temperatures are linearly related to CPU voltage.

2. Yes, BSODs are the most common symptom of a failed overclock. You might just see programs crashing in windows 7 (such as benchmark tests) before you get to a BSOD. Windows 7 is more stable than older operating systems and I've seen some failed overclocks just lead to prime95 crashing rather than a full BSOD. You shouldn't see any lag without also seeing a BSOD or program crash.

3. Not technically. You can get your RAM to run at 2000 Mhz by increasing the Base clock (BCLK), then decreasing the CPU multiplier. (BCLK x CPU multiplier = CPU speed). By doing so, you leave your CPU at its stock speed while effectively raising your RAM speed (Ram speed = BCLK x RAM multiplier). I hope this makes sense.
 
^ To point 3 above, you could increase BCLK and descrease the CPU multplier to not OC your CPU. But this is what turning on XMP does. And as we've discussed by changing your CPU multiplier so it's no longer equal to 21 you lose Turbo.
 
I ran some tests yesterday, first playing with V settings and then bumping up my BCLK. My DRAM voltage was previously set to 1.5V, my timings to 9-9-9-24-2T, and SPD to x12. This is also with Turbo, ESIT, C-states set to Enabled.

First, before chaning settings I used Prime95 to run some tests and CPU-Z and Task Manager-Performance to monitor the results. The 3 tests I ran were: system idle, P95 8 threads (system maxed out) and P95 1 thread (Turbo maxed out). I used Task Manager to make sure 0, 8, or 1 thread were running, respectively. I used CPU-Z to measure CPU Multiplier, Core Speed and Core Voltage.

I don't have the baseline results in front of me but they were in line with:
Idle: x9 Multiplier, ~1.197 GHz, 0.862 V Core Voltage
8 threads: x22 Multiplier, ~2.926 GHz, 1.132 V Core Voltage
1 Thread: x26 Multiplier, ~3.458 GHz, 1.230 V Core Voltage

Having established this baseline, I went into the Advanced Voltage Settings** and set:
CPU Vcore = Normal
Dynamic Vcore (DVID) = -0.01250 (as I said a minor undervolt)

I ran the same tests and observed lower voltage. So I continued on. I then changed my BCLK to 145 and SPD to x10 (my memory is only rated at 1600 MHz so that's the threshold I have to stay under).

I reran the same tests with Prime95. According to CPU-Z I am getting the following (Voltages are the same as when I tested with the DVID change but BCLK at 133):
Idle: x9 Multiplier, ~1.197 GHz, 0.848 V Core Voltage
8 threads: x22 Multiplier, ~3.190 GHz, 1.120 V Core Voltage
1 Thread: x26 Multiplier, ~3.768 GHz, 1.216 V Core Voltage

I have not done extensive stress testing. But preliminary results showed a ~260 (4 cores) - 310 (1 core) MHz increases with a nominal decrease in Voltage. While I was not testing for temperature and power draw from the wall at this point (measured with a kill a watt), I did keep an eye on them and they were basically the same. As expected since V didn't increase (since I decreased it if anything I'd expect to see less, or the same).


**I also changed the other settings on the Advanced Voltage Settings. They were at Auto. I set them all to Normal. I'm wondering if I should have just set actual values for them? I think most people leave them on Auto but nobody every really comments about them so I've been unsure. The settings I changed from Auto to Normal are:
QPI/Vtt Voltage
PCH Core
CPU PLL
DRAM Termination
Ch-A Data VRef.
Ch-B Data VRef.
Ch-A Address VRef.
Ch-B Address VRef.
I left Load Line Callibration at Auto
 
Alrighty,

It seems that my BCLK can go right up to 153 without getting a blue screen (I haven't tried anything between 153 and 157 as of yet). This puts my RAM clockspeed to 1836. I've had it at 153 for just over a week now.

I haven't mastered Prime95 as of yet, so I'm still using IntelBurnTest. Now, I know that certain CPU core features need to be disabled when you overclock and I've listed how I think the layout should be, but can anyone confirm this?

• Intel Turbo Boost Tech: Enabled
• CPU Cores Enabled: All
• CPU Multi-Threading: Enabled
• CPU Enhanced Halt (C1E): Disabled
• C3/C6/C7 State Support: Disabled
• CPU Thermal Monitor: Enabled
• CPU EIST Function: Enabled
• Bi-Directional PROCHOT: Enabled

Also, at the normal voltage, when all eight of my threads are being stressed, the average core temperature is ~72C. This doesn't seem normal, as I only got to ~50C when my BCLK was at 133. Is it just the paranoid part of my conscious saying "you shouldn't be doing this...", or do I have a real problem here?

Also, what temperature should I keep my CPU at generally? What is the "general" maximum temperature my CPU should be kept at (80C, 90C?)? I'm inclined to think that 75-80C is the max, but this is the very first Intel I've owned and the very first overclocked CPU I've done, so it might be different to stock AMD Athlons. I don't really want any burn marks on my CPU.

EDIT: What exactly is the Dynamic VCore (DVID), and why do you suggest a slight undervolt? I would have thought that you'd leave it at 0.00V or normal like the rest?
 
72c is fine under max load. You will never see the i7 under maximum load during anything but prime95/burn test anytime in the near future. As long as you are idleing near or under 40c, you'll never see temps above the mid 50s during gaming/heavy usage from my experience.
 
I agree with foolycooly. Some people are comfortable at 90. I like to stay under 80 at max. But in normal operating conditions its much lower anyway.

I didn't mean to suggest anyone had to undervolt DVID. It's used as an offset to allow dynamic voltage reduction of VCore between idle and overclocked modes. Without it, if you set the Vcore setting your board will always supply that much voltage. With Vcore set to Normal and DVID populated, your board will run at Normal V when idle (EIST) but apply additional DVID V when it ramps up to the manual overclock. You may need to enable the C-states for dynamic voltage to be in play. The easiest way to find out it to screenshot CPU-Z when your processor is at idle and at max.
Here's some threads where DVID has been discussed (some have off topic info in them, too):
http://forums.anandtech.com/showpost.php?p=28882027&postcount=18
http://forums.anandtech.com/showthread.php?t=335080
http://forums.anandtech.com/showthread.php?t=330645

I chose to undervolt in this round of testing for two reasons: 1) I wanted my first round of V testing to be using less V - rather than more; and 2) I wanted to test the theory that "at stock" my board was supplying more V than it needed to.


Does Turbo still work for you? I thought that C1E and C3/C6/C7 were required for Turbo, but I may be remembering that wrong.

Which Voltages did you change from Auto and what did you set them to?
 
If there isn't going to be much that'll drive my CPU beyond 50C, then that's great. If the worst comes, I might just need to install a push fan on my cooling unit (then I've got cool air coming on to it, hot air off in the opposite direction and the mega fan sucking up as much as possible). My idling temperature is between 24C and 35C.

Turbo works, but only after I turned all of the Core features back on.

I put every voltage to "Normal", except for my DRAM voltage, which is at 1.600v.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom